Parameter adjustment device and computer-readable recording medium

Abstract

A parameter adjustment device according to the present disclosure includes: a simulation unit that simulates an operation of a control object in which a value of a predetermined control parameter is set, on the basis of a control object model obtained by modeling an operation of each part of the control object; a first search unit that performs a simulation process while changing the values of the control parameters within a predetermined first range, and searches for candidates of a group of values of the control parameters near the boundary of a predetermined constraint condition on the basis of a first index value calculated from the result of the simulation process; a candidate implementation unit that sets a set of candidate control parameter values to a control target to perform operation verification, and calculates a second index value on the basis of the result; and a search range determination unit that determines a second range that is narrower than the first range on the basis of the first index value and the second index value.

Description

Technical Field

[0001] The present disclosure relates to a parameter adjustment device and a computer-readable recording medium. Background Art

[0002] As a method of adjusting control parameters set when controlling industrial machinery, there is a method of making flexible use of simulation. In this method, first, a predetermined first range is determined in advance for the control parameters, and simulation is repeatedly performed while changing the value of the control parameter within the first range. Then, candidates for the optimal value of the control parameter are extracted based on the results of the simulation. After that, a predetermined second range is determined near each of the extracted candidates for the value of the control parameter. Then, the machinery is repeatedly operated while adjusting the control parameter within the second range. Based on the results of this operation, the optimal value of the control parameter is determined (for example, Patent Document 1, etc.).

[0003] Prior Art Documents

[0004] Patent Documents

[0005] Patent Document 1: Japanese Unexamined Patent Application Publication No. 2017-102619 Summary of the Invention

[0006] Problems to be Solved by the Invention

[0007] If the search range of the control parameter is set narrow, it may not be possible to obtain an appropriate control parameter. However, if the search range of the control parameter is set wide, the calculation of the simulation and the operation of using the machinery take a long time. Especially when actually operating the machinery to adjust the control parameter, the burden on the operator is large.

[0008] Therefore, a technique for appropriately limiting within the search range of the control parameter is desired.

[0009] Means for Solving the Problems

[0010] One aspect of the present disclosure is a parameter adjustment device, comprising: a controlled object model that models the operations of each part of an industrial machine that is a controlled object; a simulation unit that simulates the operations of the controlled object with a predetermined control parameter value set based on the controlled object model; a first search unit that repeatedly performs an operation of varying the value of the control parameter within a predetermined first range and performing a simulation process using the simulation unit, and searches for a candidate for a set of values of the control parameter that obtains the first index value near the boundary of a predetermined constraint condition based on a first index value calculated from the result of the simulation process; a candidate implementation unit that sets the set of values of the control parameter that is a candidate searched by the first search unit to the controlled object and performs action verification, and calculates a second index value based on the result of the action verification; and a search range determination unit that determines a second range, which is a range of values narrower than the first range, based on the first index value and the second index value. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 FIG. is a schematic hardware configuration diagram of the parameter adjustment device according to the first embodiment of the present disclosure.

[0012] Figure 2 FIG. is a diagram schematically showing the functions of the parameter adjustment device according to the first embodiment of the present disclosure as a block diagram.

[0013] Figure 3 FIG. is a flowchart schematically showing the process related to search preparation.

[0014] Figure 4 FIG. is a flowchart schematically showing the first search process.

[0015] Figure 5 FIG. is a flowchart schematically showing the second range determination process.

[0016] Figure 6 FIG. is a flowchart schematically showing the second search process. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] Hereinafter, each embodiment of the present disclosure will be described with reference to the drawings.

[0018] [First Embodiment]

[0019] Figure 1FIG. 0 is a schematic hardware structure diagram of the main part of a parameter adjustment device according to an embodiment of the present disclosure. The parameter adjustment device 1 of this embodiment can be installed, for example, as a control device that controls industrial machinery based on a control program. In addition, the parameter adjustment device 1 can be installed, for example, as a personal computer provided together with a control device that controls industrial machinery, a computer connected to the control device via a wired / wireless network, a fog computer, a cloud server, or other computers. In this embodiment, an example is shown in which the parameter adjustment device 1 is installed as a computer connected to a control device that controls industrial machinery 4 via a network 5.

[0020] The CPU 11 included in the parameter adjustment device 1 of this embodiment is a processor that controls the parameter adjustment device 1 as a whole. The CPU 11 reads out the system program stored in the ROM 12 via the bus 22 and controls the entire parameter adjustment device 1 according to the system program. Temporary calculation data, display data, and various data input from the outside are temporarily stored in the RAM 13.

