Patent Application
20190258759
The present invention relates to a total integration analysis model assistance device.
The background art of the relevant technical field is disclosed in Japanese Unexamined Patent Application Publication No. 2002-259888 (Patent Literature 1). The abstract of Patent Literature 1 is described as below. “The model selection unit selects the simulation model based on the selection condition set by the condition input unit so as to read the selected simulation model from the model database. Using the read simulation model, the simulation calculation unit executes the simulation calculation based on the initial state set by the condition input unit, and the simulation condition. Based on the model selection condition, the simulation calculation is executed by selecting the simulation model from those with different levels of details. For example, the highly accurate simulation will be executed for the important part using the model with high level of details, and the simulation will be executed for the other part with less importance in a short time using the model with low level of details.”
The background art of the relevant technical field is disclosed in Japanese Unexamined Patent Application Publication No. 2015-90639 (Patent Literature 2). The abstract of Patent Literature 2 is described as below. “The control unit equalizes the power consumption information every 30 minutes for each category so as to create each average power consumption pattern in daytime and nighttime of the day. The control unit then derives the correlation between the temperature and the power consumption for each category from the temperature information and the power consumption information of each day. Under the circumstance, in the case of the daytime, the correlation between the highest temperature and the total power consumption in the daytime is obtained. In the case of the nighttime, the correlation between the lowest temperature and the total power consumption in the nighttime is obtained. Then estimated temperatures (estimated highest temperature and estimated lowest temperature) of the day are acquired via the network. In accordance with the category of the day, the estimated power consumption in the daytime or the nighttime corresponding to the acquired estimated temperature is obtained.”
Patent Literature 1: Japanese Unexamined Patent Application Publication No. 2002-259888
Patent Literature 2: Japanese Unexamined Patent Application Publication No. 2015-90639
The generally employed analysis has been proposed to execute the analysis calculation for simulation with high accuracy of the important part using the model with high level of details, and for simulation with low accuracy of the other less important part using the model with low level of details by using the single unit of the analysis model, or selecting one of analysis models in accordance with the analysis condition. The analysis calculation is intended to estimate the stress generated in the mechanical structure to be analyzed, and the performance such as efficiency. In the above-described circumstances, calculation of the manufacturing cost of the mechanical structure may be largely influenced by the method or procedure of processing the mechanical structure. Therefore, it is difficult for the generally employed analysis method to calculate such manufacturing cost. Execution of the analysis individually while taking into account of the respective processing methods may cause the problem of prolonging the calculation time.
In the case of Patent Literature 1, the method of calculating the manufacturing cost is not sufficiently considered.
The calculation method for estimating the cost or the like based on the past data is implemented by generating the correlation (continuous) function while setting the cost required to be estimated as the dependent variable, and the temperature or the like as the independent variable. The calculation of the manufacturing cost of the mechanical structure may be largely influenced by the method or procedure of processing the mechanical structure. Individual consideration of the “process method 1”, “process method 2” and the like in the discrete manner may cause difficulties in applying the task to the continuous function. Even if the correlation function is made forcibly, the resultant estimation may exhibit low accuracy.
In the case of Patent Literature 2, estimation in the discontinuous situation is not sufficiently considered.
In order to solve the above-described problem of the related art, the present invention provides a total integration analysis model assistance device which simultaneously estimates a performance and a manufacturing cost of a mechanical structure to be analyzed. The device includes means for displaying an analysis process input screen on which an analysis process is displayed by selecting an analysis node having an analysis program, means for displaying an analysis condition input screen on which an input condition required for the analysis is displayed, means for creating an analysis model in accordance with the analysis process to analyze a performance estimation, and analyzing the manufacturing cost in accordance with a result of the performance estimation, means for acquiring past manufacturing cost information, means for displaying a data analysis condition input screen on which an analysis condition is displayed, means for grouping the past manufacturing cost information using a clustering method, means for visualizing a result derived from the unit for grouping as a graph, and means for acquiring and displaying an analysis result.
The present invention ensures to improve the analysis accuracy, and shorten the analysis time.
This example relates to a computer-aided analysis assistance device by integrally calculating both the efficiency performance and the manufacturing cost of the mechanical structure. The example will be described referring to the drawings.
The analysis process defining unit 101 displays an analysis process input screen through which an operator inputs an analysis process through drag and drop of an analysis model name and an analysis node containing the analysis program (synonymously, simply selecting the analysis node). The input analysis process information is displayed and further input to the database 109.
