Patent Application
20240302040
This application claims the benefit of Italian Patent Application No. 102023000004521, filed on Mar. 10, 2023, which application is hereby incorporated herein by reference.
The present disclosure relates to a sensor device for flame presence detection, in particular for an ignition system. Furthermore, it relates to an ignition system comprising the sensor device, to a flame presence detection method and to a computer program product thereof.
As is known, ignition systems use sparks to ignite flammable fuel-air mixtures. These sparks are usually generated electrically, for example using piezoelectric material. This ignition mode is exploited, for example, in the combustion chambers of combustion engines, as well as in kitchen hobs and gas cookers.
An important functionality for ignition systems is the possibility of shutting off the fuel supply when the flame is not lit or goes out. This avoids that, in the absence of a flame, the fuel continues to be supplied to the ignition system and is dispersed in the environment, thus risking the occurrence of accidents such as fires or bursts in the environments where the fuel is dispersed.
Furthermore, a cause of lack of flame ignition may be the degradation over time or a malfunction of a spark generator of the ignition system. The impossibility of generating a spark by the spark generator causes for example a non-start condition of the combustion engine burner and, if fuel is introduced during this non-start condition, an explosion of the same burner may potentially occur. Therefore, the spark generator may be one of the causes responsible for the lack of flame ignition.
The issue of controlling the spark generation and flame detection for the correct and safe use of the ignition system is therefore known.
In other words, performing the following activities are needed: shutting off the fuel supply if the flame is not detected, shutting off the fuel supply if the spark generator does not work correctly, and promptly calling for maintenance in case of malfunction of the spark generator.
Nowadays, there exist several solutions for flame detection.
For example, known solutions are based on optical detection and exploit analyses of ultraviolet radiation (UV) and infrared radiation (IR), individually or in combination with each other. Infrared sensors are the most common technique to verify the presence of a flame and use optical filters to filter the radiation to be detected as a function of the fuel used. In fact, exemplarily considering the case of domestic boilers or industrial burners which use methane as fuel, the methane-air mixture is converted through combustion into carbon dioxide and water; as a result, the optical detection at wavelengths corresponding to the presence of carbon dioxide allows to correlate an increase in the concentration of carbon dioxide molecules to the condition of igniting a flame. However, these optical approaches have detection accuracy issues mainly due to the presence of water vapor and other elements (e.g., dirt on optical filters and IR detectors) which negatively affect the measurement. Furthermore, radiation generated by heat sources other than sparks and flames (e.g., the sun, surrounding hot bodies, incandescent lamps, etc.) causes a high risk of false positives.
Another approach for flame detection is based on visible radiation sensors, i.e. on digital cameras and video processing. In this case, however, sophisticated and expensive electronics is required for real-time processing of the acquired images (in detail, for high-frequency frame grabbing) and this makes this solution acceptable only for specific applications.
Furthermore, monitoring of the spark generator activity may also be based on acoustic analysis through microphone. However, in this case a fast, high-frequency data processing is required to analyze the audio signal and classify it based on the presence or absence of the spark. However, this solution is affected by the ambient acoustic noise and therefore has poor accuracy and reliability.
The aim of the present disclosure is to provide a sensor device for flame presence detection, an ignition system comprising the sensor device, a flame presence detection method and a non-transitory computer program product including computer instructions thereof, which overcome the drawbacks of the prior art.
According to the present disclosure there are provided a sensor device for flame presence detection, an ignition system comprising the sensor device, a flame presence detection method and a non-transitory computer program product including computer instructions thereof, as defined in the annexed claims.
For a better understanding of the present disclosure, preferred embodiments are now described, purely by way of non-limiting example, with reference to the attached drawings, wherein:
In particular, the Figures are shown with reference to a triaxial Cartesian system defined by an axis X, an axis Y and an axis Z, orthogonal to each other.
In the following description, elements common to the different embodiments have been indicated with the same reference numerals.
For example, and as better shown in
By way of non-limiting example, the ignition system 100 may be used in an apparatus such as a kitchen hob, as shown in
As shown in
The dispensing device 102 is coupleable to a fuel source (e.g., gas cylinder, petrol tank, etc.) which contains fuel to be burned to produce the flame 12. As a result, in use the dispensing device 102 receives the fuel from the fuel source.
The dispensing device 102 is further controllable, for example by the main control unit 106, to dispense the fuel received from the fuel source. For example, the fuel may be dispensed at the stove of the kitchen hob or in the boiler combustion chamber, or in any case, in the place where the flame 12 is to be generated.
The spark generator 104 is controllable, for example by the main control unit 106, to generate a spark for igniting the fuel, giving rise to the flame 12. For example, the spark generator 104 comprises an electronic spark generator or a piezoelectric element which, when mechanically compressed, generates an electric discharge in the air and therefore a spark.
For this purpose, the spark generator 104 is coupled to the dispensing device 102 in such a way as to produce the flame 12 when the spark is generated at the fuel dispensed by the dispensing device 102. In other words, and in a per se known manner, the ignition system 100 is designed in such a way that the spark generator 104 and the dispensing device 102 are sufficiently close to each other so that the spark generated by the spark generator 104 triggers the ignition of the fuel coming from the dispensing device 102.