[0021] The non-volatile memory 14 is composed of, for example, a memory backed up by a battery (not shown) or an SSD (Solid State Drive), and maintains the storage state even when the power supply of the parameter adjustment device 1 is turned off. Programs or data read in from an external device 72 via the interface 15, programs or data input via the input device 71, programs or data obtained from the industrial machinery 4 or other devices via the network 5, etc. are stored in the non-volatile memory 14. Among the stored data, for example, data related to physical quantities such as motor current, voltage, torque, position, speed, and acceleration of the drive unit detected by the sensor 8 installed in the industrial machinery 4 may be included. The programs and data stored in the non-volatile memory 14 can also be expanded in the RAM 13 during execution / use. In addition, various system programs such as a known analysis program are pre-written in the ROM 12.

[0022] The interface 15 is an interface for connecting the CPU 11 of the parameter adjustment device 1 and an external device 72 such as a USB device. System programs, programs related to the operation of the industrial machinery 4, setting data, etc. are read in from the external device 72 side. In addition, programs, setting data, etc. generated and edited within the parameter adjustment device 1 can be stored in an external storage unit via the external device 72.

[0023] The interface 20 is an interface for connecting the CPU 11 of the parameter adjustment device 1 and a wired or wireless network 5. The network 5 can communicate using technologies such as serial communication like RS-485, Ethernet (registered trademark) communication, optical communication, wireless LAN, Wi-Fi (registered trademark), Bluetooth (registered trademark), etc. Connected to the network 5 are a control device for controlling the industrial machine 4, a fog computer 6, a cloud server 7, etc., and data is exchanged with the parameter adjustment device 1.

[0024] In the display device 70, various data read into the memory, data obtained as a result of executing a program, etc. are output and displayed via the interface 17. In addition, an input device 71 composed of a keyboard, a pointing device, etc. transmits instructions, data, etc. based on the operator's operations to the CPU 11 via the interface 18.

[0025] Figure 2 This is a diagram showing the functions of the parameter adjustment device 1 of the first embodiment of the present disclosure as a schematic block diagram. Each function of the parameter adjustment device 1 of this embodiment is realized by the CPU 11 of the parameter adjustment device 1 shown Figure 1 executing a system program to control the operations of each part of the parameter adjustment device 1.

[0026] The parameter adjustment device 1 of this embodiment includes an analog unit 100, a first search unit 110, a candidate implementation unit 120, a search range determination unit 130, a second search unit 140, and an output unit 150. In addition, in the RAM 13 or non-volatile memory 14 of the parameter adjustment device 1, an area for pre-storing a control object model that models the actions of the industrial machine 4 to be controlled, namely, a model storage unit 210, is prepared. Moreover, in the RAM 13 or non-volatile memory 14 of the parameter adjustment device 1, an evaluation program 220 for simulation processing and actual action control of the control object is pre-stored.

[0027] The analog unit 100 executes a predetermined simulation process for simulating the actions of the control object based on the control object model stored in the model storage unit 210. The control object model stored in the model storage unit 210 is a model that models the actions of each part including the drive unit such as the servo motor of the industrial machine 4, the transmission unit such as the ball screw, and the movable unit such as the workbench. The control object model is represented by mathematical expressions, transfer functions, block diagrams, etc. that reflect the characteristics of the industrial machine 4 as the control object. By preparing the control object model, the actions of the industrial machine 4 as the control object can be simulated.

[0028] The simulation process executed by the simulation unit 100 is as follows: By performing calculations based on the controlled object model, the operation of the industrial machine 4 as the controlled object when a predetermined command information is indicated is virtually executed. As a result of the simulation process, virtual feedback information of the industrial machine 4 as the controlled object is generated. The virtual feedback information exemplifies virtual physical quantities related to the operation of the industrial machine 4, such as the position, speed, acceleration, and torque of the motor in each operation cycle.

[0029] When the simulation unit 100 performs the simulation process, the provided control parameters are reflected in the calculations of the controlled object model. The control parameters vary depending on the type of the industrial machine 4 as the controlled object. For example, when the industrial machine 4 as the controlled object is a machine tool that operates by a servo motor, the control parameters exemplify set values related to each motor, such as linear acceleration, linear jerk, angular velocity difference, interpolation post-acceleration / deceleration time constant, velocity gain, position loop gain, and feedforward coefficient. In addition, in the case of an electric discharge machine, in addition to the above, set values related to the power supply, such as electric discharge machining pulse voltage and electric discharge machining current, are also exemplified.

