The present disclosure relates to a mechanical property measuring apparatus, a mechanical property measuring method, a substance manufacturing equipment, a substance management method, and a substance manufacturing method.
In the production of a steel material used as raw material of line pipes and the like, a sampling inspection is sometimes conducted as an inspection of the mechanical property of the steel material. The sampling inspection is a destructive test in which a part to be inspected is taken out from the steel material, worked into a mechanical test piece, and tested. In recent years, there has been a demand to, instead of sampling inspections, non-destructively measure or evaluate the mechanical property of a steel material product itself and assure the quality. Attempts have thus been made to measure the mechanical property of a steel material through various physical quantities relating to the mechanical property of the steel material and measured during or after the production of the steel material.
For example, JP 2008-224495 A (PTL 1) describes a technique of applying an alternating magnetic field to a metal material and detecting induced eddy current to detect a high hardness portion locally present in the metal material.
For example, WO 2019/087460 A1 (PTL 2) describes a detection device including a yoke member that has a first opening into which a long material is inserted on one side in the longitudinal direction of the long material and a second opening into which the long material is inserted on the other side in the longitudinal direction of the long material and that has a shape approximately symmetrical about an axis passing through the first opening and the second opening. The detection device in PTL 2 can reduce dead zones at the longitudinal ends of the long material, and accurately detect changes in magnetic property.
For example, JP H9-113488 A (PTL 3) describes a technique of evaluating the thickness of a coating material of an object under examination from the intensity of eddy current induced in the object and determining the degree of degradation of the object from information about the thickness reduction of the coating material.
In the case of measuring electromagnetic feature values of a steel material surface layer by a sensor to measure a mechanical property, accurate calculation is difficult with the conventional techniques because the relationship between the electromagnetic feature values and the mechanical property varies greatly. For example, steel materials manufactured to have the same mechanical property may differ in the electromagnetic feature values of the steel material surface layer measured by the sensor. There is thus a need for a technique that is usable in a steel material production process and enables accurate measurement of a mechanical property in a non destructive manner.
It could therefore be helpful to provide a mechanical property measuring apparatus and a mechanical property measuring method that can accurately measure a mechanical property through physical quantities. It could also be helpful to provide a substance manufacturing equipment and a substance manufacturing method that can improve the production yield rate of a substance by enabling accurate measurement of a mechanical property through physical quantities. It could also be helpful to provide a substance management method that can provide a high-quality substance by enabling accurate measurement of a mechanical property through physical quantities.
We studied the relationship between the physical quantities and the mechanical property of a measured object (i.e. an object to be measured), and discovered that the relationship is influenced by the property of a film of the measured object.
A mechanical property measuring apparatus according to one embodiment of the present disclosure comprises: a physical quantity measuring unit configured to measure a plurality of physical quantities of a measured object that includes a substance and a film on a surface of the substance; a calculation model generating unit configured to select a plurality of pieces of learning data from a learning data set based on selection physical quantities that are at least two of the plurality of physical quantities measured, and generate, from the plurality of pieces of learning data selected, a calculation model for calculating a mechanical property of the substance; and a mechanical property calculating unit configured to calculate the mechanical property of the substance using the calculation model generated and the at least two of the plurality of physical quantities, wherein the selection physical quantities include at least one physical quantity measured using a first measurement signal and at least one physical quantity measured using a second measurement signal.
A mechanical property measuring method according to one embodiment of the present disclosure comprises: a measuring step of measuring a plurality of physical quantities of a measured object that includes a substance and a film on a surface of the substance; a selecting step of selecting a plurality of pieces of learning data from a learning data set based on selection physical quantities that are at least two of the plurality of physical quantities measured; a generating step of generating, from the plurality of pieces of learning data selected, a calculation model for calculating a mechanical property of the substance; and a calculating step of calculating the mechanical property of the substance using the calculation model generated and the at least two of the plurality of physical quantities, wherein the selection physical quantities include at least one physical quantity measured using a first measurement signal and at least one physical quantity measured using a second measurement signal.
A substance manufacturing equipment according to one embodiment of the present disclosure comprises: a manufacturing equipment configured to manufacture a substance; and a mechanical property measuring apparatus, wherein the mechanical property measuring apparatus includes: a physical quantity measuring unit configured to measure a plurality of physical quantities of a measured object that includes the substance and a film on a surface of the substance; a calculation model generating unit configured to select a plurality of pieces of learning data from a learning data set based on selection physical quantities that are at least two of the plurality of physical quantities measured, and generate, from the plurality of pieces of learning data selected, a calculation model for calculating a mechanical property of the substance; and a mechanical property calculating unit configured to calculate the mechanical property of the substance using the calculation model generated and the at least two of the plurality of physical quantities, the selection physical quantities include at least one physical quantity measured using a first measurement signal and at least one physical quantity measured using a second measurement signal, and the mechanical property measuring apparatus is configured to measure the mechanical property of the substance manufactured by the manufacturing equipment.
A substance management method according to one embodiment of the present disclosure comprises: a measuring step of measuring a plurality of physical quantities of a measured object that includes a substance and a film on a surface of the substance; a selecting step of selecting a plurality of pieces of learning data from a learning data set based on selection physical quantities that are at least two of the plurality of physical quantities measured and include at least one physical quantity measured using a first measurement signal and at least one physical quantity measured using a second measurement signal; a generating step of generating, from the plurality of pieces of learning data selected, a calculation model for calculating a mechanical property of the substance; a calculating step of calculating the mechanical property of the substance using the calculation model generated and the at least two of the plurality of physical quantities; and a management step of classifying the calculated substance based on the mechanical property of the substance.
A substance manufacturing method according to one embodiment of the present disclosure comprises: a manufacturing step of manufacturing a substance; a measuring step of measuring a plurality of physical quantities of a measured object that includes the substance manufactured and a film on a surface of the substance; a selecting step of selecting a plurality of pieces of learning data from a learning data set based on selection physical quantities that are at least two of the plurality of physical quantities measured; a generating step of generating, from the plurality of pieces of learning data selected, a calculation model for calculating a mechanical property of the substance; and a calculating step of calculating the mechanical property of the substance using the calculation model generated and the at least two of the plurality of physical quantities, wherein the selection physical quantities in the selecting include at least one physical quantity measured using a first measurement signal and at least one physical quantity measured using a second measurement signal.