Furthermore, the sensor device 10 is operatively coupled to the dispensing device 102 and to the spark generator 104 so as to detect the presence of the flame 12 generated thereby. In other words, the sensor device 10 is arranged in proximity to the dispensing device 102 and to the spark generator 104, at a distance such as to allow the sensor device 10 to receive the ionized air from the combustion and therefore to detect the flame 12 (e.g., a few mm or cm from where the flame 12 develops).
In detail, in use the main control unit 106 receives from the sensor device 10 a flame signal SF, indicative of the presence or absence of the flame 12 and better described hereinbelow and controls the dispensing device 102 as a function the flame signal SF, in such a way as to prevent fuel dispensing if the flame signal SF is indicative of the absence of the flame 12.
In other words, the main control unit 106 blocks the flow of fuel dispensed by the dispensing device 102 when the sensor device 10 detects no flame 12, thereby preventing an accumulation of unburned fuel which may lead to uncontrolled bursts and fires. Considering the exemplary case of the combustion boiler, the main control unit 106 prevents that, in the absence of a flame 12 which consumes the fuel, the combustion chamber is filled with fuel beyond the allowed limit to the point that, when a spark is actually generated, the combustion in the combustion chamber is violent and uncontrolled and leads to damaging the boiler. Similarly, in the case of the hob, the main control unit 106 prevents the air in the kitchen from saturating with fuel, thus avoiding the risk of explosions and fires in the house.
With reference again to
The sensor device 10 comprises: a carbon dioxide sensor 20 configured to detect a concentration of carbon dioxide in the air, generated by the flame 12; a fuel sensor 30 configured to detect the presence of fuel combustion; an electrostatic charge variation sensor 40 including a first and a second electrode 40a and 40b spaced from each other and configured to detect respective electrostatic charge variations generated by the flame 12; and a control unit 50 which is operatively coupled (e.g., electrically coupled, for example through respective electrical connections) to the carbon dioxide sensor 20, the fuel sensor 30 and the electrostatic charge variation sensor 40, and is configured to implement, in use, a detection method 200 for detecting the flame 12, better described hereinbelow.
In particular, the sensor device 10 comprises a tubular body 14 which has an inlet opening 14a and an outlet opening 14b. The openings 14a, 14b are fluidically coupled to each other through a fluidic channel 15 of the tubular body 14, which extends through the tubular body 14 and which defines a fluidic path between the inlet opening 14a and the outlet opening 14b. For example, the inlet opening 14a has a distance from the flame 12 which is smaller than the respective distance of the outlet opening 14b from the flame 12.
In greater detail, the tubular body 14 (e.g., of cylindrical shape) has a first end 14′ and a second end 14″ opposite to each other along a longitudinal axis 16 of the tubular body 14. For example, the tubular body 14 has a main extension along the longitudinal axis 16, so that the fluidic channel 15 extends mainly along the longitudinal axis 16 (exemplarily shown in
Orthogonally to the longitudinal axis 16 (therefore parallel to a plane XZ defined by the axes X and Z), the tubular body 14 may have an annular section which, by way of non-limiting example, has a circular shape. Nevertheless, other shapes may similarly be chosen, for example a triangular, square, etc. shape.
In detail, the tubular body 14 has an internal surface 17a, facing the fluidic channel 15 and delimiting the latter, and an external surface 17b facing towards an environment external to the tubular body 14.
The fluidic channel 15 is in fluidic communication with the external environment and therefore with the flame 12, if any. In other words, the air present in the fluidic channel 15 is indicative of the presence or absence of the flame 12. In fact, when the flame 12 is present which heats the surrounding air causing convective motions therein, the fluidic channel 15 is flown through by an air flow which is due to the flame 12 and which transfers from the inlet opening 14a to the outlet opening 14b. This air flow is heated by the flame 12 and comprises both gases which are indicative of the presence of combustion (in detail, carbon dioxide generated by combustion and unburned fuel) and ionized particles which cause electrostatic charge variations.
The control unit 50, the carbon dioxide sensor 20, the fuel sensor 30 and the electrostatic charge variation sensor 40 are accommodated in the tubular body 14, for example they are carried thereby and are fixed thereto.
In detail, the carbon dioxide sensor 20 and the fuel sensor 30 extend at least partially into the fluidic channel 15, so as to acquire information on the air present in the fluidic channel 15.
In use, the carbon dioxide sensor 20 detects the concentration of carbon dioxide in the air by measuring the absorption of light radiation at a carbon dioxide optical absorption wavelength (equal to about 4.3 μm). Carbon dioxide is a common product of any combustion, therefore the concentration of carbon dioxide in the air may be indicative of the presence of the flame 12.
In greater detail, the carbon dioxide sensor 20 comprises a light radiation emitter 22, a carbon dioxide optical filter 23 and a carbon dioxide detector 24, which are aligned in succession to each other along a carbon dioxide alignment axis 26 extending through the fluidic channel 15 and exemplarily shown as parallel to the axis Y.
The light radiation emitter 22 comprises, for example, a light-emitting diode (LED) and in use emits light radiation. In detail, the light radiation emitted by the light radiation emitter 22 has a wavelength comprised between about 2 μm and about 15 μm, so as to cover the carbon dioxide optical absorption wavelength.
In greater detail, the light radiation emitter 22 faces the fluidic channel 15 and is controlled in use by the control unit 50 to emit the light radiation through the fluidic channel 15.