[0030] The first search unit 110 repeatedly performs the simulation process of the simulation unit 100 while varying the control parameters within the first range related to the control parameters. Then, based on the result of the simulation process, a group of values of the control parameters within the first range that is the most preferable within the range where a predetermined index value satisfies a predetermined constraint condition and a group of values of the control parameters in the vicinity thereof are searched for as candidates for the group of values of the control parameters. The predetermined index value is an index for restricting or evaluating the operation result of the controlled object. As an example of the index value, for example, machining accuracy (machining error), machining surface quality (position deviation, vibration), machining time (cycle time), etc. are exemplified.

[0031] The first range is the range of values that can be taken by each control parameter in the simulation process. The first range can be set by the operator or determined according to a predetermined rule based on the specifications of the industrial machine 4 as the controlled object, the command information included in the evaluation program 220 for simulation, etc. The first range is a predetermined range in which the index value for evaluating the result of the simulation process can be calculated. Since the first range is the search range in the simulation process of the simulation unit 100, it is okay even if it is set to a relatively wide range. For example, the range from the minimum value to the maximum value that can be set for each control parameter can be set as the first range, or the range of values that can be set from the command information can be set as the first range.

[0032] The first search unit 110 causes the simulation unit 100 to perform a simulation process based on predetermined instruction information included in the evaluation program 220 while varying the values of the respective control parameters within a first range. Then, as a result thereof, virtual feedback information is obtained, and a predetermined index value is evaluated based on the obtained virtual feedback information. The first search unit 110 executes this process using known algorithms such as machine learning-based search such as reinforcement learning and Bayesian optimization, rule-based search, and a predetermined exhaustive search algorithm. When appropriate constraint conditions and evaluation methods are set, as a result of the search by the first search unit 110, a set of values of at least one or more control parameters near the boundary of the constraint conditions is obtained. The first search unit 110 may store the result of the simulation process (index value), the determination result of the constraint conditions, and the evaluation result for the set of values of the evaluated control parameters. The first search unit 110 outputs to the candidate execution unit 120 a set of values of the control parameters that is the best within the range that satisfies the constraint conditions for a group, or two sets of values of the control parameters near the boundary of the constraint conditions, that is, a first set of values of the control parameters that satisfies the constraint conditions, and a second set of values of the control parameters that does not satisfy the constraint conditions.

[0033] The candidate execution unit 120 sets, for the industrial machine 4 that is the control object, the set of values of the control parameters near the boundary of the constraint conditions searched by the first search unit 110, and instructs to perform action verification based on the instruction information included in the evaluation program 220. Then, as an action result thereof, feedback information related to the action is obtained from the industrial machine 4. Then, an index value is calculated in order to evaluate the obtained feedback information.

[0034] The search range determination unit 130 determines a second range obtained by correcting the first range based on the action result obtained by the candidate execution unit 120 performing action verification in the industrial machine 4 that is the actual control object, for the set of values of the control parameters near the boundary of the constraint conditions searched by the first search unit 110.

[0035] The search range determination unit 130 may also determine a predetermined range based on the value group of the control parameter near the boundary of the constraint condition searched by the first search unit 110 as the second range when the value group of the control parameter near the boundary of the constraint condition searched by the first search unit 110 is one. For example, the range from the lower limit value of the first range of each control parameter to the value of the control parameter near the boundary of the constraint condition may be set as the second range. Alternatively, the range from the upper limit value of the first range of each control parameter to the value of the control parameter near the boundary of the constraint condition may be set as the second range. Regarding the second range of the value of each control parameter, when the result of the action verification performed in the industrial machine 4 that is the actual control object by the candidate implementation unit 120 does not satisfy the constraint condition, the second range may be determined in the direction that satisfies the constraint condition based on the value of the control parameter near the boundary of the constraint condition. Alternatively, when the constraint condition is not satisfied, the second range may be determined in the direction of improved evaluation based on the value of the control parameter near the boundary of the constraint condition.