The analysis condition input/display unit 102 displays an analysis condition input screen through which the operator inputs the input condition required for analyzing the analysis model input by the analysis process defining unit 101. The input analysis condition information is displayed on the input screen, and further input to the database 109.
The analysis model creation/analysis control unit 103 acquires information data input through the analysis process defining unit 101, the analysis condition input/display unit 102, and the data analysis unit 106, and creates the analysis model in accordance with the analysis process so as to execute a performance estimation analysis. In accordance with the performance estimation result, the group corresponding to the group information generated by the data analysis unit is calculated, and the analysis is executed by obtaining the arithmetic mean value of the manufacturing cost data of the corresponding group as the manufacturing cost. Upon completion of the analysis, the analysis result is input to the database 109.
The data collection unit 104 acquires the past manufacturing cost information from the database 109.
The data analysis defining unit 105 displays a data analysis condition input screen through which the operator selects variables for the X-axis and the Y-axis, respectively for plotting on the graph. The variance for the group is input, and the input data analysis condition information is displayed on the input screen. The input information is then input to the database 109.
The data analysis unit 106 divides the past manufacturing cost information into groups that coincide with the input variance using the k-means clustering method in accordance with the condition input by the data analysis defining unit 105. The result is then input to the database 109.
The data analysis result display unit 107 displays the result of grouping calculated by the data analysis unit 106 for the operator in the form of the graph.
The analysis result display unit 108 acquires the result of analysis executed by the analysis model creation/analysis control unit 103 from the database 109 so as to be displayed for the operator.
The database 109 stores data derived from the analysis model input/display unit 101, the analysis condition input/display unit 102, the analysis model creation/analysis control unit 103, the data collection unit 104, the data analysis defining unit 105, the data analysis unit 106, the data analysis result display unit 107, and the analysis result display unit 108.
The above-configured process steps according to the embodiment will be described referring to
Taking a centrifugal compressor as an example of the mechanical structure, the means of the total integration analysis for estimating the performance and the manufacturing cost will be described referring first to Phase 1. The centrifugal compressor is the machine configured to rotate the impeller for suction of a gas, and to gradually reduce the gas flow rate in the centrifugal direction for compression. The centrifugal compressor includes a plurality of impellers for gas compression. Taking the compressor as the example, the means for integrally analyzing the performance and the manufacturing cost of the compressor will be described.
In S100 of Phase 1 as shown in
In S101, the analysis process defining unit 101 displays the analysis process input screen.
In S102, the analysis process information which has been input in S101 is acquired, that is, “condition acquirement”, “performance estimation”, “cost calculation”, and “result display”.
In S103, the information obtained in S102 is acquired, and input to the database 109.
In S200 as shown in
In S201, the information input by the analysis process defining unit 101 is acquired from the database 109.
In S202, the analysis condition input/display unit 102 displays the analysis condition input screen.
In S203, the analysis conditions input in S202 are acquired.
In S204, the information obtained in S203 is acquired, and input to the database 109.
An explanation will be made with respect to Phase 2. In S300 as shown in
In S301, the data collection unit 104 acquires the past manufacturing cost information from the database 109. In this case, the manufacturing cost information of the compressor is acquired. The database 109 stores the manufacturing cost of the compressor which has been designed in the past, and the performance information such as the length of the compressor, the external diameter of the impeller, the efficiency, and the head for acquisition.
In S302, the data analysis defining unit 105 displays the data analysis input screen.
In S303, the information of the data analysis condition input in S302 is acquired.
In S304, the data analysis unit 106 executes the data analysis. In this case, the analysis is executed using the k-means clustering method. The analysis step using the k-means method will be described as follows.
1. Data xi (i=1 . . . n) is randomly assigned to groups. The code n denotes the number of variables. As the values input corresponding to the X-axis and the Y-axis in S302 are variables, the code n is set to 2. The compressor length and the external diameter of the impeller are variables.
2. Based on the data assigned to the groups, the center Vj (j=1 . . . k) of each group is calculated. The code k denotes the number of groups. The arithmetic mean of the data corresponding to the assigned group is used for calculating the Vj.
3. Each distance between the Vj and the data xi is obtained so that the xi is re-assigned to the group with the nearest center.
4. If the calculation result shows that assignment of groups to all the data xi has not been changed, or the amount of the change is smaller than the preliminarily set threshold value, the process proceeds to the next step. In the case of the opposite result, the group is newly assigned, and recalculation is executed from the analysis step 2.