It is thus possible to provide a mechanical property measuring apparatus and a mechanical property measuring method that can accurately measure a mechanical property through physical quantities. It is also possible to provide a substance manufacturing equipment and a substance manufacturing method that can improve the production yield rate of a substance by enabling accurate measurement of a mechanical property through physical quantities. It is also possible to provide a substance management method that can provide a high quality substance by enabling accurate measurement of a mechanical property through physical quantities.
In the accompanying drawings:
Although this embodiment describes an example in which the substance 1 is a steel material, the substance 1 is not limited to a steel material. Although this embodiment describes an example in which the mechanical property is hardness, the mechanical property is not limited to hardness. Although this embodiment describes an example in which the plurality of physical quantities are electromagnetic feature values, the plurality of physical quantities are not limited to electromagnetic feature values. It is conventionally known that electromagnetic feature values such as magnetic permeability and coercive force correlate with the mechanical property of metal. Hence, it is preferable to measure or evaluate the mechanical property using electromagnetic feature values. A preferable method of measuring electromagnetic feature values is, for example, eddy current examination or micromagnetic multiparameter microstructure and stress analysis (3MA). In particular, it is preferable to use an AC signal (alternating current or alternating voltage) obtained by superimposing two or more frequencies as the below-described measurement signal, because more electromagnetic feature values can be acquired. It is more preferable to set one of the frequencies to 200 Hz or less, because, even in the case where a film 2 (see
(Structure of Mechanical Property Measuring Apparatus)
As illustrated in
For example, in the case where the substance 1 is a steel material, an iron oxide film called scale or mill scale forms on the surface of the steel material during the production of the steel material. There are various types of iron oxide films, and magnetite (triiron tetraoxide, Fe3O4), wustite (ferrous oxide, FeO), and hematite (red hematite, Fe2O3) are commonly known. These scales differ not only in the composition of oxygen and iron but also in electromagnetic features (i.e. electromagnetic characteristics). For example, magnetite is magnetic, but wustite is not magnetic. To measure the mechanical property of the substance 1 (in particular the surface layer) which is a steel material, the physical quantities are measured from the surface. That is, in the present disclosure, the physical quantities are measured with the substance 1, which is a steel material, and the film 2, which is scale, together as the measured object 101.
Accordingly, the film 2 which is scale influences the measurement of the substance 1 which is a steel material. The type and composition of the scale vary depending on the state during the production of the steel material. Moreover, the steel material may have magnetic anisotropy depending on the microstructure of the steel material. Thus, the electromagnetic features differ among measured objects 101. It is therefore very difficult to, for the measured object 101 including the steel material and the scale, measure or evaluate the mechanical property, such as hardness, of the steel material by simply associating the mechanical property with the electromagnetic feature values of the measured object 101. Especially in the case of measuring the mechanical property of the surface layer of the substance 1, the electromagnetic features of the scale as the film 2 have greater influence. This makes it more difficult to, for the measured object 101 including the steel material and the scale, measure or evaluate the mechanical property, such as hardness, of the surface layer of the steel material by simply associating the mechanical property with the electromagnetic feature values of the measured object 101.
The same applies to when the substance 1 is other than a steel material and the film 2 is other than scale. In particular, in the case where the film 2 has different features from the substance 1 with regard to the plurality of physical quantities to be measured, it is very difficult to, for the measured object 101 including the substance 1 and the film 2 on its surface, measure or evaluate the mechanical property of the substance 1 by simply associating the mechanical property with the plurality of physical quantities of the measured object 101. Moreover, in the case of measuring the mechanical property of the surface layer of the substance 1, it is more difficult to, for the measured object 101 including the substance 1 and the film 2 on its surface, measure or evaluate the mechanical property of the surface layer of the substance 1 by simply associating the mechanical property with the plurality of physical quantities of the measured object 101.
The storage unit 10 stores various information and programs for causing the measuring apparatus 100 to operate. The information stored in the storage unit 10 includes the learning data set 110 which is a set of a plurality of pieces of learning data. The programs stored in the storage unit 10 include a program for causing the control unit 8 to operate as the calculation model generating unit 81, a program for causing the control unit 8 to operate as the mechanical property calculating unit 82, and a program for causing the control unit 8 to operate as the physical quantity measurement control unit 83. The storage unit 10 includes, for example, semiconductor memory or magnetic memory.
The display 11 displays various information including the mechanical property of the substance 1 to a user. In this embodiment, the display 11 includes a display capable of displaying text, images, etc. and a touch screen capable of detecting contact with the user's finger or the like. The display may be a display device such as a liquid crystal display (LCD) or an organic electroluminescence display (OELD). The detection method of the touch screen may be any method such as capacitive, resistive, surface acoustic wave, infrared, electromagnetic induction, or load detection. Alternatively, the display 11 may be composed of a display without a touch screen.
The control unit 8 controls the overall operation of the measuring apparatus 100. The control unit 8 includes one or more processors. The processors may include a general-purpose processor that reads a specific program and executes a specific function and/or a dedicated processor dedicated to a specific process. The dedicated processor may include an application specific integrated circuit (ASIC). Each processor may include a programmable logic device (PLD). The PLD may include a field-programmable gate array (FPGA). The control unit 8 may include at least one of a system in a package (SiP) and a system on a chip (SoC) in which one or more processors cooperate with each other. The control unit 8 functions as each of the calculation model generating unit 81, the mechanical property calculating unit 82, and the physical quantity measurement control unit 83 according to the corresponding program read from the storage unit 10.
The control unit 8 also collects learning data via a communication unit 7, associates a plurality of pieces of learning data for each item to generate the learning data set 110, and stores the learning data set 110 in the storage unit 10. The details of the learning data set 110 will be described later.
