Simulation adjustment method and apparatus

Abstract

To provide a simulation adjustment apparatus and a method that execute simulation adjustment considering a machine difference among a plurality of thermal treatment apparatuses represented by the same analysis model, and construct a simulation model of each thermal treatment apparatus in short time without stopping high-volume production.SOLUTION: A simulation adjustment method comprises the steps of: acquiring a thermal treatment apparatus rising temperature measurement value from a sensor unit 201 (S1); calculating a first statistical amount of each thermal treatment apparatus 10 by adding a new measured value to the rising temperature measurement value acquired in the past (S2); performing adjustment necessity determination on the basis of the first statistical amount of each thermal treatment apparatus (S3); calculating second statistical amounts of the rising temperature measurement values of total thermal treatment apparatuses 10 which is required to be adjusted (S4); performing adjustment of a simulation model based on data assimilation on the basis of the calculated second statistical amount and rising temperature measurement value (S6); and outputting the post adjustment simulation model (S7).SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present disclosure relates to a simulation tuning method and apparatus. [Background technology]

[0002] In order to understand thermal processes in manufacturing sites, efforts are being made to understand heat distribution and flow velocity distribution in advance using thermal fluid simulations. Thermal fluid simulations are simulations based on the Navier-Stokes equations, which represent the motion of fluids, or the heat advection equation, which represents the change in heat.

[0003] Thermal fluid simulation is an effective means for understanding phenomena in thermal processes where it is difficult to grasp the phenomena through direct measurement. By grasping the temperature distribution and flow velocity distribution of high-temperature heat treatment equipment as accurately as possible before the start of mass production, optimal production can be implemented, which contributes to improving product quality and energy saving in factories. It is also expected to reduce the number of actual machine tests, which will result in cost reduction. Patent Document 1 is a technology for accurately understanding phenomena such as those described above that are difficult to measure directly.

[0004] For example, Patent Document 1 uses a numerical simulation simulating an internal combustion engine to estimate the optimal ignition timing of the internal combustion engine, and is a typical example of using a numerical simulation to grasp a phenomenon that is difficult to measure directly. In an ECU of an electronically controlled internal combustion engine, it is necessary to assume in advance the ignition timing at which the torque is maximized for the operating conditions assumed in advance for ignition timing control and store it in memory. The stored ignition timing map is referred to to control the ignition of the air-fuel mixture. These ignition timing maps are created in advance based on a compatibility test using an actual machine during the development stage of the internal combustion engine. Such ignition timing maps are created by a numerical simulation device simulating an internal combustion engine. By using the numerical simulation device in this way, the compatibility process using the actual machine can be significantly reduced.

[0005] Furthermore, Patent Document 2 provides a technology for improving the uniformity of temperature distribution in a processing vessel during heat treatment using simulation. In Patent Document 2, a simulation model is modified to one that takes into account the machine differences of each heat treatment device using data acquired from a temperature sensor installed in the processing vessel, and processing settings are corrected using the simulation model. This makes it possible to maintain a uniform temperature distribution in the heat treatment device. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Patent No. 4376799 [Patent Document 2] JP 2022-168572 A Summary of the Invention [Problem to be solved by the invention]

[0007] By using a numerical simulation such as that described in Patent Document 1, the state inside the heat treatment device during the thermal process can be grasped, but it takes a lot of time and effort to adjust the simulation model. Therefore, when the simulation model needs to be adjusted many times, it is difficult to perform predictions using the simulation. Specifically, there are mass production sites where many heat treatment devices are used. Even if the internal structure of heat treatment devices is the same, there are differences in the mechanical elements such as the heater and the heat insulating material, and differences in the assembly accuracy, so that it is necessary to adjust the simulation model corresponding to each heat treatment device.

[0008] By using the configuration of Patent Document 2, it is possible to create a simulation model corresponding to each heat treatment device having machine differences. However, in the configuration of Patent Document 2, the temperature measurement value used in adjusting the simulation model is acquired from a sensor that is added later. Therefore, mass production needs to be temporarily stopped. In addition, the adjustment of the simulation model is performed based on human intuition and knack with reference to the temperature measurement value, and the timing of performing the adjustment is also based on human intuition and knack. In addition, when there are multiple heat treatment devices, it is necessary to acquire the temperature measurement value from each heat treatment device and adjust the simulation model to suit each heat treatment device, which takes a lot of time. For the reasons mentioned above, the simulation model adjustment method using the conventional configuration is not suitable for mass production sites.

