Processing system, processing method, and program

Abstract

A processing system is provided with: an input unit (11) into which setting data is input; a processing device (10) that is provided in the factory and processes the object on the basis of the setting data; a model information acquisition unit (13) that acquires model information indicating a first model (41) that is applied to the setting data so as to suppress errors included in the processing results of the processing device (10) due to characteristics unique to the processing device (10) before setting in the factory; and a learning unit (16) that learns a second model (42) for suppressing an error between a processing result of a processing device that processes the object (21) in the plant using the first model (41) on the basis of setting data input in the plant and the target value. The processing device (10) processes the object (22) on the basis of the results of applying the first model (41) and the second model (42) to the setting data.

Description

Technical Field

[0001] The present disclosure relates to a processing system, a processing method, and a program. Background Art

[0002] Factory automation (FA) sites use a variety of devices to build systems that implement production-line-like processing steps. These systems typically employ the required number of devices, each with the capabilities appropriate for the desired processing steps. However, even devices of the same model do not necessarily exhibit exactly the same capabilities when deployed on-site, and device capabilities can sometimes vary.

[0003] Therefore, it is conceivable to utilize a technique that learns a model for estimating output under changing conditions, thereby estimating device performance differences and causing the device to perform processing that takes into account these differences (for example, see Patent Document 1). When utilizing this technique, factors that may affect device performance can simply be specified as conditions. Furthermore, if the performance differences can be estimated using a model, a model can also be developed for obtaining outputs that suppress these differences.

[0004] Prior art literature

[0005] Patent Literature

[0006] Patent Document 1: Japanese Patent Application Laid-Open No. 2021-170163 Summary of the Invention

[0007] Problems to be solved by the invention

[0008] However, there are many factors that may affect the performance of the device in the field, and model learning may generate a large computational load.

[0009] The present disclosure has been made under the circumstances described above, and an object of the present disclosure is to reduce the computational load for learning a model for suppressing device performance differences.

[0010] Means for solving problems

[0011] In order to achieve the above-mentioned purpose, the processing system disclosed in the present invention comprises: an input unit, to which setting data is input; a processing device, which is installed in a factory and processes an object according to the setting data; a model information acquisition unit, which acquires model information representing a first model, and the first model is applied to the setting data to suppress a first error contained in a first processing result of the processing device due to inherent characteristics of the processing device before being installed in the factory; an environmental information acquisition unit, which acquires factory environmental information representing the environment of the processing device installed in the factory; an error information acquisition unit, which acquires first error information, the first error information representing a second error between a second processing result of the processing device that processes the object in the factory using the first model according to the setting data input in the factory and a target value; and a learning unit, which learns a second model for suppressing the second error based on the factory environmental information and the first error information, and the processing device processes the object based on the results of applying the first model and the second model to the setting data.

[0012] Effects of the Invention

[0013] According to the present disclosure, a model information acquisition unit acquires model information representing a first model, which is applied to setting data to suppress a first error contained in a first processing result of a processing device due to inherent characteristics of the processing device before being installed in a factory, and a learning unit learns a second model for suppressing a second error between a second processing result of a processing device that processes an object in the factory and a target value. Thus, separately from the first model for suppressing capability differences generated before being installed in the factory, the second model for suppressing capability differences generated due to the environment of the processing device installed in the factory is learned. Therefore, factors that may affect the capability of the device are distinguished, and the second model for addressing capability differences caused by these factors is learned. Therefore, the computational load of learning the model for suppressing capability differences of the device can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a diagram showing an overview of the processing system according to the first embodiment.

[0015] Figure 2 This is a diagram showing the hardware configuration of the processing device according to the first embodiment.

[0016] Figure 3 This is a diagram showing the functional configuration of the processing device according to the first embodiment.

[0017] Figure 4 This is a diagram showing an example of model information of the first model in the first embodiment.

[0018] Figure 5 This is a diagram showing an example of the second model in the first embodiment.

[0019] Figure 6 This is a flowchart showing the model application process in the first embodiment.

[0020] Figure 7 It is a diagram showing the functional configuration of a processing device according to a modified example.

[0021] Figure 8 This is a diagram showing an overview of a processing device according to a second embodiment.

[0022] Figure 9 This is a diagram showing the functional configuration of a processing device according to the second embodiment.