The calculation model generating unit 81 selects a plurality of pieces of learning data from the learning data set 110, based on at least two of the plurality of physical quantities of the measured object 101 measured by the physical quantity measuring unit 5. The physical quantities used to select the plurality of pieces of learning data are hereafter referred to as “selection physical quantities”. As an example, suppose all of current waveform phase change, harmonic amplitude, and incremental magnetic permeability which are electromagnetic feature values are selection physical quantities. First, the calculation model generating unit 81 acquires the learning data set 110 from the storage unit 10. The calculation model generating unit 81 then selects a plurality of pieces of learning data close to the combination of the acquired values of current waveform phase change, harmonic amplitude, and incremental magnetic permeability. The calculation model generating unit 81 generates a calculation model from the selected plurality of pieces of learning data. The generated calculation model is used by the mechanical property calculating unit 82. Examples of the calculation model used in the present disclosure include a regression model using the k-nearest neighbor algorithm, a local linear regression model, and a regression model using a support vector machine. Of these, a regression model using the k-nearest neighbor algorithm and a local linear regression model are preferable, for the following reason. In a typical regression model, the prediction accuracy may decrease due to, for example, distribution bias of a whole dataset. This is because the model is usually constructed so as to have minimal error for the whole dataset. Hence, a method of extracting k pieces of data that are, in Euclidean distance, near the data to be evaluated by the k-nearest neighbor algorithm, generating a regression model using a dataset of the extracted k pieces of data, and evaluating the data is preferable in terms of enhancing accuracy. A local linear regression model can achieve the same effects as the k nearest neighbor algorithm. In the case of the local linear regression model, a sequential model is constructed while weighting a whole dataset so that a dataset near the data to be evaluated will have influence on regression and a dataset farther from the data to be evaluated will have less influence on regression, and the data is evaluated. This contributes to higher accuracy.
The mechanical property calculating unit 82 calculates the mechanical property of the substance 1, using the calculation model generated by the calculation model generating unit 81 and at least two of the plurality of physical quantities measured by the physical quantity measuring unit 5. As an example, suppose the plurality of physical quantities include the foregoing electromagnetic feature values, and all of current waveform phase change, harmonic amplitude, and incremental magnetic permeability are used to calculate the mechanical property of the substance 1. The mechanical property calculating unit 82 acquires the calculation model generated by the calculation model generating unit 81. The mechanical property calculating unit 82 inputs the acquired values of current waveform phase change, harmonic amplitude, and incremental magnetic permeability to the calculation model, to calculate the mechanical property of the substance 1. The mechanical property calculating unit 82 may output the calculated hardness of the steel material to the display 11 to present it to the user.
Although the calculation model generating unit 81 uses all of the electromagnetic feature values as the selection physical quantities when generating the calculation model in the foregoing example, a combination of two or more but not all of the electromagnetic feature values may be used. Although the mechanical property calculating unit 82 uses all of the electromagnetic feature values when calculating the mechanical property of the substance 1 in the foregoing example, two or more but not all of the electromagnetic feature values may be input to the calculation model. Here, the electromagnetic feature values input to the calculation model may be different from the electromagnetic feature values used when the calculation model generating unit 81 generates the calculation model. For example, the calculation model generating unit 81 generates the calculation model using a combination of current waveform phase change and incremental magnetic permeability, and the mechanical property calculating unit 82 inputs current waveform phase change and harmonic amplitude to the calculation model and calculates the mechanical property of the substance 1.
The physical quantity measurement control unit 83 controls the operation of the physical quantity measuring unit 5. For example, the physical quantity measurement control unit 83 causes the sensor 3 to operate and measure the electromagnetic feature values.
(Structure of physical quantity measuring unit)
The sensor 3 measures the physical quantities of the measured object 101 including the substance 1 and the film 2. Although this embodiment describes an example in which the sensor 3 is a magnetic sensor, the sensor 3 is not limited to a magnetic sensor. The sensor 3 may be composed of one or more sensors. The measurement results of the sensor 3 indicate the physical quantities involving the influence of the film 2, that is, the physical quantities in a state in which not only the substance 1 but also the film 2 is included. On the other hand, the mechanical property calculated by the mechanical property calculating unit 82 relates to the substance 1 without the film 2.
As the physical quantities of the steel material to be measured, the electromagnetic feature values of the surface layer may be used. It is known that, in a steel material, changes in a magnetic hysteresis curve and Barkhausen noise correlate with the mechanical property of the material such as tensile strength and hardness. Hence, it is preferable to measure the electromagnetic feature values of the surface layer by the magnetic sensor illustrated in
As defined in formula (1), the penetration depth is shallower when the frequency is higher. In other words, the penetration depth is deeper when the frequency is lower. Hence, the penetration depth can be adjusted by adjusting the frequency according to the surface layer depth range to be measured or evaluated. For example, to measure or evaluate the mechanical property up to about 0.25 mm of the surface layer, the frequency is determined so that the penetration depth will be about 0.25 mm. Preferably, ¾ of the penetration depth is greater than 0.25 mm with respect to the surface layer depth, in consideration of attenuation.
The scanning unit 6 moves the sensor 3 relative to the measured object 101. The scanning unit 6 may move the sensor 3 to an evaluation location designated by the physical quantity measurement control unit 83. The scanning unit 6 may acquire information of the moving speed of the substance 1, and adjust the sensor 3 to move at an appropriate relative speed.
(Selection Physical Quantities)
Preferable conditions for the selection physical quantities will be described below. The selection physical quantities are the most important concept in the present disclosure. The selection physical quantities are at least two of the plurality of physical quantities measured by the physical quantity measuring unit 5. The selection physical quantities include at least one physical quantity measured using a first measurement signal and at least one physical quantity measured using a second measurement signal.
That is, the physical quantities of the measured object 101 measured by the physical quantity measuring unit 5 include one or more physical quantities measured using the first measurement signal and one or more physical quantities measured using the second measurement signal. In the case where the physical quantities to be measured are electromagnetic feature values, the first measurement signal may be an AC signal having a first frequency and the second measurement signal may be an AC signal having a second frequency higher than the first frequency. In other words, the first measurement signal and the second measurement signal by the physical quantity measuring unit 5 may be a low frequency signal and a high frequency signal respectively.