[0009] The present disclosure provides a simulation adjustment method and device for performing simulation adjustment that takes into account machine differences between multiple heat treatment devices represented by the same analytical model based on start-up temperature measurement values ​​obtained from sensors installed in the heat treatment devices, and for constructing a simulation model for each heat treatment device in a short period of time without stopping mass production. [Means for solving the problem]

[0010] A simulation adjustment method according to one aspect of the present disclosure is a simulation adjustment method taking into account machine differences between a plurality of heat treatment devices represented by the same analytical model, comprising the steps of: Obtain a simulation model, In each of the plurality of heat treatment devices, a rise temperature measurement value from a temperature rise start of the heat treatment device until a steady state is obtained from each temperature sensor; calculating a first statistic of each of the heat treatment devices in a data processing unit based on a rise-up temperature measurement value of the heat treatment device and past data that is a rise-up temperature measurement value of the heat treatment device previously obtained; determining, by the data processing unit, whether or not adjustment of the simulation model is required based on the first statistical amount and a threshold value for determining whether or not adjustment of the simulation model is required; calculating a second statistic based on a rise-up temperature measurement value of at least one of the plurality of heat treatment devices by the data processing unit; A data assimilation unit adjusts the simulation model by data assimilation based on the second statistical amount and a start-up temperature measurement value of the heat treatment device that requires adjustment of the simulation model; The adjusted simulation model is output to a storage device.

[0011] A simulation adjustment device according to another aspect of the present disclosure is a simulation adjustment device that performs simulation adjustment taking into account machine differences between a plurality of heat treatment devices represented by the same analytical model, comprising: a data processing unit that acquires a simulation model used in a simulation for estimating a state inside the heat treatment device, calculates a first statistical amount for each of the heat treatment devices based on a rise-up temperature measurement value of the heat treatment device acquired from each temperature sensor of the heat treatment device from the start of temperature rise to a steady state and past data that is the rise-up temperature measurement value of the heat treatment device acquired in the past, determines whether or not adjustment of the simulation model is necessary based on the first statistical amount and a threshold value that determines whether or not adjustment of the simulation model is necessary, and calculates a second statistical amount based on the rise-up temperature measurement value of at least one of the heat treatment devices; a data assimilation unit that adjusts the simulation model by data assimilation based on the second statistics and a start-up temperature measurement value of the heat treatment device that requires adjustment of the simulation model; and an output unit that outputs the adjusted simulation model to a storage device. Effect of the Invention

[0012] According to the simulation adjustment method and device of the above aspect of the present disclosure, the measurement value of the rising temperature of each of the heat treatment devices, which is likely to have a large machine difference during temperature change, is focused on, and the first statistical amount based on the time-series rising temperature measurement value of each of the heat treatment devices from the past is used to determine whether or not the simulation model needs to be adjusted, and the second statistical amount based on the rising temperature measurement value of the plurality of heat treatment devices is used to adjust the simulation model by data assimilation. As a result, the simulation model that takes into account the machine difference between the plurality of heat treatment devices represented by the same analysis model can be adjusted in a short time without stopping mass production, based on the rising temperature measurement value acquired from each heat treatment device. [Brief description of the drawings]

[0013] [Figure 1] FIG. 1 is a block diagram showing a configuration example of a simulation adjustment device and a sensor unit according to an embodiment of the present disclosure. [Diagram 2] FIG. 2 is a schematic diagram showing a heat treatment apparatus to which a temperature sensor of a sensor unit is attached; [Diagram 3] 1 is a flowchart illustrating a procedure of a simulation adjustment method. [Figure 4] 1 is a schematic graph showing an example of a rise temperature measurement value acquired from a temperature sensor; [Diagram 5] FIG. 1 is a schematic diagram showing time series data of the start-up temperature measurement values ​​of each heat treatment device; [Figure 6] FIG. 13 is a diagram showing an example of a data set of time-series data used when calculating a first statistical amount for each heat treatment device. [Figure 7] 4 is a flowchart illustrating a process flow of the first statistical quantity calculation S3 of each heat treatment device shown in FIG. 3. [Figure 8] 4 is a flowchart illustrating a process flow of decision S4 shown in FIG. 3. [Figure 9] FIG. 13 is a diagram showing an example of a data set of time-series data used when calculating a second statistical quantity for all heat treatment devices. [Figure 10]A flowchart illustrating a process flow of calculating a second statistic S5 of the start-up temperature measurement values ​​of all the heat treatment devices shown in FIG. 3. [Figure 11] Diagram explaining data assimilation technology [Figure 12] Conceptual diagram of sequential data assimilation [Figure 13] A flowchart illustrating a process flow of the adjustment of the simulation model by sequential data assimilation S7 shown in FIG. [Figure 14] A diagram showing the adjustment process when the simulation model to be adjusted is set to temperature. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] Hereinafter, the present embodiment will be described with reference to the drawings.