The carbon dioxide optical filter 23 is an optical filter, in particular a band-pass filter (in detail, a narrow-band filter) and designed in such a way as to transmit the light radiation with a wavelength comprised in a carbon dioxide wavelength range (which comprises the carbon dioxide absorption wavelength at about 4.3 μm) and to block the light radiation with a wavelength not comprised in the carbon dioxide wavelength range. For example, the bandwidth at half of the transmission peak is equal to about 5% of the carbon dioxide wavelength, therefore it is equal to about 0.2 μm (e.g., the carbon dioxide optical filter 23 has a lower cut-off frequency equal to about 4.175 μm and an upper cut-off frequency equal to about 5.512 μm). As a result, in use the carbon dioxide optical filter 23 is transparent to the light radiation emitted by the light radiation emitter 22 and with a wavelength equal to about the carbon dioxide absorption wavelength, while blocking the remaining part of the light radiation spectrum.
The carbon dioxide detector 24 is an optical detector, in particular sensitive to infrared radiation (IR) and for example of MEMS type.
For example, the carbon dioxide detector 24 is a thermal MOS (TMOS) transistor. The TMOS transistor is a known field effect transistor device and typically used in sensor applications to determine the amount of infrared radiation (IR) emitted by an object or a body under examination. The IR radiation emitted, received by the TMOS transistor, causes the generation of charge carriers in the conductive channel of the TMOS transistor and, therefore, a variation of the output current of the latter. Greater details regarding the TMOS transistor may be found for example in the document EP3689816 of the present Applicant.
The carbon dioxide detector 24 is arranged at a distance from the light radiation emitter 22. In particular, the carbon dioxide detector 24 may extend on the opposite side of the fluidic channel 15 with respect to the light radiation emitter 22, along the carbon dioxide alignment axis 26, so as to maximize the optical path of the light radiation between the light radiation emitter 22 and the carbon dioxide detector 24 and therefore to maximize the optical absorption by the carbon dioxide of the light radiation at the carbon dioxide absorption wavelength.
The carbon dioxide optical filter 23 faces the fluidic channel 15 and is interposed, along the carbon dioxide alignment axis 26, between the light radiation emitter 22 and the carbon dioxide detector 24. In detail, the carbon dioxide optical filter 23 is arranged at the carbon dioxide detector 24, for example it is closer to the latter than to the light radiation emitter 22.
As a result, the carbon dioxide detector 24 detects in use the light radiation emitted by the light radiation emitter 22 and filtered by the carbon dioxide optical filter 23 and generates a carbon dioxide signal SA (an electrical signal, for example a digital signal) indicative of the concentration of carbon dioxide in the fluidic channel 15.
In particular, generating the carbon dioxide signal SA starting from the light radiation impinging on the carbon dioxide detector 24 occurs in a per se known manner. For example, the concentration of carbon dioxide may be calculated as a function of the light radiation absorption intensity at the carbon dioxide absorption wavelength (given by the ratio between the light radiation intensity measured by the carbon dioxide detector 24 and the total intensity of light radiation provided by the light radiation emitter 22, at the carbon dioxide absorption wavelength) and as a function of a relative distance (known as decided during the design step) between the light radiation emitter 22 and the carbon dioxide detector 24 (e.g., using the well-known Beer-Lambert law). By way of example,
In use, the fuel sensor 30 detects the fuel combustion. This occurs by measuring the emission spectrum of the flame 12, and in particular by analyzing the emission of the flame 12 at an optical emission wavelength identifying the fuel used. In fact, the fuel combustion generates an optical emission spectrum which identifies the fuel used (i.e. it has one or more emission peaks at respective wavelengths which are specific for each fuel). As a result, detecting an absorption peak at the optical emission wavelength of interest for the fuel used entails detecting that the combustion of the fuel is in progress, and therefore allows the presence of the flame to be deduced.
In the following, reference is exemplarily made to the case in which the fuel is methane (CH4), with a fuel emission wavelength equal to about 3.3 μm; however, other fuels may be similarly considered (e.g., LPG with a fuel emission wavelength comprised between about 400 μm and about 440 μm, or hydrogen) and the fuel sensor 30 may be modified accordingly (e.g., so as to emit and detect ultraviolet radiation in the case of LPG). Nonetheless, these values are exemplary and may vary, as known to the person skilled in the art, as a function of different environmental factors and conditions; greater details in this regard may for example be found in the document “Infrared Emission Spectra of Flames”, by Earle K. Plyler et al, 1948.
In detail, the fuel sensor 30 comprises a fuel optical filter 33 and a fuel detector 34.
The inlet opening 14a, the fuel optical filter 33, and the fuel detector 34 are aligned in succession to each other along a fuel alignment axis 36 extending through the fluidic channel 15 and exemplarily shown as parallel to the axis Y and to the carbon dioxide alignment axis 26. In this manner, the radiation emitted by the flame 12 arrives at the fuel detector 34 first traversing the inlet opening 14a and then the fuel optical filter 33.