In the case where the signal obtained by superimposing the high frequency signal on the low frequency signal illustrated in
For example, in the case where the measured object 101 is a steel material having scale, it is preferable to apply, to the excitation coil 31 of the electromagnetic sensor, a voltage or a current obtained by superimposing a sine wave with a frequency of 1 kHz or more on a sine wave with a frequency of 150 Hz or less. As a result of the low frequency signal being a sine wave of 150 Hz or less, an alternating magnetic field excited by the electromagnetic sensor can penetrate to a depth of about 300 μm from the surface of the steel material. The electromagnetic feature values measured using the low frequency signal preferably include characteristics relating to current waveform phase change. In the measurement using the low frequency signal, the alternating magnetic field penetrates relatively deep, so that more information of the substance 1 than the film 2 can be included. Current waveform phase change includes information about coercive force. Accordingly, by measuring the characteristics relating to current waveform phase change using the low frequency signal, information about the coercive force of the substance 1 can be obtained. The electromagnetic feature values measured using the high frequency signal preferably include characteristics relating to incremental magnetic permeability. In the measurement using the high frequency signal, the penetration of the alternating magnetic field is relatively shallow, so that more information of the film 2 than the substance 1 can be included. Incremental magnetic permeability includes information of the magnetic property of the film 2 in a state in which a magnetic field that changes according to a low frequency signal is applied. Accordingly, by measuring the characteristics relating to incremental magnetic permeability using the high frequency signal, information about the magnetic property of the film 2 can be obtained. The obtainment of the magnetic property of the film 2 benefits accurate prediction of the property of the substance 1 while compensating for the influence of the film 2. The selection physical quantities preferably include at least one physical quantity measured using the first measurement signal and at least one physical quantity measured using the second measurement signal so as to contain accurate information of both the substance 1 and the film 2.
(Collection of Learning Data)
The mechanical property measuring apparatus 100 according to this embodiment calculates the mechanical property of the substance 1 based on the physical quantities of the measured object 101 measured by the physical quantity measuring unit 5 and the selected plurality of pieces of learning data. For example, the measured object 101 is a steel material having scale. For example, the physical quantities include electromagnetic feature values. For example, the mechanical property of the substance 1 is the hardness of the steel material. In the calculation of the mechanical property of the substance 1, a plurality of pieces of learning data are selected from the learning data set prepared beforehand to calculate the mechanical property of the substance, and a calculation model is generated. For accurate measurement of the mechanical property, it is necessary to select appropriate learning data based on the physical quantities and generate a correct calculation model. It is therefore preferable to carefully collect the learning data set that is the basis of the calculation model. For example, a measurement system including the measuring apparatus 100 and the physical quantity measuring unit 5 collects learning data in the following manner.
The control unit 8 causes the physical quantity measuring unit 5 to measure the physical quantities in the set evaluation location (step S2). In the learning data, the physical quantities of the measured object 101 are explanatory variables.
The control unit 8 performs pretreatment (step S3). The pretreatment is, for example, removing the film 2 from the measured object 101 to enable the measurement of the mechanical property in the evaluation location. For example, in the case where the measured object 101 is a steel material having scale on its surface, the scale may be removed by etching, grinding, or the like. The pretreatment may include cutting the measured object 101 at the evaluation location to expose a cross-section of the substance 1.
The control unit 8 measures the mechanical property in the evaluation location (step S4). The learning data includes the mechanical property as an objective variable. The mechanical property may be, for example, the hardness of the cross-section of the steel material in the evaluation location. As the mechanical property, for example, a value obtained by converting the Leeb hardness of the surface of the steel material measured by a rebound hardness meter into the hardness of the cross-section using a conversion formula obtained from past tests may be used. For more accurate conversion, a value obtained by normalizing the converted value with respect to the thickness of the steel material may be used. That is, a process of conversion into a value at the reference thickness of the steel material may be performed. The reference thickness of the steel material is, for example, 28 mm. In the case where the foregoing pretreatment involves cutting the measured object 101 at the evaluation location, the mechanical property may be the Vickers hardness obtained by directly measuring the cut surface. The control unit 8 acquires the measured mechanical property. The control unit 8 stores a data label such as the management number and the evaluation location of the substance 1, the explanatory variables, and the objective variable in the storage unit 10 in association with each other as one piece of learning data.
In the case where the control unit 8 determines that learning data sufficient for model generation has not been collected (step S5: No), the control unit 8 returns to the process in step S1 and further collects learning data.
In the case where the control unit 8 determines that learning data sufficient for model generation has been collected and the collection has been completed (step S5: Yes), the control unit 8 ends the series of processes.
The learning data set 110 stored in the storage unit 10 by the control unit 8, i.e. a set of a plurality of pieces of learning data, may include objective variables obtained by different measurement methods. In the foregoing example, the learning data set 110 may include objective variables obtained by at least two methods from among the Vickers hardness obtained by directly measuring the cut surface, the value obtained by converting the Leeb hardness of the surface of the steel material into the hardness of the cross-section, and the value obtained by normalizing the converted value with respect to the thickness of the steel material. For example, the Vickers hardness is accurate, but measuring the Vickers hardness takes time because the steel material is cut. By allowing for a mixture of objective variables obtained by different measurement methods, an accurate learning data set 110 can be generated within a realistic time.
(Mechanical Property Measuring Method)
The mechanical property measuring apparatus 100 according to this embodiment calculates the mechanical property of the substance 1 based on the physical quantities of the measured object 101 measured by the physical quantity measuring unit 5. For example, the measured object 101 is a steel material having scale. For example, the substance 1 is the steel material. For example, the film 2 on the surface of the substance 1 is the scale. For example, the physical quantities include electromagnetic feature values. For example, the mechanical property of the substance 1 is the hardness of the steel material. For example, the sensor 3 is the magnetic sensor illustrated in
The control unit 8 causes the physical quantity measuring unit 5 to measure the physical quantities of the measured object 101 (measurement step, step S11). To measure the mechanical property of the substance 1 (in particular the surface layer), the physical quantities are measured from the surface of the substance 1 on which the film 2 is formed. That is, in this measurement method, the physical quantities are measured with the substance 1, which is a steel material, and the film 2, which is scale, together as the measured object 101. The same applies to when the substance 1 is other than a steel material and the film 2 is other than scale. Specifically, the sensor 3 in the physical quantity measuring unit 5 is located at the surface of the film 2. The measurement results of the sensor 3 indicate the physical quantities involving the influence of the film 2, that is, the physical quantities in a state in which not only the substance 1 but also the film 2 is included. The scanning unit 6 moves the sensor 3 relative to the measured object 101. As a result, the sensor 3 applies an alternating magnetic field to an evaluation location in the measured object 101 designated by the physical quantity measurement control unit 83. The sensor 3 measures the influence of eddy current or the like induced in the measured object 101 by the alternating magnetic field, as changes in electromagnetic feature values. The physical quantity measuring unit 5 outputs the measured electromagnetic feature values to the control unit 8 as a plurality of physical quantities.