[0015] <Configuration Example of Simulation Adjustment Device 100 and Sensor Unit 200> FIG. 1 is a block diagram showing an example configuration of a simulation adjustment device 100 and a sensor unit 200 capable of implementing a simulation adjustment method according to an embodiment of the present disclosure.

[0016] The simulation adjustment device 100 is equipped with a memory device 110, a control unit 120, and an input / output unit 130 as an example of an output unit, and implements a simulation adjustment method that takes into account machine differences between multiple heat treatment devices 10 represented by the same analytical model.

[0017] The sensor unit 200 includes a temperature sensor 201 .

[0018] The input / output unit 130 is connected to user interface devices such as a display, a keyboard, and a mouse, for example, and data can be exchanged via these.

[0019] The storage device 110 stores a program 111, simulation model data 112, model parameter data 113, time-series data 114, and statistical data 115. The storage device 110 is realized by, for example, a semiconductor storage device such as a flash memory, a solid-state drive (SSD), a magnetic storage device such as a hard disk drive (HDD), or other recording media, either alone or in combination. The storage device 110 may include a volatile memory such as an SRAM or a DRAM.

[0020] The control unit 120 includes a data processing unit 121, a simulation unit 122, and a data assimilation unit 123. The control unit 120 includes a CPU and performs information processing to realize a simulation adjustment function of the simulation adjustment device 100 described later. Such information processing is realized, for example, by the CPU operating in accordance with the instructions of the storage device 110 and the program 111 stored in the storage device 110. The CPU is an example of a processor in the present disclosure. The processor is not limited to a CPU as long as it includes an arithmetic circuit that performs calculations for information processing. For example, the processor may be configured with a circuit such as an MPU or an FPGA.

[0021] More specifically, the data processing unit 121 of the control unit 120 performs various calculation processes as described later, to obtain a simulation model, calculate a first statistical quantity, determine whether or not adjustments to the simulation model are necessary, and calculate a second statistical quantity.

[0022] The simulation unit 122 estimates the state inside the heat treatment device based on the simulation model.

[0023] The data assimilation unit 123 adjusts the simulation model by data assimilation based on the second statistics and the start-up temperature measurement value of the heat treatment device for which adjustment of the simulation model is required.

[0024] The temperature sensor 201 of the sensor unit 200 is installed in each heat treatment device 10 as described below, and is capable of measuring the temperature. The measurement results are input to the simulation adjustment device 100.

[0025] <Outline of Heat Treatment Apparatus 10> FIG. 2 is a schematic diagram showing each heat treatment device 10 in which a temperature sensor 201 of a sensor unit 200 is installed. A heater 11 is installed inside each heat treatment device 10, and heat treatment is performed by making the inside of each heat treatment device 10 a high-temperature environment using this heater 11. A temperature sensor 201 is installed in each heat treatment device 10. The temperature sensor 201 is, for example, a thermocouple, and acquires the temperature inside each heat treatment device 10. At least one temperature sensor 201 must be installed, and even if multiple temperature sensors are installed, the simulation can be adjusted in the same manner as in this embodiment. Note that in this specification, only the case in which one temperature sensor 201 is installed in each heat treatment device 10 will be described. Also, in this embodiment, it is assumed that there are multiple heat treatment devices 10, that is, N (N is an integer of 2 or more). In the multiple heat treatment devices 10, variations in internal state occur even if the processing settings are the same due to individual differences or deterioration of mechanism elements such as the heater 11 or heat insulating material, and differences in assembly accuracy. Hereinafter, this variation is defined as the machine difference between the plurality of heat treatment apparatuses 10. In this embodiment, taking the machine difference into consideration, a simulation adjustment corresponding to the state of each of the plurality of heat treatment apparatuses 10 is performed.

[0026] <Example of procedure for adjusting a simulation> 3 is a flow chart illustrating the procedure of a simulation adjustment method according to the present embodiment, which is a simulation adjustment method taking into account differences between a plurality of heat treatment apparatuses 10 represented by the same analytical model. S1 to S8 represent each step of the process performed by the simulation adjustment device 100. Also, the relationship between each step and the configuration of FIG. 1 is described here.