The fuel optical filter 33 is an optical filter, in particular a band-pass filter and designed in such a way as to transmit the light radiation with a wavelength comprised in a fuel wavelength range (which comprises the emission wavelength characteristic of the fuel, hereinafter also referred to as fuel emission wavelength) and to block the light radiation with a wavelength not comprised in the fuel wavelength range. For example, the fuel wavelength range is comprised between about 2.84 μm and about 3.55 μm and therefore the fuel optical filter 33 is a narrow-band filter and is centered at about 3.3 μm. As a result, in use the fuel optical filter 33 is transparent to the light radiation emitted by the flame 12 and with a wavelength equal to about the fuel emission wavelength, while blocking the remaining part of the spectrum of the radiation emitted by the flame 12.
The fuel detector 34 is an optical detector, in particular sensitive to infrared radiation (IR) and for example of MEMS type. For example, the fuel detector 34 is a TMOS transistor.
The fuel detector 34 is arranged at a distance from the inlet opening 14a. In particular, the fuel detector 34 may extend on the opposite side of the fluidic channel 15 with respect to the inlet opening 14a, along the fuel alignment axis 36.
The fuel optical filter 33 faces the fluidic channel 15 and is interposed, along the fuel alignment axis 36, between the inlet opening 14a and the fuel detector 34. In detail, the fuel optical filter 33 is arranged at the fuel detector 34, for example it is closer to the latter than to the inlet opening 14a.
As a result, the fuel detector 34 detects in use the emission spectrum of the radiation emitted by the flame 12 and filtered by the fuel optical filter 33 and generates a fuel signal SC (an electrical signal, for example a digital signal) indicative of the presence of the combustion. For example, the fuel signal SC assumes at each instant a respective value corresponding to the amplitude at that instant of the emission peak characteristic of fuel combustion, which is substantially proportional to the amount of fuel burned and, as a result, is proportional to the intensity of the flame 12 at that instant.
In particular, generating the fuel signal SC starting from the radiation impinging on the fuel detector 34 occurs in a per se known manner. For example, the fuel concentration may be calculated as a function of the measured emission intensity of the radiation emitted by the flame 12 at the fuel emission wavelength. By way of example,
In use, the electrostatic charge variation sensor 40 detects environmental electrostatic charge variations indicative of the flame 12 as they are caused by the presence in the air of particles which have been ionized by the combustion which generates the flame 12.
In detail, the first electrode 40a of the electrostatic charge variation sensor 40 extends at the first end 14′ of the tubular body 14 and the second electrode 40b of the electrostatic charge variation sensor 40 extends at the second end 14″ of the tubular body 14, so as to maximize the relative distance between the electrodes 40a and 40b and therefore so as to maximize the difference of electrostatic charge variations detected by the electrodes 40a and 40b.
In detail, each electrode 40a, 40b may have a metal surface or be of a totally metal material coated with a dielectric material, or still have a metal surface arranged under an external case of the tubular body 14. In any case, during use, each electrode 40a, 40b is electrostatically coupled to the environment wherein the sensor device 10 and the flame 12 are present, in order to detect the electrostatic charge variation induced by the fuel combustion.
According to one embodiment, each electrode 40a, 40b is a metal element carried by the tubular body 14. Optionally, when a possible use of the sensor device 10 in a humid environment is envisaged, each electrode 40a, 40b is inserted inside a waterproof case or in any case it is shielded by means of one or more protective layers, thereby preventing a direct contact of the electrode 40a, 40b with water or humidity: in this case, the waterproof case or the one or more protective layers are of a material (e.g., dielectric material, such as plastic material) such as not to shield the electrostatic charge generated by the flame 12, which is to be acquired by the electrode 40a, 40b. Other embodiments are possible, as evident to the person skilled in the art, so that the electrodes 40a, 40b are electrostatically coupled to the flame 12 during use.
In use, each electrode 40a, 40b detects a respective electrostatic charge variation caused by the combustion which originates the flame 12, and generates a respective detection signal SR indicative of the electrostatic charge variation.
Furthermore, the electrostatic charge variation sensor 40 generates at output an electrostatic charge variation signal SQ (an electrical signal, for example a digital signal) indicative of a difference between the electrostatic charge variations detected by the first and the second electrodes 40a, 40b. In detail, the electrostatic charge variation signal SQ is a function of the mutual difference of the detection signals SR measured by the electrodes 40a, 40b. For example, and in a manner not shown, the electrostatic charge variation sensor 40 may comprise an analog subtractor (e.g., a differential amplifier) or a sensor control unit to calculate the difference of the detection signals SR measured by the electrodes 40a, 40b.
For illustrative purposes,
On the other hand,
In a manner not shown, the control unit 50 (such as a processor, e.g., a microprocessor, a microcontroller or a dedicated calculation unit) may comprise, coupled to each other, a data storage unit (such as a memory, e.g. a non-volatile memory) for storing the acquired data, and a processing unit including computer instructions for processing the acquired data. For example, the control unit 50 is integrated into the tubular body 14.
In use, the control unit 50 implements the detection method 200 of the presence of the flame 12.
An embodiment of the detection method 200 is shown in
In detail, the detection method 200 is performed iteratively, so as to update the information on the presence or absence of the flame 12 in real time. For the sake of simplicity, an iteration of the detection method 200, hereinafter referred to as current iteration and also indicated with the reference k, is described hereinbelow.
At a step S05 of the detection method 200, the carbon dioxide signal SA is acquired through the carbon dioxide sensor 20, as previously described.