The control unit 8 selects a plurality of pieces of learning data from the learning data set 110, based on selection physical quantities that are at least two of the acquired physical quantities (selection step, step S12). Here, the control unit 8 selects learning data close to the acquired selection physical quantities, from the learning data set 110 stored in the storage unit 10.
The control unit 8 generates a calculation model for calculating the mechanical property of the substance 1, from the selected plurality of pieces of learning data (generation step, step S13). The calculation model may be prepared as a linear regression model or a nonlinear regression model that links the explanatory variables and the objective variable of the learning data. As the linear regression model, a generalized linear model, a generalized linear mixed model, or the like may be used. From the viewpoint of improvement in calculation accuracy, the calculation model is preferably generated by local linear regression in combination with the process in step S12. Here, in the generation of the calculation model, it is preferable to weight each of the plurality of pieces of learning data selected in the process in step S12 according to the distance from the selection physical quantities. That is, it is preferable to assign a larger weight to a piece of learning data whose distance from the selection physical quantities is shorter.
The control unit 8 calculates the mechanical property of the substance 1 based on the generated calculation model (calculation step, step S14). The control unit 8 calculates the mechanical property of the substance 1 using the generated calculation model and at least two physical quantities necessary as input.
The mechanical property of the substance 1 may be, for example, the hardness of the cross-section of the steel material in the evaluation location. As the mechanical property, for example, a value obtained by converting the Leeb hardness of the surface of the steel material measured by a rebound hardness meter into the hardness of the cross section using a conversion formula obtained from past tests may be used. For more accurate conversion, a value obtained by normalizing the converted value with respect to the thickness of the steel material may be used. That is, a process of conversion into a value at the reference thickness of the steel material may be performed. The reference thickness of the steel material is, for example, 28 mm. In the case where the foregoing pretreatment involves cutting the measured object 101 at the evaluation location, the mechanical property may be the Vickers hardness obtained by directly measuring the cut surface.
The control unit 8 outputs the calculated mechanical property of the substance 1 to the display 11 (output step, step S15), and ends the series of processes. The mechanical property of the substance 1 displayed by the display 11 is recognized by the user. The user may, for example, perform quality management for the substance 1 or issue an instruction to change the production parameters of the substance 1, based on the displayed mechanical property of the substance 1.
As described above, with the mechanical property measuring apparatus 100 and the mechanical property measuring method executed by the measuring apparatus 100 according to this embodiment, the mechanical property can be measured accurately through the physical quantities. In particular, in the case where the film 2 has different features from the substance 1 with regard to the plurality of physical quantities to be measured, a more appropriate calculation model can be generated by the calculation model generating unit 81 or the selection step and the generation step (steps S12 and S13), so that the foregoing effects can be further enhanced. Moreover, in the case of measuring the mechanical property of the surface layer of the substance 1, a more appropriate calculation model can be generated by the calculation model generating unit 81 or the selection step and the generation step (steps S12 and S13), so that the foregoing effects can be further enhanced. The foregoing effects can be equally achieved in the below-described Embodiment 2.
The effects according to the present disclosure will be described in detail below by way of examples, although the present disclosure is not limited to these examples.
In Example 1, the measuring apparatus 100 is a apparatus that measures the hardness of the surface layer of a steel material. In this example, the substance 1 is the steel material. The film 2 is scale formed on the surface of the steel material. The sensor 3 is an electromagnetic sensor. The physical quantities of the measured object 101 are the electromagnetic feature values of the steel material having the scale. The mechanical property to be measured in this example is the hardness of a cross-section of the steel material at a depth of 0.25 mm.
The steel material was manufactured by subjecting a continuously cast slab to rough rolling and further to quenching by continuous cooling online. To collect learning data, the hardness of a cross-section at a depth of 0.25 mm was measured for the steel material manufactured by this production process.
In this example, electromagnetic sensors capable of measuring electromagnetic feature values were mounted in the measuring apparatus 100, and the electromagnetic feature values of the surface layer of the steel material having scale on its surface were measured. As the scanning unit 6, a truck moved by human power was used. Eight electromagnetic sensors were arranged side by side in the truck. The eight electromagnetic sensors scanned the whole surface of the steel material.
A voltage obtained by superimposing, on a sine wave having a first frequency, a sine wave having a second frequency higher than the first frequency was applied to each electromagnetic sensor. The first frequency was set to 150 Hz or less. The second frequency was set to 1 kHz or more. A plurality of types of electromagnetic feature values were extracted from the current waveforms observed by the electromagnetic sensors. In this example, 20 feature values such as the distortion amount, amplitude, and phase change of the current waveform, the amplitude and phase change of the harmonic, the maximum value, minimum value, and average value of the incremental magnetic permeability, and the coercive force were extracted as the electromagnetic feature values. The 20 feature values consist of four physical quantities measured using the low frequency signal and 16 physical quantities measured using the high frequency signal. Here, the frequency of the sine wave applied was limited to 150 Hz or less so that an alternating magnetic field excited by each electromagnetic sensor would penetrate to a depth of about 300 μm from the surface of the steel material. The incremental magnetic permeability is a value indicating magnetizability in a state in which a magnetic field is applied, and is expressed by the gradient of a minor loop in a magnetization curve that indicates the relationship between the magnetic flux density and the magnetic field.
A sufficient number of pieces of learning data were collected, and the learning data set 110 was stored in the storage unit 10. The sufficient number of pieces of learning data is, for example, 100.
To calculate the hardness of the surface layer of the steel material, the measuring apparatus 100 measured the electromagnetic feature values by the physical quantity measuring unit 5. The selection physical quantities were set to include at least one physical quantity measured using the low frequency signal and at least one physical quantity measured using the high frequency signal from among the electromagnetic feature values. Specifically, the selection physical quantities were set to include at least the characteristics relating to current waveform phase change measured using the low frequency signal and the characteristics relating to incremental magnetic permeability measured using the high frequency signal. The control unit 8 selected a plurality of pieces of learning data from the learning data set 110 in the storage unit 10, based on the selection physical quantities. The control unit 8 generated a calculation model by local linear regression, using the selected plurality of pieces of learning data. The control unit 8 then calculated the hardness using the generated calculation model.