[0027] First, the data processing unit 121 of the simulation adjustment device 100 generates a simulation model (step S1). If a simulation model has already been generated, the generated simulation model is acquired by the data processing unit 121. The simulation model simulates all the heat treatment devices 10 and is necessary for performing a thermal fluid simulation of each of the heat treatment devices 10. The generated simulation model is stored in the simulation model data 112 of the storage device 110.

[0028] Following step S1, the temperature sensor 201 of the sensor unit 200 of each heat treatment device 10 acquires the start-up temperature measurement value of each heat treatment device 10 (step S2). The acquired start-up temperature measurement value of each heat treatment device 10 is stored in the time-series data 114. In this embodiment, it is assumed that all the heat treatment devices 10 operate in parallel. Therefore, data for N units is acquired in one heat treatment.

[0029] Following step S2, the data processing unit 121 calculates a first statistic of each heat treatment device 10, which will be described later, for the time series data of each heat treatment device 10 (step S3). More specifically, for each heat treatment device 10, the data processing unit 121 calculates the first statistic of the heat treatment device based on the rise temperature measurement value of the heat treatment device and past data, which are the rise temperature measurement values ​​of the heat treatment device previously obtained.

[0030] Following step S3, the data processing unit 121 determines whether or not to adjust the simulation model for each heat processing device 10 (step S4). Based on the determination process performed in step S4, the data processing unit 121 determines which of the multiple heat processing devices 10 should have their simulation models adjusted. More specifically, the data processing unit 121 determines whether or not to adjust the simulation model based on the first statistical amount and a threshold value that determines whether or not to adjust the simulation model, which will be described later. The heat processing device 10 for which it is determined that the simulation model should not be adjusted does not execute processing in the steps described later. The heat processing device 10 for which it is determined that the simulation model should be adjusted executes processing in the steps described later.

[0031] Following step S4, for the heat treatment apparatuses 10 determined to require adjustment of the simulation model, the rising temperature measurement values ​​of all the heat treatment apparatuses 10 acquired in step S2 are acquired from the time series data 114, and a second statistical amount, which will be described later, is calculated by the data processing unit 121 (step S5). The calculation of the second statistical amount by the data processing unit 121 is not limited to the calculation by acquiring the rising temperature measurement values ​​of all the heat treatment apparatuses 10 acquired in step S2 from the time series data 114, but may be performed by acquiring the rising temperature measurement value of at least one of all the heat treatment apparatuses 10 acquired in step S2 from the time series data 114.

[0032] Following step S5, the data processing unit 121 calculates an initial ensemble (step S6). The initial ensemble is an initial setting required when performing a simulation using sequential data assimilation in step S7 described later. In this embodiment, the calculation of the initial ensemble is performed by the data processing unit 121 based on the variance of the rise temperature measurement values ​​of all heat treatment devices 10 of the second statistics calculated in step S5. Specifically, the initial ensemble is calculated by the data processing unit 121 as random numbers based on a normal distribution with the room temperature as the average and the value calculated in step S5 as the variance. In this embodiment, the Box-Muller method is used as a method for generating random numbers based on a normal distribution, but the initial ensemble may be calculated using other random number generation methods. In addition, the average is set to the room temperature, but the temperature before the rise of the time series data of each heat treatment device 10 stored in the time series data 114 may be used as the room temperature. In this embodiment, the number of initial ensembles to be generated is set to 10, but may be increased or decreased depending on the calculation resources. The calculated initial ensemble is stored in the model parameter data 113 .

[0033] Following step S6, the data processor 121 uses the initial ensemble based on the second statistics to adjust the simulation model by sequential data assimilation based on the start-up temperature measurement value of the heat treatment device 10 for which the simulation model needs to be adjusted (step S7). In this embodiment, an ensemble Kalman filter, which is sequential data assimilation, is used, but other sequential data assimilation methods may be used.

[0034] Following step S7, the simulation model after adjustment is output to the storage device 110 (step S8). The adjusted simulation model is adjusted based on the start-up temperature measurement values ​​acquired from each heat treatment device 10, and is therefore a simulation model that takes into account differences between the heat treatment devices.

[0035] As described above, by carrying out each step of the flowchart in FIG. 3, it is possible to adjust the simulation model taking into consideration the differences between the multiple heat treatment apparatuses 10.