5. The variance of the distance between the xi and Vj is calculated. If the absolute value of the difference between the calculated variance and the one input in S302 is smaller than the preliminarily set threshold value, the process ends. If the calculated variance is larger than the input one, the number of the groups is increased. If the calculated variance is smaller than the input one, the number of the groups is decreased. The process then returns to the analysis step 1.
In S305, the data analysis result display unit 107 displays the analysis result.
In S306, referring to the result displayed in S305, if the operator requires increasing the number of groups, the variance is decreased. If the number of groups is required to be decreased, the variance is increased. Then the analysis button is pressed to return to S303. If the operator determines that the result is appropriate, Enter button is pressed.
In S307, the analysis result obtained in S304 is acquired, and input to the database 109.
Phase 3 will be described. In S400 as shown in
In S401, the analysis model creation/analysis control unit 103 acquires the analysis process input in S100, the analysis conditions input in S200, and the group information input in S300.
In S402, the analysis model is created and analyzed. In this case, in accordance with the analysis process input in S100, the analysis model is created and analyzed. Firstly, the calculation conditions are acquired. The conditions such as the suction pressure, the discharge pressure, and the suction temperature are acquired. Then the performance calculation is executed. Using the input conditions, the compressor length, the external diameter of the impeller, the efficiency, and the head are calculated. Then the cost calculation is executed. In this case, the manufacturing cost is calculated using the compressor length and the external diameter of the impeller in reference to the group information analyzed in S300. In other words, using the compressor length and the external diameter of the impeller, which have been calculated in S402, the corresponding group is obtained. The arithmetic mean of the manufacturing cost of the corresponding group is set as the manufacturing cost.
In S403, the analysis result is input to the database 109. In this case, the compressor length, the external diameter of the impeller, the efficiency, the head, and the manufacturing cost are input to the database.
In S501, the analysis result display unit 108 displays the result of the total integration analysis which has been executed by the analysis model creation/analysis control unit 103.
As described above, the information of the past manufacturing cost and the performance is collected for integrated calculation of the performance and the manufacturing cost of the mechanical structure. The collected data is then grouped through the clustering method. The results are stored in the database. This ensures to evaluate the performance of the mechanical structure, and simultaneously the creation cost in reference to the performance information derived from the group information to which the mechanical structure is corresponded. This makes it possible to shorten the estimation time.
The embodiment provides the total integration analysis device, and the method thereof. The device includes: means for displaying the analysis process input screen, inputs the analysis model name and the analysis process through the operator's operation for dragging and dropping the analysis node containing the analysis program, and displaying the input analysis process information; means for displaying the analysis condition input screen through which the operator inputs the input conditions required for analyzing the input analysis model, and displaying the input analysis condition information on the input screen; means for creating the analysis model in accordance with the analysis process for analyzing the performance estimation, calculating the group corresponding to the created group information in accordance with the performance estimation result, and executing the analysis by setting the arithmetic mean of the manufacturing cost data of the corresponding group as the manufacturing cost; means for acquiring the past manufacturing cost information; means for displaying the data analysis condition input screen through which the operator selects variables for the X-axis and the Y-axis to be plotted on the graph so as to be displayed, inputting the variance for grouping, and displaying the input data analysis condition information on the input screen; means for dividing the past manufacturing cost information into groups each corresponding to the input variance through the k-means clustering method; means for displaying the grouping result on the graph; and means for acquiring the analysis results and displaying the analysis results for the operator.
As described above, the past manufacturing cost data is grouped through the clustering method, and the grouping results are stored in the database. It is possible to evaluate the performance of the mechanical structure, and simultaneously, to estimate the manufacturing cost from the information of the group to which the mechanical structure is corresponded, thus shortening the time for estimation. The past manufacturing cost data is grouped through the clustering method, and the results are stored in the database for estimating the performance and calculating the manufacturing cost of the mechanical structure to be analyzed. The operator is allowed to analyze the performance of the mechanical structure for estimation, and simultaneously, to estimate the manufacturing cost, thus improving the analysis accuracy and shortening the time for analysis.
In this example, the grouping is executed based on the two-dimensional information including variables set for the X-axis and the Y-axis. However, the grouping process may be expanded to the use of the N-dimensional information.
The k-means clustering method is employed as described above. However, it is possible to employ any other clustering method such as the self-organizing mapping.
As described above, the analysis node constituting the analysis process is analyzed using the same computer. However, it is possible to use another computer in addition to the one as described above utilizing the network environment.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2016-155187 | Aug 2016 | JP | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/JP2017/024424 | 7/4/2017 | WO | 00 |
