Processing control device and control method for processing device

Abstract

According to the present invention, a processing control device that controls a processing device that performs a plurality of processes on a processed product comprises: a target physical amount setting unit that sets a target physical amount in each process of the plurality of processes; a physical amount acquisition unit that acquires a physical amount of the processed product after the processing of each process; a processing condition acquisition unit that acquires a processing condition of each process; and a storage unit that stores a prediction model associating, among any two consecutive processes included in the plurality of processes, the physical amount after the processing of the process in a relatively preceding step, the target physical amount of the process in a relatively subsequent step, and the processing condition of the process in the subsequent step. The processing condition setting unit uses the prediction model to acquire the processing condition of each of two different processes included in the plurality of processes from the target physical amounts of the respective processes and the physical amounts after the processing of the process in the preceding step of the respective processes.

Description

Processing control apparatus and control method for processing apparatus

[0001] The present invention relates to a processing control apparatus and a control method for a processing apparatus.

[0002] There has been known a molding control apparatus that calculates processing conditions using a prediction model in which processing conditions and predicted values of physical quantities after processing are associated with each other in advance (see, for example, Patent Document 1).

[0003] Japanese Unexamined Patent Application Publication No. 2024-021184

[0004] Kanta Suzuki, Kazuya Yamauchi, Kosuke Kojima, Ichiro Kiriyama, Kotaro Ito, Yuta Yokoyama, Hirofumi Sugiyama, Shigenobu Okazawa, "Prediction of Stress Distribution During Plastic Deformation of Bent Plate Members Using Machine Learning", JSAE Transactions, Vol. 53, No. 5, pp. 892-897 (2022)

[0005] However, there is room for improvement in the technique of calculating processing conditions using the prediction model described in Patent Document 1.

[0006] The present invention is intended to solve at least part of the above-described problems, and an object of the present invention is to efficiently calculate processing conditions using a prediction model in a processing control apparatus.

[0007] The present invention is intended to solve at least part of the above-described problems, and can be implemented as the following aspects.

[0008] (1) According to one embodiment of the present invention, a processing control device is provided. This processing control device controls a processing device that performs a plurality of processing operations on a workpiece, and comprises: a target physical quantity setting unit that sets a target physical quantity for each of the plurality of processing operations; a physical quantity acquisition unit that acquires a physical quantity of the workpiece after processing for each of the processing operations; a processing condition acquisition unit that acquires processing conditions for each of the processing operations; and a storage unit that stores a prediction model relating the physical quantity after processing of a processing operation that is relatively the preceding process among any two consecutive processing operations included in the plurality of processing operations, the target physical quantity of a processing operation that is relatively the following process, and the processing conditions of the following process, wherein the processing condition setting unit uses the prediction model to acquire the processing conditions for each of two different processing operations included in the plurality of processing operations from the target physical quantity of each processing operation and the physical quantity after processing of the processing operation that is the preceding process for each processing operation.

[0009] With this configuration, the machining conditions for two different machining processes included in the aforementioned multiple machining processes can be obtained using a predictive model from the target physical quantity of each machining process and the post-machining physical quantity of the machining process preceding each machining process. Therefore, with this configuration, machining conditions can be calculated efficiently.

[0010] (2) In the processing control device according to the above embodiment, the processing condition acquisition unit may include a correction unit that corrects the prediction model based on the target physical quantity of the Nth (where N is a natural number) processing among the plurality of processing and the physical quantity after processing of the Nth processing, and the processing conditions for the (N+1)th processing may be acquired from the physical quantity after processing of the Nth processing and the target physical quantity of the (N+1)th processing using the prediction model corrected by the correction unit.

[0011] In this configuration, the prediction model is corrected based on the target physical quantity in the Nth machining process and the physical quantity after machining in the Nth machining process. Using the corrected prediction model, the machining conditions for the (N+1)th machining process are calculated from the physical quantity after machining in the Nth machining process and the target physical quantity for the (N+1)th machining process. Therefore, this configuration improves the accuracy of the calculated machining conditions and allows for efficient calculation of machining conditions.

[0012] (3) In the machining control device according to the above embodiment, the correction unit may further correct the prediction model based on the target physical quantity in the N+1th machining process among the plurality of machining processes and the actual physical quantity after machining in the N+1th machining process.

[0013] With this configuration, the prediction model is corrected based on the target physical quantity in the (N+1)th machining process and the physical quantity after the (N+1)th machining process. Using the corrected prediction model, it becomes possible to calculate the machining conditions for the (N+1)th process. Therefore, with this configuration, the machining conditions for each of the three or more machining processes can be calculated efficiently.

[0014] (4) In the processing control device according to the above embodiment, the processing device may be a feed bending device, the processed product may be a long member made from a metal material, and the physical quantity may be a quantity relating to the bending of the processed product.

[0015] With this configuration, in multiple bending processes of long members made from metal materials, a predictive model can be used to obtain the processing conditions for two different bending processes included in the multiple bending processes, based on the target bending angle of each bending process and the bending angle after processing of the preceding bending process. Therefore, with this configuration, the processing conditions for bending processes can be calculated efficiently.

[0016] Furthermore, the present invention can be realized in various forms, for example, in the form of a processing apparatus including a processing control device, a system equipped with a processing control device, a processing method, a method for controlling a system including a processing apparatus, a computer program for controlling a system including a processing apparatus, a server device for distributing a computer program for controlling a system including a processing apparatus, a non-temporary storage medium storing a computer program for controlling a system including a processing apparatus, and so on.