[0036] <Example of temperature measurement value at start-up in step S2> 4 is a graph showing an example of the rise-up temperature measurement value of the heat treatment apparatus 10 acquired in step S2. After the start of heat treatment, the temperature inside the heat treatment apparatus 10 starts to rise and then reaches a steady state. The rise-up temperature measurement value is a measurement value for the period from the point at which the temperature starts to rise (i.e., the temperature of the heat treatment apparatus starts to rise) to the point at which the steady state is reached. The acquired rise-up temperature measurement value is stored in the time series data 114 of the storage device 110.

[0037] <An example of the rising temperature measurement value of the time series data 114> 5 is an example of the rising temperature measurement values ​​stored in the time series data 114 of the storage device 110. In the time series data 114, the rising temperature measurement values ​​acquired by the temperature sensors 201 provided in the multiple heat treatment devices 10 are stored as time series data from the past to the present.

[0038] <Example of Data Set Used When Calculating the First Statistical Quantity by the Data Processing Unit 121 in Step S3> 6 shows an example of a data set used for calculating the first statistical quantity of each heat processing device 10 for the time series data of the heat processing device 10 in step S3 in which the data processing unit 121 calculates the first statistical quantity of each heat processing device for the time series data of the heat processing device 10. The data set surrounded by a dashed line is the time series data of the start-up temperature measurement value of the first unit. In step S3, the data processing unit 121 calculates the first statistical quantity for the data set shown in this example for the number of heat processing devices 10.

[0039] <Example of processing flow of step S3> FIG. 7 is a flowchart illustrating the process flow of step S3.

[0040] In step S31, the start-up temperature measurement values ​​of each heat treatment device 10 stored in the time-series data 114 are acquired.

[0041] Next, in step S32, the data processing unit 121 calculates an average waveform of the rising temperature measurement values ​​of each heat treatment device 10. The average waveform of the rising temperature measurement values ​​of each heat treatment device 10 is a waveform that represents the average rise of the temperature of each heat treatment device 10, and is calculated by the data processing unit 121 from the time series data stored in the time series data 114. In the case of the first heat treatment device 10 in FIG. 5 as an example, there are K pieces of time series data up to the Kth time for the rising temperature measurement value of the first heat treatment device 10, so that in order to calculate the average waveform of the temperature measurement value in the data processing unit 121, the data processing unit 121 adds up the temperature values ​​of the K pieces of time series data, and divides the added temperature value by the number of data K, so that the data processing unit 121 can calculate the average waveform of the temperature measurement value (i.e., the first average waveform) of the time series data of the first heat treatment device 10. Here, K is an integer of 2 or more.

[0042] Next, in step S33, the Euclidean distance between the average waveform calculated in step S32 and each piece of time series data is calculated by the data processing unit 121. In addition, the Euclidean distance between the average waveform and each piece of time series data is all added by the data processing unit 121, and the value obtained by dividing by the number K of pieces of time series data is calculated by the data processing unit 121 as the Euclidean distance average.

[0043] Next, in step S34, the first statistics calculated in steps S32 and S33 are output from the data processing unit 121 and stored in the statistics data 115. Here, the first statistics are the average waveform, Euclidean distance, and Euclidean distance average of the rising temperature measurement values ​​of each heat treatment device 10.

[0044] Next, in step S35, the data processing unit 121 determines whether the calculation of the first statistics for each heat processing device 10 has been completed. The data processing unit 121 determines whether the calculation of the first statistics for each heat processing device 10 has been completed based on the number N of heat processing devices 10. If the number of executions is equal to or less than N, the process returns to step S32 again, and the data processing unit 121 calculates the first statistics for the time-series data of the next heat processing device 10.

[0045] <Example of decision flow in step S4> FIG. 8 is a flowchart illustrating a determination flow in data processing unit 121 in step S4.

[0046] In step S41, the first statistical quantity of each heat treatment device 10 calculated by the data processing unit 121 in step S3 is acquired, specifically, the average waveform of the rising temperature measurement value of each heat treatment device 10 (i.e., the first average waveform), the Euclidean distance between the average waveform of the temperature measurement value and each time series data, and the Euclidean distance average are acquired. More specifically, the first statistical quantity is the first average waveform calculated by the data processing unit 121 by adding the rising temperature measurement value and past data of each of the multiple heat treatment devices 10 in the data processing unit 121 and dividing by the number of data in the data processing unit 121, the Euclidean distance between the rising temperature measurement value and past data of each of the multiple heat treatment devices 10 and the average waveform, and the Euclidean distance average calculated by the data processing unit 121 by adding all the Euclidean distances of each of the multiple heat treatment devices 10 in the data processing unit 121 and dividing by the number of data in the data processing unit 121.