[0023] Figure 10 This is a flowchart showing the first model generation process in the second embodiment.

[0024] Figure 11 This is a flowchart showing the model application process in the second embodiment.

[0025] Figure 12 This is a diagram showing the functional configuration of the processing device 10 according to the third embodiment.

[0026] Figure 13 This is a diagram showing the structure of a recipe library in the third embodiment.

[0027] Figure 14 This is a diagram showing the functional structure of a processing device according to a fourth embodiment.

[0028] Figure 15 This is a diagram for explaining the estimation model of the fourth embodiment.

[0029] Figure 16 This is a diagram showing an example of an estimation model according to the fourth embodiment.

[0030] Figure 17 This is a flowchart showing the estimation model application process according to the fourth embodiment.

[0031] Figure 18 This is a flowchart showing the estimation model generation process according to the fourth embodiment.

[0032] Figure 19 This is a flowchart showing the model improvement process of the fifth embodiment.

[0033] Figure 20 This is a diagram for explaining the search for a combination of the second model and the estimated model in the fifth embodiment.

[0034] Figure 21 It is a diagram showing the configuration of a processing system according to a modification.

[0035] Figure 22It is a diagram showing the configuration of a processing device according to a modified example.

[0036] Figure 23 It is a diagram showing a first model of a modified example. DETAILED DESCRIPTION

[0037] Hereinafter, a processing system according to an embodiment of the present disclosure will be described in detail with reference to the accompanying drawings.

[0038] Implementation method 1.

[0039] like Figure 1 As shown, the processing system 1000 of this embodiment is constructed by installing processing device 10b, one of processing devices 10a and 10b of the same model produced at production site 100a, in factory 100. Processing system 1000 includes processing device 10b. Hereinafter, when there is no distinction between processing devices 10a and 10b, they may be referred to as processing device 10.

[0040] The processing device 10 is a factory automation device or equipment, such as a machine tool, that processes objects 20 and 21. Object 20 is the object to be processed at the production site 100a, and object 21 is the object to be processed at the factory 100. The processing performed by the processing device 10 includes, for example, cutting or grinding of the objects 20 and 21 as workpieces, assembling the objects 20 and 21 as products, or assembling the objects 20 and 21 as parts of a product into other components. The processing device 10 processes the objects 20 and 21 based on setting data input by the user. The setting data is parameters set in the processing device 10 so that the processing device 10 processes the objects 21 and 21, and includes, for example, the position and speed of the worktable supporting the workpiece, the tool, and the tool's rotational speed.

[0041] However, the processing device 10 has mechanical errors during the production stage of the production site 100a. That is, the processing devices 10a and 10b each have inherent characteristics, and due to these characteristics, errors may occur in the processing results of the object 20 at the production site 100a and the target value. Figure 1 , processing device 10a has characteristic A, and processing device 10b has characteristic B. The processing result is, for example, the result of cutting object 20, and the error in the processing result is the dimensional error of cut object 20. The target value can be a specification value directly indicated by the setting data, or it can be a value desired by the user as a value corresponding to the setting data. The target value may or may not be included in the setting data. However, even if the target value is not included in the setting data, it generally corresponds to the setting data and can be separately input to the processing device 10 by the user.

[0042] Since the processing device 10 may produce errors in the processing results due to its inherent characteristics, the first model for suppressing the errors is learned at the production site 100a, and the learned first model is assembled to the processing device 10, thereby making the capabilities of the processing devices 10 unloaded from the production site 100a uniform.

[0043] Then, when the processing device 10 is set in the factory 100, errors may occur in the processing results of the object 21 due to the setting environment of the processing device 10 in the factory 100. The setting environment is, for example, the temperature, humidity, or the type of material inserted into the processing device 10. Therefore, the processing device 10 learns the second model for suppressing errors caused by the setting environment, and uses the learned second model to process the object 21, thereby performing processing suitable for the setting environment. The processing devices 10a and 10b respectively reduce the deviations generated in the processing results of the object 21 even in different environments by learning and using the second model, and exert the ability to be homogenized. In addition, in Figure 1 10b represents the learning related to the processing device 10b. The processing device 10a also performs learning in the same manner as the processing device 10b.