At a step S10 of the detection method 200, the fuel signal SC is acquired through the fuel sensor 30, as previously described.
At a step S15 of the detection method 200, the electrostatic charge variation signal SQ is acquired through the electrostatic charge variation sensor 40, as previously described. In particular, the electrostatic charge variation signal SQ is generated by the electrostatic charge variation sensor 40, for example by acquiring the detection signals SR through the electrodes 40a, 40b, calculating a difference between the detection signals SR and optionally digitizing, filtering and/or amplifying the signal thus obtained.
Steps S05-S15 may be performed in succession to each other or in parallel to each other.
In particular, the signals SA, SC and SQ are acquired through respective scrolling buffers. In detail, at each iteration the signals SA, SC and SQ are acquired in a time window for example having a predefined duration equal to a time window period (for example equal to a few seconds). In other words, at each iteration the signals SA, SC and SQ are considered with a time duration equal to the time window period and therefore each comprising the last sample acquired (i.e. the sample acquired at the current iteration, hereinafter also referred to as current sample) and a plurality N of samples preceding the current sample. At each iteration, the oldest sample of each signal SA, SC and SQ is discarded and a new sample of each signal SA, SC and SQ is stored in the respective buffer.
At a step S20 consecutive to step S15, a quantized signal SQ′ is determined by processing the electrostatic charge variation signal SQ.
In particular, the quantized signal SQ′ is substantially a filtered version of the electrostatic charge variation signal SQ, wherein the strong oscillations that identify the electrostatic charge variation signal SQ are not present.
One embodiment of determining the quantized signal SQ′ is better described with reference to
In detail and as shown in
At a sub-step S20B consecutive to sub-step S20A, the baseline B is subtracted from the electrostatic charge variation signal SQ to calculate a variability signal SV, which contains an information content on the flame 12 present in the oscillations of the electrostatic charge variation signal SQ around the baseline B without however depending on the same baseline B anymore.
At a sub-step S20C consecutive to sub-step S20B, a normalized signal SN is determined through a comparison (in detail, a punctual comparison) of the variability signal SV with a charge variation threshold value. The normalized signal SN is a binary signal and allows the variability of the oscillations of the variability signal SV to be reduced to a variation between only two values. In detail, at each time instant the normalized signal SN assumes a first value (e.g., 0) if the corresponding value of the variability signal SV is lower than the charge variation threshold value, or it assumes a second value (e.g., 1) if the corresponding value of the variability signal SV is greater than, or equal to, the charge variation threshold value. In particular, the charge variation threshold value is chosen heuristically as a function of the trade-off between maximum sensitivity to electrostatic charge variations (which requires a minimum charge variation threshold value) and minimum influence on the background noise measurement (which requires a maximum charge variation threshold value). By way of non-limiting example, the charge variation threshold value is equal to the multiplication of the background noise (or noise level, NL) by a multiplication factor, for example equal to about 5 times. As a result, at sub-step S20C the background noise of the variability signal SV is calculated, the charge variation threshold value is calculated, the variability signal SV is compared with the charge variation threshold value and the value of the variability signal SV as a function of this comparison is determined at each instant.
At a sub-step S20D consecutive to sub-step S20C, a clustering interval SN,k of the normalized signal SN is determined, corresponding to the current iteration k and therefore to the current time instant tk. The clustering interval SN,k is a portion of the normalized signal SN, obtained through buffering on M samples (values) of the normalized signal SN. In detail, the clustering interval SN,k comprises the value of the normalized signal SN corresponding to the time instant considered (i.e. to the current time instant tk of the current iteration k) and a predefined plurality M′ of values of the normalized signal SN corresponding to a respective plurality M′ of time instants preceding (in detail, immediately preceding) the time instant considered. In other words, the clustering interval SN,k comprises the M values of the normalized signal SN in the time period tk-M′≤t≤tk, with M=M′+1.
At a sub-step S20E consecutive to sub-step S20D, it is determined which value of the normalized signal SN, between the first value (e.g., 0) and the second value (e.g., 1), has the greatest number of occurrences in the clustering interval SN,k considered and corresponding to the current iteration k. In other words, the numbers of occurrences of 0 and 1 in the clustering interval SN,k are calculated and it is determined which of the two values has a greater occurrence.
At a sub-step S20F consecutive to sub-step S20E, the quantized signal SQ′ is generated in such a way that, at each time instant, the quantized signal SQ′ assumes a respective first value (e.g., 0) if the value of the normalized signal SN with the greatest occurrence in the clustering interval SN,k considered is the first value (e.g., 0), or assumes a respective second value (e.g., 1) if the value of the normalized signal SN with the greatest occurrence in the clustering interval SN,k considered is the second value (e.g., 1). In other words, considering the current iteration k, if the clustering interval SN,k has more 0s than 1s, the value of the quantized signal SQ′ at the current time instant tk is set equal to 0.
In this manner a rejection of the values of the normalized signal SN may be obtained which are insulated over time and are generated for example by noise in the measurement, allowing greater accuracy in the detection of the flame 12 at the expense of a greater latency in the detection (for example, in the transition from lit to extinguished flame the quantized signal SQ′ shows a delay in the change of value with respect to the real situation which is due to the buffering performed at step S20D).