Example 2 is an example in which the mechanical property measuring method executed by the measuring apparatus 100 was used to inspect the hardness of the surface layer in a steel plate manufacturing method.
In the rough rolling step S41, for example, a slab 41 is subjected to rough hot rolling at a temperature of 1000° C. or more. In the finish rolling step S42 which follows, the slab 41 is subjected to finish hot rolling at a temperature of 850° C. or more, to obtain a steel plate 42. In the cooling step S43 which follows, the steel plate 42 is cooled. In the cooling step S43, for example, the steel plate is cooled from a temperature of 800° C. or more to a temperature of about 450° C.
In the surface layer hardness measurement step S45, the mechanical property of the surface layer is measured for the whole surface of the steel plate 42 after the cooling, using the measurement method executed by the measuring apparatus 100. From the measurement result, each part harder than preset surface layer hardness is determined as a hardened portion.
In the case where the steel plate is lifted using a magnetic force such as a magnet crane, a residual magnetic field remains in the part to which the magnet part of the crane is attracted. When measuring the electromagnetic feature values to measure the mechanical property, if there is a residual magnetic field at least in the surface layer, the mechanical property measurement or evaluation accuracy may decrease. Accordingly, in the case where there is a process that causes generation of a residual magnetic field, it is preferable to add the demagnetizing step S44 immediately before the surface layer hardness measurement step S45 and demagnetize the residual magnetic field in the demagnetizing step S44. In the demagnetizing step S44, a demagnetizing apparatus performs demagnetization using a distance attenuation method so that the residual magnetic field in the surface layer will be 0.5 mT or less.
In the remeasurement step S46, the surface layer hardness of each hardened portion detected in the surface layer hardness measurement step S45 is remeasured. Here, the mechanical property of the surface layer is remeasured only for the hardened portion and its vicinity, using the measurement method executed by the measuring apparatus 100. In the case where the remeasured surface layer hardness of the hardened portion is again determined to exceed the foregoing threshold, the part is determined to have a locally hard portion, and the steel plate 42 is sent to the removal step S47.
In the removal step S47, the part determined as the hardened portion in the remeasurement step S46 is removed. Specifically, the part determined as the hardened portion is removed by grinding using a known grinding means such as a grinder. After the removal step S47, the production of the steel plate 43 from the steel plate 42 is completed, and the steel plate 43 is sent to other steps (a step of shipment to a customer, a steel pipe or tube manufacturing step, etc.). It is desirable to, for the part of the steel plate 42 ground in the removal step S47, measure the thickness of the steel plate 42 at the grinding position using a known or existing thickness meter, and determine whether it is within a dimensional tolerance set beforehand in the steel plate production. Moreover, it is desirable to, after removing the hardened portion, measure the surface layer hardness for the hardened portion once again using a known contact hardness meter. From this measurement result, whether the surface layer hardness is less than or equal to preset surface layer hardness is determined. If the surface layer hardness is determined to be less than or equal to the preset surface layer hardness, the production of the steel plate 43 from the steel plate 42 is completed.
In the case where it is determined that there is no hardened portion in the surface layer hardness measurement step S45 or in the case where it is determined that the part is not a hardened portion in the remeasurement step S46, the production of the steel plate 43 from the steel plate 42 is completed without performing the removal step S47, and the steel plate 43 is sent to another step (a step of shipment to a customer, a steel pipe or tube manufacturing step, etc.).
The steel plate manufacturing method in this example may further include an annealing step S48 (not illustrated) and the like after the cooling step S43 and before the surface layer hardness measurement step S45. Particularly in the case where the surface layer hardness (more specifically, the Vickers hardness measured from the top surface from which oxide scale has been removed, according to ASTM A 956/A 956MA Standard Test Method for Leeb Hardness Testing of Steel Products) of the steel plate 43 to be manufactured is 230 Hv or more and the steel is of a type that tends to develop warping in the steel plate 43, it is desirable to perform the annealing step S48 after the cooling step S43 and before the surface layer hardness measurement step S45. The annealing step S48 can facilitate microstructure softening by tempering. Since microstructure softening leads to reduction of occurrences of hardened portions, reduction of removal regions can be expected.
In the surface layer hardness measurement step S45, the hardness is measured from the top surface from which oxide scale has been removed according to ASTM A 956/A 956MA Standard Test Method for Leeb hardness Testing of Steel products to determine the hardness, as mentioned above. Here, in rebound hardness measurement, the thickness of the measured object influences the measured value. Hence, the value of the cross-sectional Vickers hardness at a depth of 0.25 mm and the value of the hardness of the surface layer by a rebound hardness meter are studied for each thickness and a relational formula is constructed beforehand. The value of hardness determined as a hardened portion may be adjusted based on the preconstructed relational formula in consideration of the influence of the thickness, with respect to the cross-sectional hardness at a depth of 0.25 mm. Although the reference depth is 0.25 mm in this example, the reference depth is not limited to such.
Although a known grinding means is used as the method of removing the hardened portion in the surface layer of the steel plate 42 determined in the surface layer hardness measurement step S45 in this example, the removal method is not limited to such. Any known method (e.g. heat treatment) that can remove the hardened portion, other than grinding, may be equally used.
In the case where the mechanical property measuring method executed by the measuring apparatus 100 is used in the method of manufacturing the steel plate 43 as in this example, the steel plate 43 which is the substance 1 of high quality can be provided because the mechanical property can be measured accurately through the physical quantities. More specifically, the steel plate 43 without hardened portions can be manufactured from the steel plate 42.
With the mechanical property measuring apparatus 100, the manufacturing equipment for the substance 1 including the measuring apparatus 100, the mechanical property measuring method executed by the measuring apparatus 100, and the management method and manufacturing method for the substance 1 using the measurement method according to this embodiment, the mechanical property can be measured accurately through the physical quantities as in Embodiment 1. Moreover, since the learning data set 110 is stored in the database 12 outside the measuring apparatus 100, the learning data set 110 exceeding the storage capacity of the internal storage unit 10 can be handled.