[0044] The processing device 10 is composed of hardware elements for functioning as a computer. Figure 2 As shown, the processing device 10 includes a processor 101, a main storage unit 102, an auxiliary storage unit 103, an input unit 104, an output unit 105, and a communication unit 106. The main storage unit 102, the auxiliary storage unit 103, the input unit 104, the output unit 105, and the communication unit 106 are connected to the processor 101 via an internal bus 107.

[0045] The processor 101 includes a CPU (Central Processing Unit) as a processing circuit. The processor 101 realizes various functions by executing a program P1 stored in the auxiliary storage unit 103 and executes the processing described below.

[0046] The main storage unit 102 includes a RAM. The program P1 is loaded from the auxiliary storage unit 103 into the main storage unit 102. The main storage unit 102 is used as a work area for the processor 101.

[0047] The auxiliary storage unit 103 includes nonvolatile memory, typically EEPROM (Electrically Erasable Programmable Read-Only Memory) and an HDD (Hard Disk Drive). In addition to the program P1, the auxiliary storage unit 103 also stores various data used by the processor 101. Following instructions from the processor 101, the auxiliary storage unit 103 provides the processor 101 with data used by the processor 101. Furthermore, the auxiliary storage unit 103 stores data provided by the processor 101.

[0048] The input unit 104 includes input devices such as hardware switches, input keys, a keyboard, and a pointing device. The input unit 104 acquires information input by a user of the processing device 10 and notifies the processor 101 of the acquired information.

[0049] The output unit 105 includes output devices such as LEDs (Light Emitting Diodes), LCDs (Liquid Crystal Displays), and speakers, and presents various information to the user according to instructions from the processor 101 .

[0050] The communication unit 106 includes a communication interface circuit for communicating with an external device. The communication unit 106 receives a signal from the outside and outputs the data represented by the signal to the processor 101. In addition, the communication unit 106 transmits a signal representing the data output from the processor 101 to the external device.

[0051] Through the cooperation of the above hardware structures, the processing device 10 performs various functions in the factory 100. In detail, Figure 3 As shown, as its functions, the processing device 10 has: an input unit 11 to which setting data is input; a processing unit 12 that performs processing based on the setting data; a model information acquisition unit 13 that acquires model information representing the first model 41 after learning at the production site 100a; an environmental information acquisition unit 14 that acquires environmental information representing the installation environment; an error information acquisition unit 15 that acquires error information representing the error when the object 21 is processed using the first model 41; and a learning unit 16 that learns the second model 42 for suppressing errors based on the environmental information and the error information. Figure 3 In FIG, the solid arrows represent the flow of information before learning the second model, and the dotted arrows represent the flow of information after learning the second model.

[0052] The input unit 11 is primarily implemented by the input unit 104 or the communication unit 106. The input unit 11 receives setting data input by the user. Of the objects 21 and 22 processed based on the setting data, object 21 is the object of processing before learning the second model 42, and object 22 is the object of processing after learning the second model 42. The input unit 11 is an example of an input unit into which the setting data is input.

[0053] The processing unit 12 is primarily implemented by a processing module for processing the processor 101 and the objects 21 and 22. The processing module includes, for example, a motor for moving the tool and the workbench. Before learning the second model 42, the processing unit 12 applies the first model 41 to the setting data input to the input unit 11 to process the object 21. When the learning unit 16 learns the second model 42 by processing the object 21, the processing unit 12 sequentially applies the first model 41 and the second model 42 to the setting data to process the object 22. Furthermore, the processing unit 12 provides the output of the first model 41 to the learning unit 16 to enable learning of the second model 42.

[0054] The model information acquisition unit 13 is implemented by at least one of the processor 101, the input unit 104, and the communication unit 106. The model information acquisition unit 13 can read the model information registered in the auxiliary storage unit 103 of the processing device 10 removed from the production site 100a, or can read the model information from a recording medium such as a memory card attached to the processing device 10 when it was removed. Furthermore, the model information acquisition unit 13 can acquire model information directly input by a user or receive model information via a communication line or network.

[0055] exist Figure 4 A simple example of the model information of the first model 41 is shown in FIG. Figure 4 The model information of represents the following: when the value of the setting data is greater than or equal to zero and less than 10, the model output obtained by applying the first model 41 to the setting data is the sum of the setting data value plus 1, and when the setting data value is greater than or equal to 10 and less than 20, the sum of the setting data value plus 2 is used as the model output. Figure 4 The conversion table shown may also be a model represented as a function using a mathematical formula. The model information acquisition unit 13 is an example of a model information acquisition unit that acquires model information representing a first model. This first model is applied to the setting data to suppress a first error contained in the first processing result of the processing device due to inherent characteristics of the processing device before installation in the factory. Here, the first processing result and first error distinguish the processing result and error at the production site from the second processing result and second error in the factory 100, described later.