As a result, the number M of samples present in the clustering interval SN,k is heuristically chosen as a function of the trade-off between maximum rejection of the insulated values (i.e. false positives or false negatives) of the normalized signal SN (which requires a maximum number M of samples) and minimum latency in the transition from one value to another of the quantized signal SQ′ (which requires a minimum number M of samples). By way of non-limiting example, the number M is equal to a few tens or hundreds of samples.
With reference again to
In greater detail, the aggregate datum IA is defined, at each time instant, through a respective set of aggregate datum points (indicated by IA,k with reference to the current iteration k). In other words, each set of aggregate datum points IA,k is associated with a respective time instant.
The set of aggregate datum points IA,k of the current iteration k comprises a plurality P of aggregate datum points IA,k. Each aggregate datum point IA,k is defined by a respective value of the carbon dioxide signal SA, a respective value of the fuel signal SC and a respective value of the electrostatic charge variation signal SQ, taken at a respective time instant (different for each aggregate datum point IA,k of the set). These values of the signals SA, SC and SQ which form the aggregate datum point IA,k of the current iteration k are the values of the signals SA, SC and SQ corresponding to a time instant which may be the time instant tk of the current iteration k considered or a time instant among a plurality P′ of time instants preceding (in detail, immediately preceding) the time instant tk of the current iteration k considered. More in detail, since each aggregate datum point IA,k corresponds to a respective time instant, the set of aggregate datum points IA,k of the current iteration k is indicative of portions of the carbon dioxide signal SA, the fuel signal SC and the electrostatic charge variation signal SQ comprised in the time period tk-P′≤t≤tk, with P=P′+1. In other words, these portions of the signals SA, SC and SQ corresponding to the set of aggregate datum points IA,k are temporally defined by the time instant tk of the current iteration k (corresponding to the set of aggregate datum points IA,k considered) and by the plurality P′ of time instants preceding the time instant tk.
Each aggregate datum point IA,k therefore comprises three values and as a result is associable with a point in a three-dimensional space defined by three axes corresponding respectively to the carbon dioxide signal SA, the fuel signal SC and the electrostatic charge variation signal SQ. In view of this, the set of aggregate datum points IA,k may be graphically represented as a cluster of points in this 3D space. Examples of these 3D graphs are shown in
Furthermore, since each set of aggregate datum points IA,k corresponds to a respective iteration and therefore to a respective time instant considered, the variations over time of the signals SA, SC and SQ are graphically displayable as a video wherein each frame coincides with the 3D graph of a respective set of aggregate datum points IA,k. In detail, two consecutive frames differ from each other by only one aggregate datum point IA,k (as in the transition between the two frames the oldest sample of the first frame is discarded and the sample acquired at the time instant of the second frame is added).
At a step S30 consecutive to step S25, the flame signal SF, indicative of the presence or absence of the flame 12, is generated, as a function of the aggregate datum IA.
In particular, generating the flame signal SF is better described with reference to
Furthermore, generating the flame signal SF is described herein also with reference to
In detail and as shown in
At a sub-step S30B consecutive to sub-step S30A, a distance (or centroidal distance) DC of the centroid C from the reference centroid Cref is calculated. For example, the distance DC is shown in
At a sub-step S30C consecutive to sub-step S30B, the distance DC is compared with a threshold distance DT.
At a sub-step S30D consecutive to sub-step S30C, the value of the flame signal SF corresponding to the current time instant tk is assigned. In particular, the flame signal SF assumes, at the current iteration k, a respective first value (e.g., 0) if the distance DC is smaller than the threshold distance DT, or assumes a respective second value (e.g., 1) if the distance DC is greater than, or equal to, the threshold distance DT. In particular, the first value of the flame signal SF is indicative of the absence of the flame 12 and the second value of the flame signal SF is indicative of the presence of the flame 12. In other words, if the set of aggregate datum points IA,k has a large distance from the reference set of aggregate datum points IA,k the presence of the flame 12 is determined, whereas if the set of aggregate datum points IA,k has a reduced distance from the reference set of aggregate datum points IA,k the absence of the flame 12 is determined.
As a result, the threshold distance DT is chosen heuristically as a function of the trade-off between maximum sensitivity in recognizing the presence of the flame 12 (which requires a minimum threshold distance DT) and minimum number of false positives (which requires a maximum threshold distance DT). By way of non-limiting example, the threshold distance DT is comprised, in normalized value, between about 0.6 and about 0.7.
As may be seen in
For purposes of further illustration,
In addition to what has been previously described, the detection method 200 may also verify whether sparks generated by the spark generator 104 are detected, so as to signal, if necessary, an anomalous operating condition of the spark generator 104. This is achieved through steps S35 and S40 shown in
In particular, at step S35 consecutive to step S30, a condition of generating a spark for igniting the flame 12 is verified, as a function of the electrostatic charge variation signal SQ.
In particular, this comprises verifying whether the electrostatic charge variation signal SQ has a spark generation pattern. The spark generation pattern is indicative of the generation of a spark.