[0017] This is a schematic block diagram of a processing control system as one embodiment of the present invention. This is a diagram illustrating bending by a processing device. This is an explanatory diagram illustrating a prediction model. This is a flowchart of the processing control method of the processing control device of this embodiment. This is a diagram illustrating the parameters related to bending by the processing device. This is a diagram illustrating the operation flow of the processing control device and the processing device. This is an example of a processed product processed by the processing device. This is a diagram illustrating the operation flow of the processing control device and the processing device of a second embodiment.

[0018] <First Embodiment> Figure 1 is a schematic block diagram of a machining control system 500 as one embodiment of the present invention. The machining control system 500 shown in Figure 1 comprises a machining control device 100, a machining device 200, and a measurement unit 300.

[0019] The processing device 200 receives control from the processing control device 100 and performs processing on the workpiece. Various processing equipment can be applied to the processing device 200. In this embodiment, the processing device 200 is equipped with three rollers and performs bending of a long member. Figure 2 is a diagram illustrating the bending process performed by the processing device 200. The processing device 200 bends the long member, which is sandwiched between roller RL1 and rollers RL2 and RL3, by pushing roller RL1. Subsequently, the processing device 200 rotates rollers RL1, RL2 and RL3 to move the long member and advance the bending process by moving the position where the long member and the pressing roller RL1 are in contact.

[0020] Returning to Figure 1, the measurement unit 300 measures the physical quantities of the processed product. The physical quantities of the processed product measured by the measurement unit 300 are acquired by the physical quantity acquisition unit 115. Various devices capable of measuring physical quantities corresponding to the processing being performed can be used in the measurement unit 300. In this embodiment, since the processing apparatus 200 performs bending of a long member LM, a device for measuring the bending angle is applied.

[0021] The machining control device 100 controls the machining apparatus 200 to perform machining based on machining conditions. The machining control device 100 includes a CPU (Central Processing Unit) 110 and a storage unit 150, and is connected to an input unit 130 and a display unit 140. The machining control device 100 is composed of, for example, a personal computer.

[0022] The storage unit 150 consists of a hard disk drive (HDD) and the like. The storage unit 150 includes a prediction model storage area 151 that stores the prediction model PM. Details of the prediction model PM will be described later.

[0023] The CPU 110 controls various parts of the machining control device 100 and the machining apparatus 200 by loading a computer program stored in a ROM (Read Only Memory) (not shown) into the RAM (Random Access Memory) and executing it. In addition, it functions as a construction unit 111, a target physical quantity setting unit 112, a machining condition acquisition unit 113, and a physical quantity acquisition unit 115.

[0024] The input unit 130 consists of a keyboard, mouse, or the like, and accepts input for initial settings to control the machining process. These initial settings include, for example, the number of machining steps and the target physical quantity of the final machined product. The input unit 130 may also be fitted with various devices that allow the operator to input information, such as a touch panel.

[0025] The display unit 140 consists of an LCD monitor or the like, and displays initial setting information and the like that input by the input unit 130.

[0026] The construction unit 111 constructs a prediction model PM and stores the constructed prediction model PM in the prediction model storage area 151 of the storage unit 150. The construction unit 111 constructs the prediction model PM based on accumulated past processing data (physical quantities of the processed product before processing, physical quantities of the processing target, processing conditions, and physical quantities of the processed product after processing).

[0027] The target physical quantity setting unit 112 acquires the target physical quantity for machining. The target physical quantity is the desired physical quantity when performing machining, and is used to calculate the machining conditions.

[0028] The processing condition acquisition unit 113 calculates and acquires the processing conditions for the current processing using the prediction model PM, the physical quantities after processing from the previous processing, and the target physical quantities for the current processing. The target physical quantities for the current processing refer to the target physical quantities for each processing acquired by the target physical quantity setting unit 112.

[0029] Furthermore, the processing condition acquisition unit 113 includes a correction unit 114. The correction unit 114 corrects the prediction model based on the actual physical quantity after processing in the previous process, the target physical quantity in the previous process, and (the error between them). The actual physical quantity after processing in the previous process is a value acquired by the physical quantity acquisition unit 115, which will be described later. Therefore, the processing condition acquisition unit 113 uses the prediction model corrected by the correction unit 114 based on the actual physical quantity of the processed product after processing in the previous process, the target physical quantity in the previous process, and (the error between them) to acquire the processing conditions for the subsequent process, which are calculated based on the target physical quantity in the subsequent process. Note that the correction unit 114 does not need to correct the prediction model if the error between the actual physical quantity of the processed product after processing in the previous process and the target physical quantity in the previous process is smaller than a predetermined tolerance range.

[0030] The physical quantity acquisition unit 115 acquires the physical quantities of the processed product after processing. The physical quantities of the processed product after processing are measured by the measurement unit 300. As a method for acquiring the physical quantities of the processed product after each processing step, information may be transmitted using a conversion adapter or cable, or the measurement unit 300 may read the measured values ​​and input them to the input unit 130, which will be described later.

[0031] Figure 3 is an explanatory diagram illustrating a predictive model PM. The predictive model PM associates the processing conditions MC for the current process, the target physical quantity QN for the current process, the actual physical quantity QL after processing in the previous process, and the target physical quantity in the previous process. In other words, the predictive model PM associates the physical quantity after processing in the process that is relatively preceding, the target physical quantity for the process that is relatively following, and the processing conditions for the process that is following. The predictive model PM can associate as many parameters as necessary, depending on the number of processing types. The processing conditions MC are represented by the intersection of the plane QL, which represents the actual physical quantity after processing in the process that is preceding, shown as a dashed line, and the predictive model PM, and the intersection of the plane QN, which represents the target physical quantity of the processed product in the process that is following, shown as a thick solid line, and the predictive model PM.

[0032] Figure 4 is a flowchart showing the machining control method of the machining control device of this embodiment.