[0056] return Figure 3 The environmental information acquisition unit 14 is implemented by at least one of the processor 101, the input unit 104, and the communication unit 106. The environmental information acquisition unit 14 acquires environmental information representing the installation environment of the processing device 10 when the processing unit 12 processes the object 21 using the first model 41 instead of the second model 42. The environmental information acquisition unit 14 can read environmental information from the main storage unit 102, the auxiliary storage unit 103, or a recording medium, or can acquire environmental information directly input by a user, or can receive environmental information from a sensor measuring environmental conditions via a communication line or network. The environment represented by the environmental information can be the temperature, humidity, and type of material as described above, or it can be the power environment provided to the processing device 10 in the factory 100, the quality of air or gas, or the output of other devices connected to the processing device 10. The environmental information acquisition unit 14 is an example of an environmental information acquisition unit that acquires factory environmental information representing the environment of the processing equipment installed in the factory.

[0057] The error information acquisition unit 15 is implemented by at least one of the processor 101, the input unit 104, and the communication unit 106. The error information acquisition unit 15 acquires the error, which is the difference between the measured value and the target value of the processing result of the object 21 processed by the processing unit 12, from the measuring device 30. Alternatively, the error information acquisition unit 15 can acquire error information by separately acquiring the processing result and the target value. That is, the error information can be information indicating both the measured value and the target value of the processing result. The error information acquisition unit 15 can acquire error information through communication with the measuring device 30, read the error information from a recording medium, or acquire error information directly input by a user. The error information acquisition unit 15 is an example of error information acquisition means that acquires first error information indicating a second error between a second processing result and a target value of a processing device that processes the object in a factory using a first model based on setting data input at the factory.

[0058] The learning unit 16 is mainly implemented by the processor 101. The learning unit 16 learns the second model 42 for suppressing the error based on the output of the first model 41, the error resulting from processing the object 21 using the output, and the installation environment when the object 21 is processed using the output. Figure 5 A simple example of the learned second model 42 is shown in FIG. Figure 5The second model 42 represents the following situation: when the output of the first model 41 is greater than 1 and less than 11, when the temperature of the setting environment is less than 15°C, the sum obtained by adding 0.4 to the output of the first model 41 is used as the output of the second model 42, and when the temperature is greater than 15°C, the sum obtained by adding 0.3 to the output of the first model 41 is used as the output of the second model 42. In addition, the following situation is represented: when the output of the first model 41 is greater than 12 and less than 22, when the temperature is less than 20°C, the difference obtained by subtracting 0.2 from the output of the first model 41 is used as the output of the second model 42, and when the temperature is greater than 20°C, the difference obtained by subtracting 0.3 from the output of the second model 42 is used as the output of the second model 42. In addition, the second model 42 is not limited to the following. Figure 5 The conversion table shown can also be a model expressed as a function through mathematical formula. Figure 3 When learning the second model 42, the learning unit 16 provides the second model 42 to the processing unit 12. The learning unit 16 is an example of a learning unit that learns the second model for suppressing the second error based on the environmental information and the error information. The second model 42 is an example of a model for obtaining a value obtained by correcting the output value of the first model 41 based on the environmental information.

[0059] Next, refer to Figure 6 , the model application processing performed by the processing device 10 is explained. Figure 6 The model application process shown is executed as a setup operation or initialization process from the time the processing device 10 is installed in the factory 100 to the start of normal operation.

[0060] In the model application process, the model information acquisition unit 13 acquires model information (step S1). Then, the input unit 11 accepts the input setting data (step S2), the processing unit 12 applies the first model 41 to the setting data and processes the object 21 (step S3), the error information acquisition unit 15 acquires error information (step S4), and the environmental information acquisition unit 14 acquires environmental information (step S5).

[0061] Next, the learning unit 16 executes steps S2 to S5 to determine whether the amount of accumulated data, including the output of the first model 41, error information, and environmental information, which are mutually related, exceeds a threshold (step S6). If the data amount is determined not to exceed the threshold (step S6: No), the processing device 10 repeats the process from step S2 onward. Thus, data representing a combination of the output of the first model 41, error information, and environmental information is accumulated.