[0047] Next, in step S42, the data processing unit 121 performs a judgment using the acquired first statistics of each heat treatment device 10. In this embodiment, a threshold is set, and the data processing unit 121 performs a judgment by comparing the Euclidean distance between the rising temperature measurement value of the Kth time series data and the average waveform with the threshold. The average value of the Euclidean distance between the average waveform and each time series data calculated in step S34 is used as the threshold. By using the average value of the Euclidean distance between the average waveform and each time series data as the threshold, the data processing unit 121 judges how large the error of the rising temperature measurement value of the Kth time is from the average in the time series data, and adjusts the simulation model in the subsequent steps only for the heat treatment device 10 with a large error.

[0048] Next, in step S43, the data processing unit 121 determines whether or not to end step S4. The data processing unit 121 determines whether or not to end step S4 based on the number N of heat treatment devices 10. If the number of executions is equal to or less than N, the process returns to step S41 again, and the data processing unit 121 determines whether or not to subsequently adjust the next heat treatment device 10.

[0049] As described above, by performing the judgment in step S4 in the data processing unit 121, the adjustment of the simulation model of the heat treatment device 10 that does not require the simulation adjustment is not performed in the subsequent steps, and the time required for the simulation adjustment can be reduced. In addition, since the judgment criterion for performing the simulation adjustment is determined by the threshold value calculated by the data processing unit 121 from the first statistical amount of the time-series data of each heat treatment device 10, it can be set simply by acquiring the measured rise temperature value.

[0050] <Example of Data Set Used When Calculating the Second Statistic in the Data Processing Unit 121 in Step S5> FIG. 9 shows an example of a data set used in step S5 when the data processing unit 121 calculates the second statistical quantity of the rise temperature measurement value of the heat treatment device 10 acquired in step S2. The calculation is performed by the data processing unit 121 using the Kth rise temperature measurement values ​​of the first to Nth machines stored in the time series data 114.

[0051] <Example of process flow of step S5> FIG. 10 is a flowchart illustrating the process flow of step S5 in which the data processing unit 121 obtains the rise-up temperature measurement value of the heat treatment device 10 obtained in step S2 from the time-series data 114 and calculates the second statistical amount thereof.

[0052] In step S51, the data processing unit 121 calculates an average waveform (i.e., a second average waveform) of the rising temperature measurement values ​​of all the heat processing devices 10 as a second statistic. The average waveform of the rising temperature measurement values ​​of all the heat processing devices 10 is a waveform representing the average rise of the temperature of all the heat processing devices 10, and is calculated by the data processing unit 121 from the time series data stored in the time series data 114. In this embodiment, the average waveform of the rising temperature measurement values ​​of all the heat processing devices 10 is that of the Kth heat processing. In the case of the Kth heat processing, since there are N rising temperature measurement values ​​from the 1st unit to the Nth unit, the data processing unit 121 adds up the N rising temperature values ​​and divides the temperature value added by the data processing unit 121 by N, so that the data processing unit 121 can calculate the average waveform of the rising temperature measurement values ​​of all the heat processing devices 10.

[0053] Subsequently, in step S52, the data processing unit 121 calculates the variance of the rising temperature measurement values ​​of all the heat processing devices 10. Specifically, the data processing unit 121 calculates the Euclidean distance between the average waveform of the rising temperature measurement values ​​of all the heat processing devices 10 (i.e., the second average waveform) calculated in step S51 and the rising temperature measurement value of each heat processing device 10, obtains the sum of squares of the calculated value, and divides the sum by the number N of the heat processing devices 10 to calculate the variance in the data processing unit 121. The calculated variance is stored in the statistics data 115 as the second statistic.

[0054] <Data Assimilation Concept> FIG. 11 is a conceptual diagram of data assimilation. Data assimilation is a method for predicting a phenomenon using measured values ​​obtained from a sensor and predicted values ​​obtained from a numerical simulation. In data assimilation, it is assumed that the measured values ​​and predicted values ​​each contain an error, and the phenomenon is predicted statistically. In this embodiment, the data assimilation unit 123 uses sequential data assimilation to adjust the simulation model.