[0036] The search range determination unit 130 may also determine the second range based on the processing results obtained by respectively performing simulation processing on the value groups of the control parameters near the boundary of the constraint condition and the results obtained by performing action verification on the industrial machine 4 that is the actual control object when the value group of the control parameter near the boundary of the constraint condition searched by the first search unit 110 is two. For example, when the result of the action verification performed in the industrial machine 4 that is the actual control object is that the constraint condition is satisfied even when operating with any value group of the control parameter, the search range determination unit 130 calculates the difference between the respective index values of the processing result of the simulation processing and the result of the action verification performed in the industrial machine 4 that is the actual control object as the modeling error for the value group of the second control parameter. Then, considering the calculated modeling error, the value group of the third control parameter that does not satisfy the constraint condition is selected from the value groups of the control parameters searched by the first search unit 110. Then, the range between the selected value group of the third control parameter and the value group of the second control parameter may be determined as the second range.

[0037] Alternatively, when the result of the action verification performed in the industrial machine 4 that is the actual control object is that the index value satisfies the constraint condition when operating with one value group of the control parameter and does not satisfy the constraint condition when operating with the other value group of the control parameter, the range between the value group of the first control parameter and the value group of the second control parameter may be determined as the second range.

[0038] Also, as a result of conducting an operation verification on the industrial machine 4 which is the actual control object, when the constraint conditions are not satisfied even when operating with any set of values of the control parameters, the search range determination unit 130 calculates the difference between the respective index values of the processing result of the simulation process and the result of the operation verification conducted on the industrial machine 4 which is the actual control object as the modeling error for the set of values of the first control parameter. Then, considering the calculated modeling error, it selects a set of values of the third control parameter that satisfies the constraint conditions from the sets of values of the control parameters searched by the first search unit 110. Then, it is sufficient to determine the range between the selected set of values of the third control parameter and the set of values of the first control parameter as the second range.

[0039] The second search unit 140 searches for an appropriate set of control parameters while changing the control parameters within the second range determined by the search range determination unit 130. The second search unit 140 of the present embodiment repeatedly performs an operation of causing the industrial machine 4 which is the actual control object to operate according to the instruction information included in the evaluation program 220. Then, based on the result of the operation of the industrial machine 4 which is the actual control object, it searches for the set of values of the control parameter within the range of the set of values of the control parameter within the second range whose index value satisfies the constraint conditions and is the most optimal. The second search unit 140 uses a known algorithm such as machine learning type search such as reinforcement learning and Bayesian optimization, rule-based search, and a predetermined full search algorithm to execute this process. The second search unit 140 causes the industrial machine 4 which is the actual control object to operate, and calculates the index value based on the result of its operation. Then, it determines whether the calculated index value satisfies the constraint conditions and conducts an evaluation using the index value. The operation of the industrial machine 4 with a set of values of the control parameter takes the same time as the actual cycle time. However, since the second range determined by the search range determination unit 130 is sufficiently narrower than the first range, this search does not take a large amount of time. The second search unit 140 searches for the set of values of the best control parameter that satisfies at least one set of constraint conditions. The second search unit 140 outputs the searched set of values of the control parameter to the output unit 150.

[0040] The output unit 150 outputs the set of values of the control parameter searched by the second search unit 140 to a predetermined device, equipment, etc. The output unit 150 can, for example, display the set of values of the control parameter on the display device 70, or record it in a predetermined area provided on the non-volatile memory 14. In addition, it can be sent and set to the control device of the industrial machine 4, or sent to an upper computer such as the fog computer 6 or the cloud server 7.

[0041] Use Figures 3 to 6 , a specific operation example of the parameter adjustment device 1 of the present embodiment having the above structure will be described.

[0042] In this operation example, consider the case where the industrial machine 4 is a machine tool whose respective axes are actuated by a servo motor. The linear acceleration [mm / sec 2 and the angular velocity difference [mm / sec] in each of the axes of the industrial machine 4 are used as control parameters to be searched for. In addition, the maximum value of the machining error and the cycle time are selected as index values, the condition that the maximum value of the machining error is below a predetermined target error is set as a constraint condition, and the smaller the cycle time, the more preferable, is set as the evaluation method. In addition, as the index value, the machining surface quality, etc. can also be used. In this case, it is only necessary to set the value of the predetermined machining surface quality as the constraint condition. The machining error and the cycle time are in a trade-off relationship. If the machining error is to be reduced, there is a tendency for the cycle time to be extended. If the cycle time is to be shortened, the machining error becomes larger. When the machining error is within the target error, among the sets of values of any control parameters near its boundary, there is a solution that minimizes the cycle time.

[0043] Figure 3 is a flowchart showing a schematic process related to search preparation.