[0033] First, the processing control device 100 initially acquires the number of processing steps and the target physical quantity of the final processed product from the input unit 130 as initial settings (step S1). Here, the input unit 130 inputs 3 as the number of processing steps and 90° as the final bending angle target for the long member LM made of metal material. Since 3 processing steps and 90° as the final bending angle target have been input, the processing of bending the long member LM to 90° by repeating the bending process shown in Figures 2(A) to (E) three times will be described. Therefore, here the number of processing steps is 3 and the final target bending angle is 90°, the first bending process is referred to as the "first processing," the second bending process as the "second processing," and the third bending process as the "third processing." Furthermore, the allowable error range for the actual bending angle Ameasure_n (n=1,2,3) after each processing step is defined as an error of 3° or less compared to the target bending angle Atage_n in the processing step.

[0034] The bending process performed by the processing device 200 is as shown in Figures 2(A) to (E). First, as shown in Figure 2(A), the long member LM is placed between roller RL1 and rollers RL2 and RL3 of the processing device 200.

[0035] Next, as shown in Figure 2(B), the processing device 200 moves roller RL1 towards the long member LM, and bends the long member LM by sandwiching it between roller RL1, rollers RL2 and RL3. The point where the long member LM and roller RL1 come into contact at this time (the position where the long member LM is bent as roller RL1 is pressed in) will be called the workpiece position of the long member LM.

[0036] Next, as shown in Figure 2(C), the processing device 200 rotates rollers RL1, RL2, and RL3 respectively while maintaining the pressure of roller LR1, thereby moving the long member LM. This movement of the long member LM moves the workpiece position of the long member LM, and the bending process of the long member LM progresses.

[0037] Next, as shown in Figure 2(D), the processing device 200 rotates rollers RL1, RL2, and RL3 in the opposite direction to that shown in Figure 2(C), while maintaining the pressure of roller LR1, thereby moving the long member LM. This movement of the long member LM moves the workpiece position of the long member LM, further advancing the bending process of the long member LM.

[0038] Next, as shown in Figure 2(E), the processing device 200 maintains the indentation amount D1 of roller LR1 and rotates rollers RL1, RL2, and RL3 in the opposite direction to that shown in Figure 2(D), moving the long member LM so that its workpiece position coincides with the workpiece position in Figure 2(B).

[0039] The above describes one bending operation performed by the processing device 200. The amount by which roller RL1 is pressed in during this operation is called the indentation amount. Since the bending angle in the bending operation performed by the processing device 200 depends on this indentation amount, when the bending angle is the target physical quantity, the indentation amount becomes the processing condition.

[0040] Figure 5 is a diagram illustrating the parameters related to the bending process of the processing device. The parameters when the above processing is performed are as shown in Figure 5. Here, when the processing device 200 performs processing with a pressure D of roller RL1 and a feed amount L to the left and right of the workpiece position of the long member LM, the bending angle of the long member LM is A. Since the bending angle A in the bending process performed by the processing device 200 depends on the pressure D of roller RL1, if the target physical quantity is the bending angle A, the processing condition is the pressure D of roller RL1.

[0041] Returning to Figure 4, the target physical quantity setting unit 112 acquires the target physical quantity for each process calculated based on the number of processing steps and the target physical quantity of the final processed product (step S2). Here, since the number of processing steps is 3 and the final target bending angle is 90°, the difference between the bending angle in each process and the bending angle after the previous processing is divided equally, resulting in the target bending angle Atarget_1 in the first process being 30°, the target bending angle Atarget_2 in the second process being 60°, and the target bending angle Atarget_3 in the third process being 90°.

[0042] Next, the construction unit 111 constructs a prediction model PM (step S3). The constructed prediction model PM is stored in the prediction model storage area 151 of the storage unit 150. Here, the construction unit 111 constructs the prediction model PM from accumulated past processing data and simulation results. The construction unit 111 constructs the prediction model PM using, for example, actual physical quantities before processing, target physical quantities for processing, processing conditions, and actual physical quantities after processing as accumulated past processing data.

[0043] Next, the processing condition acquisition unit 113 calculates and acquires the indentation amount D1 as the processing condition MC1 for the first processing, which is calculated based on the prediction model PM and the target bending angle Atarget_1 for the first processing (step S4). Note that the indentation amount D1 as the processing condition MC1 for the first processing, which is the first processing among multiple processing operations, may be calculated using a pre-prepared correspondence table that shows the correspondence between the target bending angle Atarget_1 and the indentation amount D1.

[0044] Next, the processing apparatus 200 performs first processing based on the indentation amount D1 as the processing condition MC1 acquired by the processing condition acquisition unit 113 (step S5). Here, the processing apparatus 200 performs the processing shown in FIG. 2 with the roller RL1 set to the indentation amount D1.

[0045] Next, the physical quantity acquisition unit 115 acquires the actual bending angle Ameasure_1 after processing measured by the measurement unit 300 (step S6).

[0046] Next, the correction unit 114 corrects the prediction model PM stored in the prediction model storage area 151 of the storage unit 150 based on (the error between) the actual bending angle Ameasure_1 after the first processing and the target bending angle Atarget_1 for the first processing (step S7). That is, the processing condition acquisition unit 113 corrects the prediction model based on a target physical quantity in the N-th processing (where N is a natural number, and N is "1" herein) among a plurality of processings, and the physical quantity after the N-th processing. Note that the correction unit 114 does not have to correct the prediction model PM if the error between the actual bending angle Ameasure_1 after the first processing and the target bending angle Atarget_1 for the first processing is within an allowable range (3°).