[0055] <Concept of sequential data assimilation> FIG. 12 is a conceptual diagram of sequential data assimilation. In sequential data assimilation, a simulation with a plurality of different initial conditions is developed over time, and at the time when a measurement value is obtained, each simulation is appropriately adjusted by the data assimilation unit 123, while the average value of the results obtained from the plurality of simulations is treated as a predicted value. In sequential data assimilation by the data assimilation unit 123, a plurality of simulations with different conditions are performed in parallel, which are called an ensemble. In addition, a plurality of different simulation initial conditions must be determined, which is called an initial ensemble. In this embodiment, by determining an initial ensemble from the start-up temperature measurement values ​​of a plurality of heat treatment devices 10 by the data assimilation unit 123, the data assimilation unit 123 can adjust the simulation by sequential data assimilation simply by acquiring the start-up temperature measurement values ​​from the sensor unit 200.

[0056] <Example of processing flow for step 7> FIG. 13 is a flowchart illustrating a process flow of step 7 in which the data assimilation unit 123 adjusts the simulation model by sequential data assimilation.

[0057] In step S71, the data assimilation unit 123 acquires the initial ensemble calculated in step S6.

[0058] Next, in step S72, the data assimilation unit 123 acquires the Kth rise-up temperature measurement value of the heat treatment device 10 to be adjusted.

[0059] In step S73, the data assimilation unit 123 adjusts the simulation model using an ensemble Kalman filter.

[0060] Next, in step S74, the data assimilation unit 123 determines whether adjustments have been completed for all heat treatment devices 10 for which it was determined in step S4 that simulation model adjustments were required. If there are still simulation models of heat treatment devices 10 that require adjustments, the process returns to step S72, and the data assimilation unit 123 performs adjustments again.

[0061] <Adjustment of simulation model> FIG. 14 is a diagram showing the state of adjustment of a simulation model when the temperature of the heater 11 is used as the setting of the simulation model to be adjusted by sequential data assimilation in the data assimilation unit 123. The solid line in the graph shows the ensemble average value in the graph. Using the initial ensemble calculated in step S6 as the initial setting, the simulation unit 122 performs multiple thermal fluid simulations until the time when the rising temperature measurement value is acquired. At the timing when the rising temperature measurement value is acquired, the data assimilation unit 123 performs data assimilation and corrects each ensemble. Using the corrected ensemble, multiple numerical simulations are time-evolved in the simulation unit 122 until the time when the temperature measurement value is obtained again, and the data assimilation unit 123 performs correction of the ensemble by data assimilation. This is repeated until the last time when the rising temperature measurement value is obtained, and the data assimilation unit 123 acquires the final average value of the temperature of each ensemble as the setting value of the adjusted simulation model.

[0062] <Effects of the embodiment> According to the simulation adjustment method and device of the present embodiment, focusing on the rise temperature measurement value of each heat treatment device 10, which is likely to have a large machine difference among the plurality of heat treatment devices 10 during temperature change, a first statistic based on the time-series rise temperature measurement value from the past of each heat treatment device 10 is used to determine whether or not adjustment of the simulation model is necessary, and a second statistic based on the rise temperature measurement values ​​of the plurality of heat treatment devices 10 is used to adjust the simulation model by data assimilation. As a result, it is possible to build an adjustment of the simulation model that takes into account the machine difference among the plurality of heat treatment devices 10 represented by the same analytical model, based on the rise temperature measurement value acquired from each heat treatment device 10, in a short time without stopping mass production.

[0063] In this embodiment, the temperature is used as the ensemble, but there is no problem if another physical quantity in the simulation model setting, such as wind speed, is added to the ensemble. In this case, the simulation model can be adjusted using the temperature measurement value in the same way.

[0064] Furthermore, in the simulation adjustment device 100 of this embodiment, the data assimilation unit 123 adjusts the simulation model using the rise temperature measurement value acquired by the temperature sensor 201 installed in the heat treatment device 10, so there is no need to conduct a separate experiment for adjusting the simulation. Therefore, since the data assimilation unit 123 can adjust the simulation without stopping mass production, a simulation environment can be constructed relatively easily at the mass production site.

[0065] Furthermore, by storing the rise temperature measurement values ​​obtained by the temperature sensor 201 as time-series data, the first and second statistics are calculated by the data processing unit 121, and the data processing unit 121 uses the calculated statistics to make a judgment, thereby automatically determining the timing of simulation adjustment in the data assimilation unit 123. As a result, adjustment is not performed on simulation models that do not require adjustment, and the time required for simulation adjustment can be shortened.