[0062] If the data volume is determined to exceed the threshold (step S6: Yes), the learning unit 16 uses the accumulated data to learn the second model 42 (step S7). For example, the learning unit 16 generates a temporary model representing the relationship between the output, error, and installation environment of the first model 41 through regression analysis, and then obtains a transformation formula for the output of the first model 41 that reduces the error as the second model 42. However, the learning unit 16 is not limited to this method for learning the second model 42 and may also use supervised learning, typified by neural networks, or reinforcement learning.

[0063] Next, the input unit 11 receives the newly input setting data (step S8), and the processing unit 12 applies the first model 41 and the second model 42 to the new setting data to process the object 22 (step S9).

[0064] As described above, the model information acquisition unit 13 acquires model information representing the first model 41, which is applied to the setting data to suppress errors in the processing results of the processing device 10 due to the inherent characteristics of the processing device 10 before installation in the factory 100. Furthermore, the learning unit 16 learns the second model 42 for suppressing errors in the processing results of the processing device 10 processing the object 21 in the factory 100. Thus, separately from the first model 41 for suppressing performance differences that occur before installation in the factory 100, the second model 42 for suppressing performance differences that occur due to the environment of the processing device 10 installed in the factory 100 is learned. Thus, factors that may affect device performance are classified, and the second model 42 is learned in the factory 100 to address performance differences caused by these factors. Consequently, the computational load of learning the model for suppressing performance differences in the device can be reduced.

[0065] That is, information indicating characteristics unique to the processing device 10 is excluded from information used by the learning unit 16 to learn the second model 42. This avoids unnecessary computational processing caused by performing the same type of learning at both the production site 100a and the factory 100.

[0066] In addition, the second model 42 for obtaining the output of the processing device 10 based on the output of the first model 41 has been described, but the present invention is not limited to this. Figure 7 As shown, the first model 41 and the second model 42 may be applied in parallel to the setting data, and after learning the second model 42, the sum of the output of the first model 41 and the output of the second model 42 is input to the processing unit 17 that processes the objects 21 and 22. As long as it is used to correct the processing of the object 21 based on the first model 41, the method of applying the second model 42 is arbitrary. Figure 7In the example of , the learning unit 16 may also collect setting data to learn the second model 42.

[0067] Implementation method 2.

[0068] Next, Embodiment 2 will be described with a focus on the differences from Embodiment 1. The same reference numerals are used for the same or equivalent structures as those in Embodiment 1. Figure 8 As shown, this embodiment differs from Embodiment 1 in that processing device 10 has the ability to learn the first model. Furthermore, this embodiment differs from Embodiment 1 in that learning unit 16 further distinguishes the installation environment into a fixed environment that remains fixed within factory 100 and a variable environment that may change over time, and learns the second model.

[0069] like Figure 9 As shown in FIG. 1 , the processing device 10 of this embodiment includes: a characteristic information acquisition unit 18 that indicates the characteristic inherent to the processing device 10; and a model generation unit 19 that generates a first model 41. Figure 9 In FIG. 1 , the flow of information in the production site 100 a is shown by bold arrows.

[0070] The characteristic information acquisition unit 18 is implemented by at least one of the processor 101, the input unit 104, and the communication unit 106. The characteristic information acquisition unit 18 acquires characteristic information representing the characteristics inherent to the processing devices 10 at the production site 100a. The characteristic information acquisition unit 18 can read the characteristic information from the auxiliary storage unit 103 or an external recording medium, acquire characteristic information directly input by a user, or receive characteristic information via a communication line or network. The characteristic information may, for example, represent the results of quality inspections performed on multiple processing devices 10 at the production site 100a. The characteristic information acquisition unit 18 serves as an example of characteristic information acquisition means for acquiring characteristic information representing the characteristics inherent to the processing devices 10.