[0047] Next, the processing condition acquisition unit 113 calculates and acquires the indentation amount D2 as the processing condition MC2 for the second processing, which is calculated based on the prediction model PM, the target bending angle A target_2 for the second processing, and the actual bending angle A measure_1 after the first processing (step S4). As described above, the processing condition acquisition unit 113 calculates and acquires the indentation amount D1 as the processing condition MC1 for the first processing, which is calculated based on the prediction model PM and the target bending angle A target_1 for the first processing. Therefore, the processing condition acquisition unit 113 uses the prediction model PM to acquire the respective processing conditions for the first processing and the second processing from the target physical quantity of each processing and the physical quantity after the processing which is the preceding step of each processing. That is, the processing condition acquisition unit 113 uses the prediction model to acquire the respective processing conditions for two mutually different processings included in a plurality of processings from the target physical quantity of each processing and the physical quantity after the processing which is the preceding step of each processing.

[0048] Next, the processing apparatus 200 performs the second processing based on the indentation amount D2 as the processing condition MC2 acquired by the processing condition acquisition unit 113 (step S5). Here, the processing apparatus 200 performs the processing shown in Fig. 2 by setting the roller RL1 to the indentation amount D2 from the position of the indentation amount D1.

[0049] Next, the physical quantity acquisition unit 115 acquires the actual bending angle A measure_2 after processing measured by the measurement unit 300 (step S6).

[0050] Next, the correction unit 114 corrects the prediction model PM stored in the prediction model storage area 151 of the storage unit 150 again based on the error between the actual bending angle Ameasure_2 after the second processing and the target bending angle Atarget_2 in the second processing (step S7). In other words, the correction unit 114 of the processing condition acquisition unit 113 further corrects the prediction model based on the target physical quantity in the (N+1)th processing and the actual physical quantity after the (N+1)th processing. Note that the correction unit 114 does not need to correct the prediction model PM if the error between the actual bending angle Ameasure_2 after the second processing and the target bending angle Atarget_2 in the second processing is within the allowable range (3°).

[0051] Next, the processing condition acquisition unit 113 calculates and acquires the indentation amount D3 as the processing condition MC3 for the third processing, which is calculated based on the prediction model PM, the target bending angle Atarget_3 in the third processing, and the actual bending angle Ameasure_2 after processing in the second processing (step S4).

[0052] Next, the processing device 200 performs a third processing step (step S5) based on the indentation amount D3, which is the processing condition MC3 acquired by the processing condition acquisition unit 113. Here, the processing device 200 performs the processing shown in Figure 2, with the indentation amount D3 being the indentation amount D3 from the position of indentation amount D2 of the roller RL1.

[0053] Next, the physical quantity acquisition unit 115 acquires the actual bending angle Ameasure_3 after processing, which was measured by the measurement unit 300 (step S6). After this, the processing control device 100 compares the actual bending angle Ameasure_3 after processing of the third processing with the final bending angle Atarget of the long member LM, and terminates the processing if the error is within an acceptable range.

[0054] Figure 6 is a diagram illustrating the operation flow of the processing control device 100 and the processing device 200. Here, we will explain the operation of the processing control device 100 and the processing device 200 when the processing control device 100 performs control according to the control flowchart shown in Figure 4. First, as shown in Figure 6(A), when the initial settings are input from the input unit 130 (step S1 in Figure 4), the target physical quantity setting unit 112 of the processing control device 100 sets the target bending angle Atarget_1 for the first processing to 30°, the target bending angle Atarget_2 for the second processing to 60°, and the target bending angle Atarget_3 for the third processing to 90°, as explained in step S2 in Figure 4.

[0055] Next, as shown in Figure 6(B), the processing condition acquisition unit 113 of the processing control device 100 calculates and acquires the amount D1 of roller RL1 being pressed as processing condition MC1 (step S4 in Figure 4).

[0056] Next, as shown in Figure 6(C), the processing device 200 performs processing using the amount D1 of roller RL1 being pressed as the processing condition MC1 for the first processing (step S5 in Figure 4). Then, as shown in Figure 6(D), the physical quantity acquisition unit 115 of the processing control device 100 acquires the actual bending angle Ameasure_1 after processing for the first processing (step S6 in Figure 4).

[0057] Next, as shown in Figure 6(E), the correction unit 114 of the processing control device 100 corrects the prediction model PM as described in step S7 of Figure 4. The correction unit 114 does not need to correct the prediction model PM if the error between the actual bending angle Measure_1 after the first processing and the target bending angle Atarget_1 in the first processing is within a predetermined tolerance range. Therefore, the correction unit 114 of the processing condition acquisition unit 113 can correct the prediction model PM based on the target bending angle Atarget_1 in the first processing and the actual bending angle Measure_1 after the first processing. In other words, the correction unit 114 of the processing condition acquisition unit 113 can correct the prediction model based on the target physical quantity in the Nth (N is a natural number) processing among multiple processing operations and the physical quantity after processing in the Nth processing operation. Next, the processing condition acquisition unit 113 calculates and acquires the amount of indentation D2 from the position D1 into which the roller RL1 was indented in the first processing, as the processing condition MC2 for the second processing. Therefore, the processing condition acquisition unit 113 uses the prediction model PM to acquire the processing conditions for the first processing and the second processing from the target physical quantity of each processing and the physical quantity after processing of the processing preceding each processing. In other words, the processing condition acquisition unit 113 uses the prediction model to acquire the processing conditions for two different processing steps included in a plurality of processing steps from the target physical quantity of each processing and the physical quantity after processing of the processing preceding each processing.

[0058] Next, as shown in Figure 6(F), the processing apparatus 200 performs processing using the indentation amount D2 as the processing condition MC2 for the second processing (step S5 in Figure 4).