[0044] First, the simulation unit 100 reads out, according to the operator's instruction, the model of the machine tool whose respective axes are actuated by a servo motor from the model storage unit 210 (step SA01). Next, the first search unit 110 selects, according to the operator's instruction, the linear acceleration and the angular velocity difference as the control parameters to be searched for (step SA02). In addition, as a constraint condition, the machining error is set to be 10 [μm] or less, and as an evaluation method, it is set that the smaller the cycle time, the higher the evaluation (step SA03).

[0045] Next, the first search unit 110 sets a first range for each control parameter (step SA04). The first range is set to 10 to 10000 [mm / sec 2 for the linear acceleration and 10 to 5000 [mm / sec] for the angular velocity difference. This setting can also be based on the operator's instruction. In addition, it is also possible to inquire of the control device of the industrial machine 4 to be controlled about the maximum and minimum values of each control parameter and make the setting according to the response.

[0046] Figure 4 is a flowchart showing a schematic process of the first search process.

[0047] If the above settings are completed and the operator instructs the start of the search, the first search unit 110 selects an initial value from the first range for each control parameter and sets the simulation model read by the simulation unit 100 (step SB01). Next, the first search unit 110 sets the evaluation program 220 to be executed in the simulation process for the simulation unit 100 (step SB02). Then, the first search unit 110 instructs the selected simulation model to execute the simulation process with the set value of the control parameter.

[0048] The simulation unit 100 executes the simulation process according to the instruction of the first search unit 110 (step SB03). When the simulation process ends, the first search unit 110 obtains the processing result from the simulation unit 100 (step SB04). The obtained processing result is the machining path and cycle time used for the calculation of the index value. The first search unit 110 calculates the index value based on the obtained processing result (step SB05). The machining error as the index value can be calculated as the maximum value of the difference between the path indicated by the instruction information and the machining path in the virtual machining during the simulation process. The cycle time directly uses the cycle time obtained as the processing result as the index value. When the search in the first range has not ended at this time point (No in step SB06), the first search unit 110 sets the simulation model after adjusting the value of the control parameter (step SB07). Then, the process proceeds to step SB03. On the other hand, when the search in the first range ends at this time point (Yes in step SB06), the control parameter near the boundary of the constraint condition is selected as a candidate from the searched control parameters (step SB08). As described above, the processing of steps SB03 to SB07 is executed using a known search algorithm such as machine learning-based search, rule-based search, a prescribed full search algorithm, and Bayesian optimization.

[0049] Here, as a result of the first search process, the first search unit 110 selects the set A of the values of the first control parameters that satisfy the constraint conditions (linear acceleration = 2300 [mm / sec 2 , angular velocity difference = 1100 [mm / sec], at this time, the machining error is 9 [μm], and the cycle time is 35.2 [sec]), and the set B of the values of the second control parameters that do not satisfy the constraint conditions (linear acceleration = 2400 [mm / sec 2 , angular velocity difference = 1200 [mm / sec], at this time, the machining error is 12 [μm], and the cycle time is 33.0 [sec]) as candidates for the control parameters.

[0050] Figure 5 It is a flowchart showing the schematic process of the second range determination process.

[0051] If a candidate for a set of control parameter values is selected, the candidate implementation unit 120 sets the value of the selected candidate control parameter for the industrial machine 4 that is the actual control object, performs action verification based on the instruction information included in the evaluation program 220, and obtains the action result when operating with each candidate for the set of control parameter values (step SC01). The obtained action result is the machining path and cycle time used for the calculation of the index value. The candidate implementation unit 120 calculates the index value based on the obtained action result (step SC02). The machining error as the index value can be calculated as the maximum value of the difference between the path indicated by the instruction information and the actual machining path. The cycle time directly uses the cycle time obtained as the processing result as the index value.