An example of the spark generation pattern is shown in
In fact, the shape of the spark generation pattern 70 depends on different factors, such as the environment wherein the spark is generated (e.g., the shape, the dimensions and the materials of the combustion chamber in the case of the combustion engine) and the spark generator 104 used (e.g., its shape, the technology used to generate the spark, etc.). As a result, the case shown in
Furthermore, the spark generation pattern 70 may not always be indicative of the spark since in certain cases it may have a non-repeatable and standard shape and therefore not easily identifiable through the previously described approach; for example, this may occur in case the spark generator 104 is based on piezoelectric technology. In view of this, the shape of the spark generation pattern 70 and the criteria which define the condition of generating a spark are obtained heuristically during the design step.
In view of what has been discussed, at step S35 the electrostatic charge variation signal SQ is processed in a per se known manner to identify any peaks thereof (i.e. positive peaks or negative peaks) and, subsequently, to verify whether the criteria on the peaks which define the condition of generating a spark are confirmed. In detail, at step S35 there are identified, if any, a number of peaks of the electrostatic charge variation signal SQ and, for each peak, a respective maximum amplitude with respect to the baseline B and a respective time position (or instant) of the maximum value of the peak. Greater details regarding the identification modes of such peaks may be found for example in the patent document US 2022/0366768 A1, of the present Applicant.
At step S40 consecutive to step S35, a spark warning signal is generated if the generation of a spark is confirmed and, for a predefined time interval starting from the generation of a spark, the flame signal SF is indicative of the absence of the flame 12. For example, the spark warning signal is a digital electrical signal which has a first value by default (e.g., 0) but which may assume a second value (e.g., 1) which allows for example a visual or sound alarm to be generated to signal a malfunction of the spark generator 104. In detail, the spark warning signal is generated when the spark generation pattern 70 is detected and if, within the predefined time interval starting from the generation of a spark (e.g., starting from the time instant corresponding to a predefined peak of the spark generation pattern 70), no flame is detected.
From an examination of the characteristics of the invention made according to the present invention, the advantages that it affords are evident.
In particular, the sensor device 10 and the detection method 200 allow data to be acquired through a plurality of sensors of different types and allow data fusion to be performed on these data in order to detect the presence or absence of the flame 12 with high accuracy, preventing the risk of fires or bursts due to uncontrolled combustions.
The high accuracy results from the redundancy of information (different from each other) acquired through the sensors 20, 30 and 40, which allows specific limits of each of these sensors 20, 30 and 40 to be compensated for. For example, the sole carbon dioxide sensor 20 carries out a measurement of carbon dioxide, whose concentration however depends on the environment in which the sensor device 10 is present (e.g., it may be higher in a closed and crowded room while it may be lower in the open air); as a result, this measurement is strongly influenced by factors external to the flame 12. Furthermore, the fuel sensor 30 performs, to detect the fuel concentration, an optical measurement of the emission spectrum of the flame 12 at wavelengths comprised in the IR; as a result, each hot body external and close to the sensor device 10 (e.g., people, animals or metal elements which have previously been heated by a heat source) emits radiation in the IR which may influence the measurement. Furthermore, the electrostatic charge variation sensor 40 is sensitive to any electrostatic charge variation induced in the environment external to the sensor device 10, therefore also this measurement is not sufficiently accurate if considered alone. Instead, the strategy of aggregating the signals coming from these sensors 20, 30, 40 allows to overcome the intrinsic limits of each of them and to increase the detection accuracy.
Furthermore, the present solution is immune to ambient acoustic noise, which may instead influence the known detection performed acoustically (e.g., through microphones).
The detection method 200 also requires lower computational cost, energy consumption and time compared to the solutions currently on the market with similar accuracies.
Furthermore, the sensor device 10 allows to detect a flame 12 which is even several tens of centimeters away.
Furthermore, generating the spark warning signal is useful for signaling a malfunction of the spark generator 104 which could lead to bursts and fires of the fuel. In fact, exemplarily considering the case of a hob, when a user manually controls the spark generator 104 to ignite the fire, the dispensing device 102 starts to dispense the fuel and at the same time the spark generator 104 generates the spark. If a malfunction of the spark generator 104 is present and the user repeats the gesture of igniting the flame 12 several times, an accumulation of unburned fuel may be created in the air which, as soon as a spark is generated, generates a violent combustion which may damage surrounding objects and people.
Finally, it is clear that modifications and variations may be made to the invention described and illustrated herein without thereby departing from the scope of the present invention, as defined in the attached claims.
For example, the different embodiments described may be combined with each other so as to provide further solutions.
Furthermore, the sensor device 10 may comprise a temperature sensor 60 which, in use, detects the temperature of the air in the fluidic channel 15 or in the environment external to the sensor device 10. As evident, the temperature of the air depends on the presence or absence of the flame 12 which, when present, heats the surrounding air. In this case, the control unit 50 is also operatively coupled to the temperature sensor 60. In particular, the temperature sensor 60 is carried by the tubular body 14 (e.g., it is accommodated therein). For example, the temperature sensor 60 faces the fluidic channel 15 so as to detect the temperature of the air flow which flows therethrough.