[0059] Subsequently, as shown in Figure 6(G), the physical quantity acquisition unit 115 acquires the actual bending angle Ameasure_2 after the second processing (step S6 in Figure 4).

[0060] Next, as shown in Figure 6(H), the correction unit 114 of the processing control device 100 corrects the prediction model PM as described in step S7 of Figure 4. Note that the correction unit 114 does not need to correct the prediction model PM if the error between the actual bending angle Ameasure_2 after processing in the second processing and the target bending angle Ataget_2 in the second processing is within a predetermined tolerance range. Therefore, the correction unit 114 of the processing condition acquisition unit 113 corrects the prediction model PM based on the target bending angle Ataget_2 in the second processing and the actual bending angle Ameasure_2 after processing in the second processing. In other words, the correction unit 114 of the processing condition acquisition unit 113 further corrects the prediction model based on the target physical quantity in the N+1th processing and the actual physical quantity after processing in the N+1th processing. Next, the processing condition acquisition unit 113 calculates and acquires the amount of indentation D2 from the position D2 into which the roller RL1 was indented in the second processing, which is the processing condition MC3 for the third processing, calculated based on the corrected prediction model PM, the actual bending angle Ameasure_2 after processing in the second processing, and the target bending angle Atarget_3 in the third processing.

[0061] Next, as shown in Figure 6(I), the processing apparatus 200 performs processing using the indentation amount D3 as the processing condition MC3 for the third processing (step S5 in Figure 4).

[0062] Subsequently, as shown in Figure 6(J), the physical quantity acquisition unit 115 acquires the actual bending angle Ameasure_3 after the third processing (step S6 in Figure 4). As a result, the processing control device 100 confirms that the error between the actual bending angle Ameasure_3 measured after the third processing and the final target bending angle Ameasure_3 is within an acceptable range, and terminates the processing.

[0063] [Example] Figure 7 shows an example of a processed product processed by the processing device 200 when the processing control device 100 performed control according to the control flowchart shown in Figure 4. The processed product is a long member LM, as described above. Figure 7(A) shows the long member LM after processing in the first step. The processing device 200 (bending device) performed bending with a target bending angle Atarget_1 of 30° in the first step, and the actual bending angle Ameasure_1 after processing was 30.6°. That is, the error between the target bending angle Atarget_1 in the first step and the actual bending angle Ameasure_1 after processing in the first step was 0.6°, so the error was within the acceptable range (3°), and no correction was performed on the prediction model PM.

[0064] Figure 7(B) shows the long member LM after the second processing. The target bending angle Atarget_2 in the second processing was set to 60°, and the bending process was performed using the processing device 200. The actual bending angle Ameasure_2 after processing was 62.3°. That is, the error between the target bending angle Atarget_2 in the second processing and the actual bending angle Ameasure_2 after processing was 2.3°. Since the error was within the acceptable range (3°), no correction was made to the prediction model PM.

[0065] Figure 7(C) shows the long member LM after the third processing. The target bending angle Atarget_3 in the third processing was set to 90°, which is the final target bending angle Atarget, and the bending process was carried out using the processing device 200. The actual bending angle Ameasure_3 after processing was 88.1°. In other words, the error between the target bending angle Atarget_3 in the third processing and the actual bending angle Ameasure_3 after processing the third processing was 1.9°. Since the error was within the acceptable range (3°), no correction was made to the prediction model PM.

[0066] In the above processing, the predictive model PM, before processing the long member LM shown in Figure 7(A) to the state shown in Figure 7(B), is corrected if the error between the bending angle after the first processing (let's call the provisional bending angle after the first processing Temporary_Measure_1) and the target bending angle Atarget_1 in the first processing exceeds 3°, based on the provisional bending angle after the first processing Temporary_Measure_1 and the target bending angle Atarget_1 in the first processing. Furthermore, the prediction model PM, before processing the long member LM shown in Figure 7(B) to the state shown in Figure 7(C), is corrected again based on the provisional bending angle after the second processing (let's call the provisional bending angle after the second processing Temporary_Measure_2) and the target bending angle Atarget_2 in the second processing if the error between these two values ​​exceeds 3°.

[0067] The error between the actual bending angles Measure_1, Measure_2, and Measure_3 after processing in the first to third processes, and the target bending angles Atarget_1, Atarget_2, and Atarget_3 in each process, was within 3° in all cases, which falls within the acceptable range. Therefore, by using the processing control device 100 of this embodiment, the time required to construct the prediction model is reduced compared to constructing a prediction model for each processing process, as only one prediction model needs to be constructed. Furthermore, it is possible to keep the error between the actual physical quantity of the processed product measured after each process and the target physical quantity for each processing step within a predetermined range.

[0068] As described above, the processing control device 100 of this embodiment corrects the prediction model PM based on the actual bending angle Temporary_Measure_1 measured after the provisional first processing and the target bending angle Atarget_1 in the first processing, and uses the corrected prediction model to calculate the indentation amount D2 as the processing condition MC2 for the second processing based on the target bending angle Atarget_2 in the second processing. Furthermore, the prediction model PM is corrected again based on the actual physical quantity Temporary_Measure_2 measured after the provisional second processing and the target physical quantity Atarget_2 in the second processing, and uses the twice corrected prediction model PM to calculate the indentation amount D3 as the processing condition MC3 for the third processing. By repeating similar operations, instead of building a predictive model for each process, the predictive model built when calculating the processing conditions for the first process can be repeatedly corrected based on the target physical quantity of the previous process and the actual physical quantity measured after the previous process. In this way, compared to building a predictive model for each processing step, the time required to build the predictive model is reduced because the predictive model is built only once. Furthermore, one predictive model can be used for processing multiple processes, reducing the capacity of the storage unit 150 required to store the predictive model. In addition, easier management becomes possible. In contrast, when using the predictive model described in Patent Document 1 to perform processing multiple times as described above, the predictive model described in Patent Document 1 does not have data on the actual physical quantity of the processed product after each process as an input variable, so it is necessary to build a predictive model for each process. Therefore, a predictive model must be built for each process, which takes an enormous amount of time. Furthermore, when performing the multi-step processing described above using the machine learning model described in Non-Patent Document 1, the prediction model described in Non-Patent Document 1 does not have data on the actual physical quantities of the processed product after each processing step as input variables. Therefore, it becomes necessary to construct a prediction model for each processing step. Consequently, a prediction model must be constructed for each process, which would require an enormous amount of time.