[0052] Next, the search range determination unit 130 determines a second range after correcting the first range based on the index value calculated by the candidate implementation unit 120 (step SC03). For example, a set A of values of the first control parameter is set for the industrial machine 4 that is the actual control object and action verification is performed. As a result, the machining error as the index value is 12 [μm]. In addition, a set B of values of the second control parameter is set for the industrial machine 4 that is the actual control object and action verification is performed. As a result, the machining error as the index value is 14 [μm]. This does not satisfy the constraint condition in either the set of values of the first control parameter or the set of values of the second control parameter. In such a case, the search range determination unit 130 calculates the difference between the index values of the processing results of the simulation process and the results of the action verification in the industrial machine 4 that is the actual control object for the set A of values of the first control parameter. Here, when the set A of values of the first control parameter is set for the simulation process, the machining error is 9 [μm]. In addition, as a result of setting and performing action verification on the industrial machine 4 that is the actual control object, the machining error as the index value is 12 [μm], and the difference is 3 [μm]. It is presumed that this value is the modeling error generated when modeling the industrial machine that is the actual control object. Therefore, candidates for the set of values of the control parameter that takes into account the constraint condition and the modeling error are reselected. In this case, the target error set as the constraint condition is 10 [μm], and the modeling error is 3 [μm]. Therefore, if it is a set of values of the control parameter that converges with a machining error of 10 [μm] - 3 [μm] = 7 [μm] or less in the result of the simulation process, even when set and operated on the industrial machine 4 that is the actual control object, the possibility of satisfying the constraint condition is high. Therefore, from the sets of values of the control parameter in which the machining error converges to 7 [μm] or less in the result of the simulation process, the set of values of the control parameter with the fastest cycle time is selected as the set C of the third control parameter. After that, when the set C of the third control parameter is selected, the candidate implementation unit 120 performs action verification on the set C of the third control parameter in the industrial machine 4 that is the actual control object. Then, when the index value calculated based on the verification result satisfies the constraint condition, the range between the set C of the third control parameter and the set A of the first control parameter may be defined as the second range. On the other hand, if the constraint condition is not satisfied, a set of values of the control parameter with a smaller machining error is reselected as the set C of the third control parameter, and the same process may be repeated. According to the above processing flow, for example, as the set C of the third control parameter, the linear acceleration = 2000 [mm / sec 2, the set of values of the control parameters with an angular velocity difference = 900 [mm / sec] (at this time, the machining error is 7 [μm] and the cycle time is 36.0 [sec]) is selected as the set C of values of the third control parameter. At this time, the search range determination unit 130 determines the range of the linear acceleration in the second range to be 2000 to 2300 [mm / sec 2 , and determines the range of the angular velocity difference to be 900 to 1100 [mm / sec].

[0053] In addition, the case where both the group of values of the first control parameter and the group of values of the second control parameter satisfy the constraint conditions is also mentioned. For example, as a result of setting the group A of values of the first control parameter to the industrial machine 4 that is the actual control object and performing action verification, the machining error as the index value is 7 [μm]. In addition, as a result of setting the group B of values of the second control parameter to the industrial machine 4 that is the actual control object and performing action verification, the machining error as the index value is 9 [μm]. This satisfies the constraint conditions in any of the group of values of the first control parameter and the group of values of the second control parameter. In such a case, the search range determination unit 130 calculates the difference between the index values of the processing result of the simulation process and the result of the action verification in the industrial machine 4 that is the actual control object for the group B of values of the second control parameter. Here, when the group B of values of the second control parameter is set for the simulation process, the machining error is 12 [μm]. In addition, as a result of setting and performing action verification on the industrial machine 4 that is the actual control object, the machining error as the index value is 9 [μm], and the difference is 3 [μm]. It is presumed that this value is the modeling error generated when modeling the industrial machine that is the actual control object. Therefore, reselect the candidate of the group of values of the control parameter considering the constraint conditions and the modeling error. In this case, the target error set as the constraint condition is 10 [μm], and the modeling error is 3 [μm]. Therefore, if it is a group of values of the control parameter that generates a machining error of 10 [μm] + 3 [μm] = 13 [μm] or more in the result of the simulation process, even when set and operated on the industrial machine 4 that is the actual control object, the possibility of not satisfying the constraint conditions is high. Therefore, among the groups of values of the control parameter with a machining error of 13 [μm] or more in the result of the simulation process, select the group of values of the control parameter with the minimum cycle time as the group C of values of the third control parameter. After that, when the group C of values of the third control parameter is selected, the candidate execution unit 120 performs action verification on the group C of values of the third control parameter in the industrial machine 4 that is the actual control object. Moreover, in the case where the index value calculated based on the verification result does not satisfy the constraint conditions, it is sufficient to define the range between the group C of values of the third control parameter and the group B of values of the second control parameter as the second range. On the other hand, in the case where the constraint conditions are satisfied, reselect the group of values of the control parameter with a larger machining error as the group C of values of the third control parameter, and repeat the same process.

[0054] Figure 6 It is a flowchart showing a schematic process of the second search process.