[0066] In addition, by appropriately combining any of the various embodiments or modifications described above, it is possible to achieve the effects of each of them. In addition, combinations of embodiments, combinations of examples, or combinations of embodiments and examples are possible, and combinations of features of different embodiments or examples are also possible. [Industrial Applicability]

[0067] The simulation adjustment method and apparatus according to the above aspects of the present disclosure can construct a simulation model that takes into account machine differences between multiple heat treatment devices represented by the same analytical model in a short period of time without stopping mass production, and is useful for understanding processes, for example, in thermal processes at manufacturing sites. [Explanation of symbols]

[0068] 10 Heat treatment device 11 Heater 100 Simulation Adjustment Device 110 Storage device 111 Program 112 Simulation model data 113 Model Parameter Data 114 Time Series Data 115 Statistical Data 120 Control section 121 Data Processing Unit 122 Simulation Department 123 Data Assimilation Department 130 Input / output section 200 Sensor section 201 Temperature Sensor

Claims

1. A simulation adjustment method taking into account machine differences between a plurality of heat treatment devices represented by the same analytical model, comprising: Obtain a simulation model, In each of the plurality of heat treatment devices, a rise temperature measurement value from a temperature rise start of the heat treatment device until a steady state is obtained from each temperature sensor; calculating a first statistic of each of the heat treatment devices in a data processing unit based on a rise-up temperature measurement value of the heat treatment device and past data that is a rise-up temperature measurement value of the heat treatment device previously obtained; determining, by the data processing unit, whether or not adjustment of the simulation model is required based on the first statistical amount and a threshold value for determining whether or not adjustment of the simulation model is required; calculating a second statistic based on a rise-up temperature measurement value of at least one of the plurality of heat treatment devices by the data processing unit; A data assimilation unit adjusts the simulation model by data assimilation based on the second statistical amount and a start-up temperature measurement value of the heat treatment device that requires adjustment of the simulation model; The simulation model after adjustment is output to a storage device.

2. The first statistic is a first average waveform calculated by the data processing unit by adding the rising temperature measurement value and the past data of each of the plurality of heat treatment devices in the data processing unit and dividing the result by the number of data in the data processing unit; the rising temperature measurement value of each of the plurality of heat treatment devices, and a Euclidean distance between the past data and the average waveform; The Euclidean distances of the respective heat treatment devices are all added up in the data processing unit, and the sum is divided by the number of data to obtain a Euclidean distance average calculated in the data processing unit. The simulation adjustment method according to claim 1 , wherein

3. 3. The simulation adjustment method according to claim 2, wherein a value used as the threshold value used for determining whether or not the adjustment of the simulation model should be performed is a Euclidean distance average calculated using the first statistic.

4. The second statistic is a second average waveform calculated by the data processing unit by adding up the rising temperature measurement values ​​acquired from the plurality of heat treatment devices in the data processing unit and dividing the sum by the number of the heat treatment devices in the data processing unit; The data processing unit calculates Euclidean distances between the rising temperature measurement values ​​acquired from the plurality of heat treatment devices and the second average waveform, and the data processing unit divides the sum of squares of the calculated values ​​by the number of the heat treatment devices, thereby calculating a variance of the rising temperature measurement values. The simulation adjustment method according to claim 1 , wherein

5. The simulation adjustment method according to claim 4, wherein when adjusting the simulation model by the data assimilation, an initial ensemble at the start of data assimilation is calculated in the data processing unit using random numbers generated according to a normal distribution determined by the variance of the rise temperature measurement value.

6. 6. The simulation adjustment method according to claim 1, wherein in adjusting the simulation model, the simulation model is adjusted by sequential data assimilation using the rise-up temperature measurement values ​​acquired from the plurality of heat treatment devices.

7. A simulation adjustment device that performs simulation adjustment taking into account machine differences between a plurality of heat treatment devices represented by the same analytical model, a data processing unit that acquires a simulation model used in a simulation for estimating a state inside the heat treatment device, calculates a first statistical quantity for each of the heat treatment devices based on a rise-up temperature measurement value of the heat treatment device acquired from each temperature sensor of the heat treatment device from the start of temperature rise to a steady state and past data that is the rise-up temperature measurement value of the heat treatment device acquired in the past, determines whether or not adjustment of the simulation model is necessary based on the first statistical quantity and a threshold value that determines whether or not adjustment of the simulation model is necessary, and calculates a second statistical quantity based on the rise-up temperature measurement value of at least one of the heat treatment devices; a data assimilation unit that adjusts the simulation model by data assimilation based on the second statistics and a start-up temperature measurement value of the heat treatment device that requires adjustment of the simulation model; and an output unit that outputs the adjusted simulation model to a storage device.

Images