[0071] The model generation unit 19 is primarily implemented by the processor 101. The model generation unit 19 acquires the setting data input to the input unit 11 and obtains error information from the error information acquisition unit 15, which has acquired error information, indicating the error of the object 20 after processing by the processing unit 12 based on the setting data without using the first model 41. This error information is an example of the second error information indicating the first error described above. Furthermore, the model generation unit 19 acquires environmental information indicating the production environment of the production site of the production processing device 10 from the environmental information acquisition unit 14, which has acquired environmental information, and obtains characteristic information indicating the characteristics of the processing device 10 itself from the characteristic information acquisition unit 18. The model generation unit 19 then generates the first model 41 through learning based on this acquired information and provides it to the model information acquisition unit 13. For example, the model generation unit 19 generates the first model 41 for obtaining setting data that minimizes the error through regression analysis or supervised learning using the setting data as the target variable and the error information, environmental information, and characteristic information as explanatory variables. Furthermore, even when correction values for setting data corresponding to environmental conditions are statistically predetermined, if deviations due to the characteristics of the processing device occur, the model generation unit 19 may use the characteristic information to learn the first model 41 to reduce such deviations. Furthermore, the model generation unit 19 may generate the first model 41 through regression analysis or supervised learning using error information as the target variable and setting data, environmental information, and characteristic information as explanatory variables. The model generation unit 19 serves as an example of a model generation unit that generates the first model based on the characteristic information, environmental information of the production environment, and the second error information.

[0072] Furthermore, the method used by the model generation unit 19 to generate the first model 41 can be arbitrarily modified. For example, one or both of the characteristic information and the environmental information can be omitted from the information used to generate the first model 41. Even in cases where one or both of the characteristic information and the environmental information are omitted, the first model 41 generated by the model generation unit 19 ultimately serves to suppress errors caused by the characteristics of the processing device 10. Furthermore, the model generation unit 19 can generate the first model 41 by applying the characteristic information and the environmental information to a template model provided externally at the production site 100a.

[0073] Next, the first model generation process executed by the processing device 10 at the production site 100 a and the model application process executed at the factory 100 will be described in order.

[0074] In the first model generation process, Figure 10As shown, the characteristic information acquisition unit 18 acquires characteristic information (step S11). Then, the input unit 11 accepts the input setting data (step S12), the processing unit 12 processes the object 20 according to the setting data (step S13), the error information acquisition unit 15 acquires error information (step S14), and the environmental information acquisition unit 14 acquires environmental information of the production environment (step S15). The environmental information of the production environment can be the same type of information as the setting environment, or different information. The environmental information of the production environment can be any information that indicates environmental factors that affect the quality of the processing device 10. The environmental information of the production environment is equivalent to an example of production environment information.

[0075] Next, the model generation unit 19 determines whether the amount of data obtained in steps S12 to S15 exceeds a predetermined threshold (step S16). If the amount of data does not exceed the threshold (step S16: No), the processing device 10 repeats the process from step S12 onward. This accumulates the data required to generate the first model 41.

[0076] If the data volume is determined to exceed the threshold (step S16: Yes), the model generation unit 19 generates a first model 41 through learning based on the information obtained in steps S11 to S15 (step S17). Next, the processing unit 12 applies the generated first model 41 to the setting data obtained in step S12 to process the object 21 (step S18), and the error information acquisition unit 15 acquires error information (step S19). The model generation unit 19 then determines whether the error resulting from processing the object 21 using the first model 41 and the setting data is within a predetermined range (step S20).

[0077] If it is determined that the error is not within the range (step S20: No), the process returns to step S12, and the addition of data based on the re-execution of steps S12 to S16 and the generation of the first model based on steps S17 to S19 are repeated. Thus, the model generation unit 19 continues learning the first model 41. Alternatively, during the repeated learning, the first model 41 may be learned based on newly collected data without using previously collected data. Furthermore, if the iterations of a predetermined process, such as updating the weights of each layer in deep learning, are rounded up during the learning of the first model 41 in step S17, then if the determination in step S20 is negative (step S20: No), the process returns to step S17 without additionally collecting data, and the learning iterations may continue. If it is determined in step S20 that the error is within the range (step S20: Yes), the processing device 10 terminates the first model generation process.

[0078] like Figure 11As shown, in the model application process of this embodiment, after executing step S1 similarly to the first embodiment, the environment information acquisition unit 14 acquires environment information representing a fixed environment (step S21). The fixed environment is an environment that is subsequently fixed by the installation of the processing device 10 or adjustments during installation. Specifically, the fixed environment includes whether A / D conversion of an externally applied voltage is performed or the type of installed components selected based on the space.