[0069] <Second Embodiment> In the second embodiment, similar to the first embodiment, the processing control system 500 includes a processing control device 100, a processing device 200, and a measuring unit 300. In the processing control system 500 of the second embodiment, processing is performed on a metal member in four or more steps, including multiple types of processing. The multiple types of processing include bending and heat treatment, and we will focus on and explain two processing steps in which the preceding process is bending and the subsequent process is heat treatment. Therefore, for a metal member whose bending angle before processing (after the completion of the previous processing) is Ainitial and hardness is Hinital, the processing consists of two steps, with the first processing step (preceding process) being bending and the second processing step (subsequent process) being heat treatment. The bending process is the same as described in <First Embodiment>, and the explanation will be omitted. The heat treatment is one of annealing, quenching, or tempering. Specifically, the processing apparatus 200 includes a bending apparatus for performing bending as shown in Figure 2, and a heat treatment apparatus for performing heat treatment. Therefore, the measuring unit 300 includes a device capable of measuring the bending angle and a device capable of measuring the hardness of the metal member. Furthermore, the final target bending angle after the first and second processing is set to A target, and the final target hardness is set to H target.

[0070] It should be assumed that processing has been performed prior to the first processing, and that the target bending angle in the processing prior to the first processing is Atarget_0 and the target hardness is Harget_0. Furthermore, it should be assumed that the actual bending angle after processing prior to the first processing is Ameasure_0 and the actual hardness is Hmeasure_0. In other words, the actual bending angle Ameasure_0 after processing prior to the first processing is equal to Ainitial, and the actual hardness Hmeasure_0 is equal to Hinitial.

[0071] In the prediction model PM shown in Figure 3, the number of target physical quantities QN for the subsequent processing and QL for the target physical quantities QL for the previous processing are equal to the number of processing types. Here, since bending is performed in the first processing and heat treatment is performed in the second processing, the target physical quantities QN for the subsequent processing and QL for the target physical quantities QL for the previous processing are two types: bending angle A and hardness H. Also, the processing conditions MC are two types: the indentation amount D of roller RL1 and the heat treatment temperature T.

[0072] Figure 8 is a diagram illustrating the operation flow of the processing control device 100 and processing apparatus 200 of the second embodiment. First, as shown in Figure 8(A), initial settings are input from the input unit 130 (step S1 in Figure 4), and the target physical quantity setting unit 112 of the processing control device 100 sets the number of processing steps n to 2 (although actual processing is 4 or more, for convenience, we will use 2 here to explain by focusing on two consecutive processing steps), the first processing step is bending, and the second processing step is heat treatment, so bending and heat treatment will each be performed once. Therefore, since the final target bending angle is Atarget and the final target hardness is Htarget, the target bending angle Atarget_1 in the first process is set as Atarget, the target hardness Htarget_1 in the first process is set as Hintial, the target bending angle Atarget_2 in the second process is set as Atarget, and the target hardness Htarget_2 in the second process is set as Htarget.

[0073] Next, as shown in Figure 8(B), the processing condition acquisition unit 113 of the processing control device 100 calculates and acquires the processing condition MC1, which is the amount of indentation D1 of the roller RL1 and the heat treatment temperature T1, using the prediction model PM, with the target bending angle Atarget_1 set to Atarget and the target hardness Harget_1 set to Hintial (step S4 in Figure 4). Note that in the first processing, since bending is performed, the heat treatment temperature T1 is 0.

[0074] Next, as shown in Figure 8(C), the processing apparatus 200 performs processing (bending) using the processing conditions MC1 for the first processing, which are the amount of indentation D1 of the roller RL1 and the heat treatment temperature T1 (=0) (step S5 in Figure 4). After that, as shown in Figure 8(D), the physical quantity acquisition unit 115 of the processing control device 100 acquires the actual bending angle Ameasure_1 and the actual hardness Hmeasure_1 after processing for the first processing (step S6 in Figure 4).

[0075] Next, as shown in Figure 8(E), the correction unit 114 of the processing control device 100 corrects the prediction model PM as described in step S7 of Figure 4. Note that the correction unit 114 does not need to correct the prediction model PM if the error between the actual bending angle Ameasure_1 and actual hardness Hmeasure_1 after processing in the first processing and the target bending angle Atarget_1 and target hardness Htarget_1 in the first processing is within a predetermined tolerance range. Therefore, the correction unit 114 of the processing condition acquisition unit 113 can correct the prediction model PM based on the target bending angle Atarget_1 and target hardness Htarget_1 in the first processing and the actual bending angle Ameasure_1 and actual hardness Hmeasure_1 after processing in the first processing. In other words, the correction unit 114 of the processing condition acquisition unit 113 can correct the prediction model based on the target physical quantity in the Nth (N is a natural number) processing step among multiple processing steps, and the physical quantity after processing in the Nth processing step. Next, as explained in step S4 of Figure 4, the processing condition acquisition unit 113 calculates and acquires the indentation amount D2 from the indentation position D1 of the roller RL1 in the first processing step, and the heat treatment temperature T2, as processing conditions MC2 for the second processing step. Note that in the second processing step, heat treatment is performed, so the indentation amount D2 is 0. As previously described, the processing condition acquisition unit 113 uses the prediction model PM to calculate and acquire the indentation amount D1 of the roller RL1 as processing condition MC1 and the heat treatment temperature T1, with the target bending angle Atarget_1 as Atarget and the target hardness Htarget_1 as Hintial in the first processing step. Therefore, the processing condition acquisition unit 113 uses the prediction model PM to acquire the processing conditions for the first processing and the second processing from the target physical quantity of each processing and the post-processing physical quantity of the processing preceding each processing. In other words, the processing condition acquisition unit 113 uses the prediction model to acquire the processing conditions for two different processing steps included in a plurality of processing steps from the target physical quantity of each processing and the post-processing physical quantity of the processing preceding each processing.