The communication method by the communication unit 7 may be a short-range wireless communication standard, a wireless communication standard connecting to a mobile phone network, or a wired communication standard. Examples of the short-range wireless communication standard include Wi-Fi® (Wi-Fi is a registered trademark in Japan, other countries, or both), Bluetooth® (Bluetooth is a registered trademark in Japan, other countries, or both), infrared, and Near Field Communication (NFC). Examples of the wireless communication standard connecting to a mobile phone network include Long Term Evolution (LTE) and a mobile communication system after 4G. Examples of the communication method used for communication between the communication unit 7 and the physical quantity measuring unit 5 include communication standards such as low power wide area (LPWA) and low power wide area network (LPWAN).
While the presently disclosed techniques have been described by way of the drawings and the embodiments, various changes and modifications may be easily made by those of ordinary skill in the art based on the present disclosure. Such changes and modifications are therefore included in the scope of the present disclosure. For example, the functions included in the means, steps, etc. may be rearranged without logical inconsistency, and a plurality of means, steps, etc. may be combined into one means, step, etc. and a means, step, etc. may be divided into a plurality of means, steps, etc.
The structures of the measuring apparatus 100 and the physical quantity measuring unit 5 described in the foregoing embodiments are examples, and all of the components may not necessarily be included. For example, the measuring apparatus 100 may not include the display 11. The measuring apparatus 100 and the physical quantity measuring unit 5 may include other components. For example, there may be a physical distance between the physical quantity measuring unit 5 and each of the control unit 8 and the storage unit 10. In this case, the physical quantity measuring unit 5 and the control unit 8 in the measuring apparatus 100 are electrically connected wiredly or wirelessly. A known technique may be used for the connection.
For example, the presently disclosed techniques can be implemented as programs including processes for achieving the functions of the measuring apparatus 100 or storage media storing such programs, which are also included in the scope of the present disclosure.
For example, although the foregoing embodiments describe the case where the measuring apparatus 100 according to the present disclosure in
For example, although the foregoing embodiments describe an example in which the measuring apparatus 100 generates the calculation model, another information processing apparatus may generate the calculation model. In such a case, the other information processing apparatus acquires the learning data set 110 and generates the calculation model. The other information processing apparatus transmits the generated calculation model to the measuring apparatus 100. That is, the calculation model generated by the other apparatus is installed in the control unit 8 in the measuring apparatus 100 and used as part of the measuring apparatus 100.
For example, although the foregoing embodiments describe an example in which the scanning unit 6 scans the sensor 3, the position of the sensor 3 may be fixed. In the case where the position of the sensor 3 is fixed, the scanning unit 6 may move the measured object 101. Although the above describes the case where the scanning unit 6 is a truck moved by human power, the scanning unit 6 may be a truck including a mechanical driving device. The scanning unit 6 may be controlled by a control unit other than the control unit 8 in the measuring apparatus 100 to scan the sensor 3. In particular, in the case where the physical quantity measuring unit 5 according to the present disclosure is installed in the manufacturing equipment for the substance 1, it is preferable to use one or more out of a known scanning device, a new scanning device, a known scanning method, a new scanning method, a known control device, a new control device, a known control method, and a new control method. The control unit of the scanning unit 6 may cooperate with a control unit (not illustrated) in another manufacturing equipment to enable automatic scanning. Conversely, automatic scanning may be enabled by the control unit 8 in the mechanical property measuring apparatus 100. In this case, the scanning unit 6 may be electrically connected to the control unit of the scanning unit, the control unit in the manufacturing equipment, or the control unit 8 in the measuring apparatus 100 wiredly or wirelessly. A known technique may be used for the connection.
For example, in the foregoing embodiments, the user may input determination based on the displayed mechanical property of the substance 1. The user may input, for example, quality determination on the display 11 by touching the touch screen with a finger or the like. The control unit 8 may perform, for example, control of determining whether or not to carry out the grinding step, depending on the quality determination result from the user. Alternatively, the control unit 8 may determine the quality of the substance 1 based on a set threshold instead of the user, to enhance the efficiency in the management step of managing the substance 1.
Although the foregoing embodiments describe an example in which the substance 1 is a steel material, the physical quantities are electromagnetic feature values, and the mechanical property is hardness, any other combination may be used. For example, the effects according to the present disclosure can be achieved even in the case where the physical quantities are temperatures. For example, the effects according to the present disclosure can be achieved even in the case where the substance 1 is a metal or a compound. The effects can be further enhanced in the case where the film 2 on the surface of the metal or compound has a different feature from the metal or compound with respect to a plurality of physical quantities to be measured. Examples of the metal include iron, steel, nickel, cobalt, aluminum, titanium, and alloys containing one or more thereof. Examples of the compound include inorganic compounds, organic compounds, and compounds containing one or more of iron, steel, nickel, cobalt, aluminum, and titanium. If the substance 1 is iron, steel, nickel, cobalt, an alloy containing one or more thereof, or a compound containing one or more thereof, the effects according to the present disclosure can be achieved more clearly in the case of using electromagnetic feature values as the plurality of physical quantities. In particular, in the case where the substance 1 is a steel material, its mechanical property is determined by the ratio of alloying elements contained in the steel material and the methods of quenching and annealing treatments. Accordingly, at least one of the surface temperatures before and after the quenching treatment and before and after the annealing treatment may be used as a physical quantity to be measured.
The mechanical property measuring apparatus 100 configured as described above and the mechanical property measuring method executed by the measuring apparatus 100 are suitable for use in, for example, the following lines or situations.
The presently disclosed techniques may be applied as part of an inspection equipment included in a manufacturing equipment for the substance 1. In detail, the mechanical property measuring apparatus 100 according to the present disclosure may be used to measure the surface of the substance 1 manufactured in a known, new, or existing manufacturing equipment together with the film 2 on the surface of the substance 1. From the measurement result and, for example, a preset mechanical property, the inspection equipment may inspect the mechanical property of the substance 1. In other words, the mechanical property measuring apparatus 100 according to the present disclosure measures the substance 1 manufactured by the manufacturing equipment. The inspection equipment including the mechanical property measuring apparatus 100 according to the present disclosure inspects the substance 1 manufactured by the manufacturing equipment, using the preset mechanical property as an example.