[0055] If the above processing is completed and the second range is determined, the second search unit 140 selects an initial value from the second range for each control parameter and sets it for the industrial machine 4 that is the actual control object (step SD01). Next, the second search unit 140 sets the evaluation program 220 to be executed for the industrial machine 4 that is the actual control object (step SD02). Then, the second search unit 140 instructs the industrial machine 4 that is the actual control object to operate with the set value of the control parameter.

[0056] In the industrial machine 4 that is the actual control object, an operation based on the instruction information included in the evaluation program 220 is performed (step SD03). When the operation based on the instruction information in the industrial machine 4 that is the actual control object is completed, the second search unit 140 obtains the operation result from the industrial machine 4 that is the actual control object (step SD04). The obtained operation result is the machining path and cycle time used for the calculation of the index value. The second search unit 140 calculates the index value based on the obtained operation result (step SD05). The machining error as the index value can be calculated as the maximum value of the difference between the path indicated by the evaluation program 220 and the actual machining path. The cycle time directly uses the obtained cycle time as the index value. If the search of the second range is not completed at this time point (in step SD06, "No"), the second search unit 140 sets the industrial machine 4 that is the actual control object after adjusting the value of the control parameter (step SD07). Then, the process proceeds to step SD03. On the other hand, if the search of the second range is completed at this time point (in step SD06, "Yes"), the control parameter that is evaluated as the highest (the cycle time as the index value is the fastest) during the period when the constraint condition is satisfied (the machining error as the index value is below the target error) is selected from the searched control parameters (step SD08). As described above, the processing of steps SD03 to SD07 is executed using known search algorithms such as machine learning-based search, rule-based search, a prescribed full search algorithm, and Bayesian optimization.

[0057] Here, as a result of the second search process, assume that the second search unit 140 selects a set of values of control parameters that satisfy the constraint condition (linear acceleration = 2250 [mm / sec 2 , angular velocity difference = 1000 [mm / sec], at this time, the machining error is 10 [μm], and the cycle time is 35.9 [sec]) as the control parameter. The output unit 150 displays the set of values of the control parameters selected by the second search unit 140 on the display device 70, and sends and sets them to the industrial machine 4 that is the actual control object (step SD09).

[0058] The parameter adjustment device 1 of the present embodiment having the above structure limits the range of the control parameter to near the boundary of the constraint condition through analog processing capable of high-speed processing. The limited range of the control parameter is narrow enough, so that, for example, adjustment of the control parameter of the industrial machine 4 which is an actual control object can be performed. In the industrial machine 4, adjustment of the control parameter can be performed with high precision, but since it takes time to obtain the action result, it is difficult to adjust the control parameter in a wide range. However, through analog processing capable of performing high-speed processing in advance, the adjustment range can be limited to a certain extent accurate range, so that it is expected to shorten the adjustment time of using the industrial machine 4.

[0059] As described above, the embodiments of the present disclosure have been described in detail, but the present disclosure is not limited to the above-described respective embodiments. These embodiments can be subjected to various additions, replacements, changes, partial deletions, etc. within the scope not departing from the gist of the invention, or within the scope not departing from the idea and gist of the present disclosure derived from the content described in the scope of patent claim and its equivalents. For example, in the above embodiments, the order of each action and the order of each process are shown as an example, and are not limited thereto. In addition, the same applies to the case where numerical values or mathematical expressions are used in the description of the above embodiments.

[0060] Hereinafter, an appendix of the embodiment of the present disclosure is shown.

[0061] (Appendix 1)

[0062] A parameter adjustment device (1) according to one aspect of the present disclosure includes: a control object model that models the actions of each part included in an industrial machine (4) that is a control object; a simulation unit (100) that simulates the action of the control object in which a value of a predetermined control parameter is set based on the control object model; a first search unit (110) that repeatedly performs an operation of performing simulation processing using the simulation unit (100) while changing the value of the control parameter within a predetermined first range, and searches for a candidate for a set of values of the control parameter that obtains the first index value near the boundary of a predetermined constraint condition; a candidate implementation unit (120) that sets the set of values of the control parameter selected as a candidate by the first search unit (110) to the control object and performs action verification, and calculates a second index value based on the result of the action verification; and a search range determination unit (130) that determines a second range which is a range of values narrower than the first range based on the first index value and the second index value.