[0076] Next, as shown in Figure 8(F), the processing apparatus 200 performs processing (heat treatment) using the processing conditions MC2 for the second processing, which are the indentation amount D2 (=0) and the heat treatment temperature T2 (step S5 in Figure 4).

[0077] Subsequently, as shown in Figure 8(G), the physical quantity acquisition unit 115 acquires the actual bending angle Ameasure_2 and the actual hardness Hmeasure_2 after the second processing (step S6 in Figure 4). As a result, the processing control device 100 confirms that the error between the actual bending angle Ameasure_2 and the actual hardness Hmeasure_2 measured after the second processing and the final target bending angle Atarget and final target hardness Htarget is within an acceptable range, and proceeds to the next processing. The above describes the first and second processing in the second embodiment.

[0078] As described above, in the machining control device 100 of the second embodiment, even in the case of machining with multiple processes where the machining content differs in each process, the prediction model PM can be corrected based on the error between the actual physical quantity measured after the machining of the previous process and the target physical quantity in the machining of the previous process, and the machining conditions for subsequent processes can be calculated using the corrected prediction model. Therefore, instead of constructing a prediction model for each process, the prediction model constructed when calculating the machining conditions for the first process can be repeatedly corrected based on the error between the target physical quantity of the previous process and the actual physical quantity of the machined product measured after the machining of the previous process. By doing so, the time required to construct the prediction model can be reduced because the prediction model only needs to be constructed once compared to when a prediction model is constructed for each machining process. In addition, multiple machining processes can be performed with a single prediction model, and the capacity of the storage unit 150 required to store the prediction model can be reduced. Furthermore, easier management becomes possible. In contrast, when performing the above-described multi-step processing using the prediction model described in Patent Document 1, the prediction model described in Patent Document 1 does not have input variables for data on the actual physical quantities of the processed product after each processing step. Therefore, it is necessary to construct a prediction model for each processing step. Consequently, a prediction model must be constructed for each step, which will require an enormous amount of time. Furthermore, when performing the above-described multi-step processing using the machine learning model described in Non-Patent Document 1, the prediction model described in Non-Patent Document 1 does not have input variables for data on the actual physical quantities of the processed product after each processing step. Therefore, a prediction model must be constructed for each step, which will require an enormous amount of time.

[0079] <Modifications of Embodiments> The present invention is not limited to the above embodiments, and can be implemented in various forms without departing from the spirit thereof. For example, the following modifications are also possible. Furthermore, in the above embodiments, some of the configurations implemented by hardware may be replaced with software, and conversely, some of the configurations implemented by software may be replaced with hardware.

[0080] The first and second embodiments are examples of machining processes involving multiple steps using the machining control device 100, and are not limited to the types of machining processes and the number of processes. The number of machining processes may be four or more. Furthermore, the machining control device 100 may be used for machining processes that involve three or more types of machining. In addition, different forming processes and molding processes (such as press working), removal processes (such as cutting, polishing, grinding, and electrical discharge machining), and additive processes (such as coating, lamination, and joining) may be arbitrarily selected from feed bending processes. The measuring instruments, etc., corresponding to the measurement unit 300 will be instruments capable of measuring the physical quantities corresponding to the machining processes performed, and the number of instruments will correspond to the number of machining processes performed.

[0081] Furthermore, regarding the processed product, in the first embodiment described above, it is a long-shaped member LM made from a metal material, and in the second embodiment, it is a metal member, but this is not limited to these, and any non-metallic member of any shape may be used.

[0082] <Modification 1> In cases where there are multiple processes that perform similar processing, the target physical quantity setting unit 112 does not need to acquire the target physical quantity for each process, such that the difference between the target bending angle in each process and the target bending angle in the previous process is obtained by dividing the target physical quantity of the final processed product by the number of processes that perform similar processing. In the first embodiment described above, in a process that bends to 90° in 3 processes, the first to third processes bend by 30° each, but this is not limited to this. For example, the target bending angle in the first process may be 40°, the target bending angle in the second process may be 70°, and the target bending angle in the third process may be 90°.

[0083] <Modification 2> The processing condition acquisition unit 113 previously acquired processing conditions for the first of multiple processing operations that were calculated using a predictive model, but this is not limited to that. For example, processing conditions may be acquired without using a predictive model, by using a correspondence table between target physical quantities and processing conditions that differs depending on the processed product and the content of the processing.

[0084] <Variation 3> The construction unit 111 constructed the predictive model using accumulated past molding data and simulation results. However, it is not limited to this, and it may be constructed using any data.