The presently disclosed techniques may be applied as part of an inspection step included in a manufacturing method for the substance 1. Specifically, the substance 1 manufactured in a known, new, or existing manufacturing step may be inspected in the inspection step in a state in which the surface of the substance 1 has the film 2 formed thereon. The inspection step includes the foregoing measurement step, selection step, generation step, and calculation step according to the present disclosure, and calculates the mechanical property of the substance 1 having the film 2 on its surface as the measured object 101. Alternatively, the inspection step calculates the mechanical property of the substance 1 having the film 2 on its surface as the measured object 101, using the mechanical property measuring apparatus 100 according to the present disclosure. More preferably, the manufacturing method may include a condition change step of, in the case where the mechanical property of the substance 1 calculated by the calculation step or the measuring apparatus 100 is outside a reference range, changing the production conditions in the manufacturing step so that the mechanical property will be within the reference range. The reference range herein may be a standard range of the mechanical property statistically obtained using substances 1 manufactured in the past. The production conditions are parameters adjustable in the manufacturing step of the substance 1. Examples of the production conditions include the heating temperature, the heating time, and the cooling time of the substance 1.
With the manufacturing equipment for the substance 1 and the manufacturing method for the substance 1, the mechanical property can be measured accurately through the physical quantities, so that the substance 1 can be manufactured at a high yield rate. In the case where the mechanical property of the substance 1 obtained by the mechanical property measuring apparatus 100 or the calculation step is the mechanical property of the surface layer of the substance 1, a more appropriate calculation model can be generated by the calculation model generating unit 81 or the selection step and the generation step (steps S12 and S13), so that the foregoing effects can be further enhanced.
An example of a manufacturing equipment for the substance 1 is as follows:
A steel plate manufacturing equipment comprising:
a rolling equipment configured to roll a slab to obtain a steel plate;
an inspection equipment including a mechanical property measuring apparatus according to the present disclosure, and configured to measure surface layer hardness of the steel plate by the measuring apparatus and determine, from the measured surface layer hardness of the steel plate, a part in a surface layer of the steel plate harder than preset surface layer hardness as a hardened portion; and
a removal equipment configured to remove the determined hardened portion in the surface layer of the steel plate.
More preferably, the manufacturing equipment optionally further comprises a demagnetization equipment configured to demagnetize the surface layer of the steel plate or the whole steel plate, between the rolling equipment and the inspection line. In this way, a decrease in mechanical property measurement or evaluation accuracy can be prevented.
An example of a manufacturing method for the substance 1 is as follows:
A steel plate manufacturing method comprising:
a rolling step of rolling a slab to obtain a steel plate;
an inspection step of measuring surface layer hardness of the steel plate by a mechanical property measuring method according to the present disclosure, and determining, from the measured surface layer hardness of the steel plate, a part in a surface layer of the steel plate harder than preset surface layer hardness as a hardened portion; and
a removal step of removing the determined hardened portion in the surface layer of the steel plate.
More preferably, the manufacturing method optionally further comprises a demagnetization step of demagnetizing the surface layer of the steel plate or the whole steel plate, between the rolling step and the inspection step. In this way, a decrease in mechanical property measurement or evaluation accuracy can be prevented.
In the foregoing steel plate manufacturing method, the rolling step is performed on the continuous slab at 850° C. or more in order to obtain a predetermined shape and mechanical property. After the rolling step, quenching and annealing may be performed as a heat treatment step. It is well known that electromagnetic feature values such as incremental magnetic permeability, coercive force, and Barkhausen noise correlate with the mechanical property of a steel material. It is therefore preferable to measure the electromagnetic feature values as the physical quantities of the measured object 101 in a state in which the microstructure of the steel material has been established through the heat treatment step. Here, the measured object 101 denotes the steel plate and a film on the surface of the steel plate. Examples of the film on the surface of the steel plate include iron oxide films such as scale and mill scale, organic coatings such as resin coating, plating films, and chemical conversion coatings. Given that the mechanical property is determined by quenching and annealing, the temperatures before and after the quenching, the temperatures before and after the annealing, etc. may be further measured and used as physical quantities of the measured object 101 in the manufacturing method.
The presently disclosed techniques may be applied to a management method for the substance 1, to inspect and thus manage the substance 1. Specifically, the substance 1 having the film 2 on its surface and prepared beforehand is inspected in an inspection step, and managed in a management step of classifying the substance 1 based on the inspection result in the inspection step. The inspection step includes the foregoing measurement step, selection step, generation step, and calculation step according to the present disclosure, and the mechanical property of the substance 1 having the film 2 on its surface and prepared beforehand as the measured object 101 is calculated. Alternatively, the inspection step calculates the mechanical property of the substance 1 having the film 2 on its surface as the measured object 101, using the mechanical property measuring apparatus according to the present disclosure. In the management step which follows, the substance 1 can be managed. In the management step, the manufactured substance 1 is classified according to a criterion designated beforehand based on the mechanical property of the substance 1 obtained by the calculation step or the mechanical property measuring apparatus 100, and thus managed. For example, in the case where the substance 1 is a steel material and the mechanical property of the substance 1 is the hardness of the steel material, the steel material can be classified into a class corresponding to the hardness. With such management method for the substance 1, the mechanical property can be measured accurately through the physical quantities, so that the substance 1 of high quality can be provided. In the case where the mechanical property of the substance 1 obtained by the mechanical property measuring apparatus 100 or the calculation step is the mechanical property of the surface layer of the substance 1, a more appropriate calculation model can be generated by the calculation model generating unit 81 or the selection step and the generation step (steps S12 and S13), so that the foregoing effects can be further enhanced.
An example of a management method for the substance 1 is as follows:
A steel plate manufacturing method comprising:
an inspection step of measuring surface layer hardness of a steel plate by a mechanical property measuring method according to the present disclosure, and determining, from the measured surface layer hardness of the steel plate, a part in a surface layer of the steel plate harder than preset surface layer hardness as a hardened portion; and
a management step of classifying the steel plate by an area and/or a position of the determined hardened portion in the surface layer of the steel plate.
Number | Date | Country | Kind |
---|---|---|---|
2020-103341 | Jun 2020 | JP | national |
Filing Document | Filing Date | Country | Kind |
---|---|---|---|
PCT/JP2021/022595 | 6/14/2021 | WO |