[0063] (Appendix 2)

[0064] Another parameter adjustment device (1) of the present disclosure further includes a second search unit (140), which searches for the control parameter by repeatedly performing an operation of causing the controlled object to operate while changing the value of the control parameter within the second range.

[0065] (Supplementary Note 3)

[0066] In another parameter adjustment device (1) of the present disclosure, the first index value and the second index value are any one of machining accuracy, machining surface quality, and machining time. At least one of the first search unit (110) and the second search unit (140) sets the constraint condition related to machining accuracy or machining surface quality, and searches for the control parameter by means of machine learning so that the machining time is the fastest under this constraint condition.

[0067] (Supplementary Note 4)

[0068] A computer-readable recording medium of one mode of the present disclosure records a program that causes a computer to operate as the following components: a simulation unit (100) that simulates the operation of the controlled object with a predetermined value of the control parameter set based on at least a controlled object model that models the operations of the respective parts of an industrial machine (4) that is the controlled object; a first search unit (110) that repeatedly performs an operation of changing the value of the control parameter within a predetermined first range and performing a simulation process using the simulation unit (100), and searches for a candidate for a set of values of the control parameter that obtains the first index value near the boundary of a predetermined constraint condition based on the first index value calculated from the result of this simulation process; a candidate implementation unit (120) that sets the set of values of the control parameter that becomes a candidate searched by the first search unit (110) to the controlled object and performs action verification, and calculates a second index value based on the result of this action verification; and a search range determination unit (130) that determines a second range that is a range of values narrower than the first range based on the first index value and the second index value.

[0069] Symbol Explanation

[0070] 1 Parameter adjustment device

[0071] 4 Industrial machine

[0072] 5 Network

[0073] 6 Fog computer

[0074] 7 Cloud server

[0075] 8 Sensor

[0076] 11 CPU

[0077] 12 ROM

[0078] 13 RAM

[0079] 14 Non-volatile memory

[0080] 15, 17, 18, 20 Interface

[0081] 22 Bus

[0082] 70 Display device

[0083] 71 Input device

[0084] 72 External device

[0085] 100 Analog section

[0086] 110 First search section

[0087] 120 Candidate implementation section

[0088] 130 Search range determination section

[0089] 140 Second search section

[0090] 150 Output section

[0091] 210 Model storage section

[0092] 220 Evaluation program

Claims

1. A parameter adjustment device, characterized in that: The parameter adjustment device includes: A control object model that models the actions of each part of an industrial machine that is the control object; A simulation unit that, based on the control object model, simulates the actions of the control object with a predetermined value of a control parameter set; A first search unit that repeatedly performs an operation of varying the value of the control parameter within a predetermined first range while performing simulation processing using the simulation unit, and searches for candidates for a set of values of the control parameter that obtains the first index value near the boundary of a predetermined constraint condition based on the first index value calculated from the result of this simulation processing; A candidate implementation unit that sets the set of values of the control parameter that becomes a candidate searched by the first search unit to the control object and performs action verification, and calculates a second index value based on the result of this action verification; And A search range determination unit that determines a range of values, i.e., a second range, that is narrower than the first range based on the first index value and the second index value.

2. The parameter adjustment device according to claim 1, characterized in that: The parameter adjustment device further includes: a second search unit that searches for the control parameter by repeatedly performing an operation of varying the value of the control parameter within the second range while operating the control object.

3. The parameter adjustment device according to claim 2, characterized in that: The first index value and the second index value are any one of machining accuracy, machining surface quality, and machining time, At least one of the first search unit and the second search unit sets the constraint condition related to machining accuracy or machining surface quality, and searches for the control parameter by a machine learning method so that the machining time is the fastest under this constraint condition.

4. A computer-readable recording medium, characterized in that: The recording medium records a program that causes a computer to act as the following components: A simulation unit that, based on a control object model that models at least the actions of each part of an industrial machine that is the control object, simulates the actions of the control object with a predetermined value of a control parameter set; A first search unit that repeatedly performs an operation of varying the value of the control parameter within a predetermined first range while performing simulation processing using the simulation unit, and searches for candidates for a set of values of the control parameter that obtains the first index value near the boundary of a predetermined constraint condition based on the first index value calculated from the result of this simulation processing; A candidate implementation unit that sets the set of values of the control parameter that becomes a candidate searched by the first search unit to the control object and performs action verification, and calculates a second index value based on the result of this action verification; And A search range determination unit that determines a range of values, i.e., a second range, that is narrower than the first range based on the first index value and the second index value.

Images