[0085] The present invention can also be realized in the following form: [Application Example 1] A processing control device for controlling a processing apparatus that performs multiple processing operations on a workpiece, comprising: a target physical quantity setting unit for setting target physical quantities for each of the multiple processing operations; a physical quantity acquisition unit for acquiring physical quantities of the workpiece after each processing operation; a processing condition acquisition unit for acquiring processing conditions for each processing operation; and a storage unit for storing a prediction model that associates the physical quantities after processing of a processing operation that is relatively the preceding process among any two consecutive processing operations included in the multiple processing operations, the target physical quantities of a processing operation that is relatively the following process, and the processing conditions of the following process, wherein the processing condition setting unit uses the prediction model to acquire the processing conditions for each of two different processing operations included in the multiple processing operations from the target physical quantities of each processing operation and the physical quantities after processing of the processing operation that is the preceding process for each processing operation. [Application Example 2] A processing control device according to Application Example 1, wherein the processing condition acquisition unit includes a correction unit that corrects the prediction model based on the target physical quantity of the Nth (where N is a natural number)-th processing among the plurality of processing and the physical quantity after processing of the Nth processing, and the control device acquires the processing conditions for the (N+1)th processing from the physical quantity after processing of the Nth processing and the target physical quantity of the (N+1)th processing using the prediction model corrected by the correction unit. [Application Example 3] A processing control device according to Application Example 1 or Application Example 2, wherein the correction unit further corrects the prediction model based on the target physical quantity of the (N+1)th processing among the plurality of processing and the actual physical quantity after processing of the (N+1)th processing. [Application Example 4] A processing control device according to any one of Application Examples 1 to 3, wherein the processing device is a feed bending device, the processed product is a long member made from a metal material, and the physical quantity is a quantity related to the bending of the processed product.[Application Example 5] A method for controlling a processing apparatus that performs multiple processing operations on a workpiece, comprising: a target physical quantity setting step of setting a target physical quantity for each of the multiple processing operations; a physical quantity acquisition step of acquiring the physical quantity of the workpiece after each processing operation; a processing condition acquisition step of acquiring the processing conditions for each processing operation; and a storage step of storing a prediction model that associates the physical quantity after processing of a processing operation that is relatively the preceding process among any two consecutive processing operations included in the multiple processing operations, the target physical quantity of a processing operation that is relatively the following process, and the processing conditions of the following process, wherein the processing condition setting step includes a step of using the prediction model to acquire the processing conditions for each of two different processing operations included in the multiple processing operations from the target physical quantity of each processing operation and the physical quantity after processing of the processing operation that is the preceding process for each processing operation.

[0086] 100... Processing control device 110... CPU 111... Construction unit 112... Target physical quantity acquisition unit 113... Processing condition acquisition unit 114... Correction unit 115... Physical quantity acquisition unit 150... Memory unit 151... Prediction model memory area 130... Input unit 140... Display unit 200... Processing device 300... Measurement unit 500... Processing control system LM... Long metal member QL... Actual physical quantity after processing in the previous process QN... Target physical quantity in the next process PM... Prediction model C1... Intersection line between the plane QL representing the actual physical quantity of the processed product after the previous process and the prediction model PM C2... Intersection line between the plane QN representing the target physical quantity of the processed product in the next process and the prediction model PM RL1, RL2, RL3... Rollers (of the feed bending device) A... Bending angle D... Push amount (of roller RL1) W1, W2... Feed direction L: Feed rate H: Hardness T: Heat treatment temperature MC, MC1, MC2, MC3: Machining conditions

Claims

1. A processing control device for controlling a processing apparatus that performs multiple processing operations on a workpiece, comprising: a target physical quantity setting unit for setting target physical quantities for each of the multiple processing operations; a physical quantity acquisition unit for acquiring physical quantities of the workpiece after each processing operation; a processing condition acquisition unit for acquiring processing conditions for each processing operation; and a storage unit for storing a prediction model that associates the physical quantities after processing of a processing operation that is relatively the preceding process among any two consecutive processing operations included in the multiple processing operations, the target physical quantities of a processing operation that is relatively the following process, and the processing conditions of the following process, wherein the processing condition setting unit uses the prediction model to acquire the processing conditions for each of two different processing operations included in the multiple processing operations from the target physical quantities of each processing operation and the physical quantities after processing of the processing operation that is the preceding process for each processing operation.

2. A processing control device according to claim 1, wherein the processing condition acquisition unit includes a correction unit that corrects the prediction model based on the target physical quantity of the Nth (where N is a natural number) processing among the plurality of processing and the physical quantity after processing of the Nth processing, and the control device acquires the processing conditions for the (N+1)th processing from the physical quantity after processing of the Nth processing and the target physical quantity of the (N+1)th processing using the prediction model corrected by the correction unit.

3. A machining control device according to claim 2, wherein the correction unit further corrects the prediction model based on the target physical quantity in the (N+1)th machining process among the plurality of machining processes and the actual physical quantity after machining in the (N+1)th machining process.

4. A processing control device according to any one of claims 1 to 3, wherein the processing device is a feed bending device, the processed product is a long member made from a metal material, and the physical quantity is a quantity relating to the bending of the processed product.

5. A method for controlling a processing apparatus that performs multiple processing operations on a workpiece, comprising: a target physical quantity setting step of setting a target physical quantity for each of the multiple processing operations; a physical quantity acquisition step of acquiring a physical quantity of the workpiece after each processing operation; a processing condition acquisition step of acquiring processing conditions for each processing operation; and a storage step of storing a prediction model that associates the physical quantity after processing of a processing operation that is relatively the preceding process among any two consecutive processing operations included in the multiple processing operations, the target physical quantity of a processing operation that is relatively the following process, and the processing conditions of the following process, wherein the processing condition setting step includes a step of using the prediction model to acquire the processing conditions for each of two different processing operations included in the multiple processing operations from the target physical quantity of each processing operation and the physical quantity after processing of the processing operation that is the preceding process for each processing operation.

Images