In case the temperature sensor 60 is also present, the detection method 200 further comprises (in a manner not shown) acquiring, through the temperature sensor 60, the temperature signal ST indicative of the temperature of the air and, if the temperature signal ST is indicative of the absence of the flame 12 while the flame signal SF is indicative of the presence of the flame 12, the modification of the flame signal SF such that it is indicative of the absence of the flame 12. Alternatively 0 in addition to the modification of the flame signal SF, a sensor warning signal may be generated which is indicative of an inconsistency between the measurements of the sensors 20, 30, 40 and 60 and therefore of the malfunction of one of the sensors 20, 30, 40 and 60. In other words, the temperature control is a further verification which is carried out through a further sensor in order to signal a malfunction condition of one of the sensors 20, 30, 40 and 60 and/or to add a further condition to the determination of the flame signal SF (which is indicative of the presence of the flame 12 only if both the temperature sensor 60 and the sensors 20, 30 and 40 detect the presence of the flame 12).
Furthermore, the electrostatic charge variation sensor 40 may have two single-ended data acquisition channels (in detail, one for each electrode) or it may comprise two electrostatic charge variation sub-sensors each including a respective electrode (i.e. the sub-sensors have non-differential and single-ended inputs). In these cases, the electrostatic charge variation signal SQ acquired by the electrostatic charge variation sensor 40 is obtained by calculating the difference of the signals acquired through the electrodes.
Furthermore, a different embodiment of the determination of the quantized signal SQ′ of step S20 of the detection method 200 is described with reference to
In detail and as shown in
After sub-step S20C, an augmented normalized signal SNA including the normalized signal SN and additional samples having the first value of the normalized signal SN (e.g., 0) is generated (sub-step S20G). These additional samples precede and follow the normalized signal SN in such a way that the augmented normalized signal SNA comprises, in succession to each other, K additional samples, the N samples of the normalized signal SN and other K additional samples. As a result, the augmented normalized signal SNA has N+2K samples, where the first K samples and the last K samples are set to 0. In other words, SNA,i=0 for 0≤i≤K−1, SNA,i=SN,i for K≤i≤K+N−1 and SNA,i=0 for K+N≤i≤N+2K−1.
At sub-step S20H a copy signal SE equal to the augmented normalized signal SNA is generated, i.e. a copy of the augmented normalized signal SNA is stored.
At sub-step S20I a counter (or index) i is initialized to the value K (i.e. i=K).
At sub-step S20J, for the sample of the augmented normalized signal SNA identified by the index i, a first cumulative value SLi and a second cumulative value SRi are calculated. The first and the second cumulative values SLi and SRi are respectively calculated according to the following mathematical expressions SLi=Σj=i−Ki−1 SNA,j and SRi=Σj=i−1i+k SNA,j. As a result, the first and the second cumulative values SLi and SRi are indicative of the sum of the samples of the augmented normalized signal SNA which precede and, respectively, follow the sample SNA,i.
At sub-step S20K it is verified whether the sample SNA,i of the augmented normalized signal SNA is equal to the second value (e.g., 1). If the sample SNA,i is not equal to 1, the method proceeds to a sub-step S20L; otherwise, if the sample SNA,i is equal to 1, the method proceeds to a sub-step S20M.
At sub-step S20L it is verified whether the first cumulative value SLi is greater than a threshold cumulative value and whether the second cumulative value SRi is greater than the threshold cumulative value.
If both of these conditions are confirmed, a sub-step S20N is performed wherein the sample SNA,i assumes the second value (i.e., it is set to be equal to 1, it is overwritten with this value).
Otherwise, if at least one of these conditions is not confirmed, a sub-step S200 is performed wherein the sample SNA,i assumes the first value (i.e., it is set to be equal to 0, it is overwritten with this value).
At sub-step S20M, it is verified whether the first cumulative value SLi is greater than the threshold cumulative value or whether the second cumulative value SRi is greater than the threshold cumulative value.
If none of these conditions is confirmed, a sub-step S20P is performed wherein the sample SNA,i assumes the first value (i.e., it is set to be equal to 0, it is overwritten with this value).
Otherwise, if at least one of these conditions is confirmed, a sub-step S20Q is performed wherein the sample SNA,i assumes the second value (i.e., it is set to be equal to 1, it is overwritten with this value).
In particular, the threshold cumulative value is indicated in
Following sub-steps S20N-S20Q, it is verified (sub-step S20R) whether the counter i is lower than the number N.
If this is confirmed, a sub-step S20S is performed wherein the counter i is updated by adding a unit (i.e. i=i+i) and the method returns to sub-step S20J.
On the other hand, if this is not confirmed (i.e. i=N), a sub-step S20T is performed wherein the augmented normalized signal SNA updated through sub-steps S20J-S20Q is compared with the copy signal SE to verify whether they match each other. In other words, it is verified whether the copy signal SE and the augmented normalized signal SNA have, at each instant, the same value.
If the copy signal SE and the augmented normalized signal SNA do not match, a sub-step S20U is performed wherein the copy signal SE is updated so as to be equal to the augmented normalized signal SNA. Thereafter the method returns to sub-step S20I to perform a new sub-iteration.
On the other hand, if the copy signal SE and the augmented normalized signal SNA match, the quantized signal SQ′ is generated as a function of the augmented normalized signal SNA updated through sub-steps S20I-S20U. In particular, the quantized signal SQ′ is equal to the portion of the augmented normalized signal SNA that is devoid of the additional 2K samples. In other words, SQ,i′=SNA,i for K≤i≤K+N−1.
The steps of
| Number | Date | Country | Kind |
|---|---|---|---|
| 102023000004521 | Mar 2023 | IT | national |