[0079] Next, processing device 10 executes steps S2 to S4 similarly to Embodiment 1 and then obtains environmental information indicating the fluctuating environment of the installation environment (step S22). A fluctuating environment is an environment that may change each time objects 21 and 22 are processed, such as temperature or humidity. Learning unit 16 then determines whether the data volume exceeds a threshold (step S6). If the data volume does not exceed the threshold (step S6: No), the process from step S2 onward is repeated.

[0080] On the other hand, if the data volume is determined to exceed the threshold (step S6: Yes), the learning unit 16 learns the second model 42 based on the fixed environment indicated by the environmental information acquired in step S11 (step S23). Here, the learning unit 16 learns the second model 42 without considering the variable environment indicated by the environmental information acquired in step S22. Specifically, the learning unit 16 learns the second model 42 without including the variable environment as a parameter, or learns the second model 42 with the parameters of the variable environment fixed, regardless of the information acquired in step S22 (step S23).

[0081] Next, the learning unit 16 determines whether the error when the second model 42 learned in step S23 is applied to the setting data is within a predetermined first range (step S24). The learning unit 16 can obtain this error by applying the second model 42 to newly input setting data, or by cross-validating the data accumulated in steps S2 to S4 and S22 repeatedly.

[0082] If the error is determined to be outside the first range (step S24: No), the process returns to step S2, and the data from steps S2 to S4, S22, and S6 are re-executed, and the learning of the second model 42 in step S23 is repeated. Furthermore, if the predetermined number of iterations of the process is rounded up during the learning of the second model 42 in step S23, and the determination in step S24 is negative (step S24: No), the process may return to step S23 without additional data collection, and the learning iterations may continue. If the error is determined to be within the first range in step S24 (step S24: Yes), the learning unit 16 continues learning the second model 42 based on the changing environment indicated by the environmental information acquired in step S22 (step S25).

[0083] Next, the learning unit 16 determines whether the error when applying the second model 42 learned in step S25 to the set data is within a predetermined second range (step S26). The second range is defined as a wider range than the first range. Because the fixed environment does not change, the first range is defined as a relatively narrow range to obtain a more appropriate second model 42 for the fixed environment in steps S23 and S24. In contrast, the second range is defined as a relatively wide range to account for changes in the variable environment.

[0084] If it is determined that the error is not within the second range (step S26: No), the process returns to step S2, and the data addition based on the re-execution of steps S2 to S4, S22, and S6 and the learning of the second model 42 in steps S23 and S25 are re-executed. Furthermore, if the predetermined number of iterations of the process is rounded up during the learning of the second model 42 in step S25, and if the determination in step S26 is negative (step S26: No), the process returns to step S25 without additional data collection, and the learning iterations may be continued.

[0085] If it is determined in step S26 that the error is within the second range (step S26: Yes), the learning unit 16 determines whether the error has decreased (step S27). Specifically, the learning unit 16 determines whether the magnitude of the error determined to be within the second range in step S26 has decreased from the magnitude of the error determined to be within the first range in step S24.

[0086] If the error is determined not to have been reduced (step S27: No), the process returns to step S2, and the data addition based on the re-execution of steps S2 to S4, S22, and S6, and the learning of the second model 42 in steps S23 and S25 are re-executed. Furthermore, if the predetermined number of iterations of the process is rounded up during the learning of the second model 42 in steps S23 and S25, and the determination in step S27 is negative (step S27: No), the process may return to step S23 without collecting additional data, and the learning iterations may continue. If the error is determined to have been reduced in step S27 (step S27: Yes), the model application process ends.

[0087] As described above, the learning unit 16 performs learning for a changing environment after performing learning for a fixed environment. Therefore, factors that may affect the performance of the processing device 10 are further classified, and models are sequentially learned to absorb performance differences caused by these factors. This further reduces the computational load associated with model learning.

[0088] Implementation method 3.

[0089] Next, Embodiment 3 will be described focusing on the differences from Embodiment 1. In addition, the same reference numerals are used for the same or equivalent structures as those in Embodiment 1. Figure 12 As shown, this embodiment differs from the first embodiment in that information used in learning is stored in advance and used when processing a new object.

[0090] The processing device 10 of this embodiment includes: a storage unit 110, which stores information used by the learning unit 16; a providing unit 111, which provides information from the storage unit 110; and a determining unit 112, which determines determination data corresponding to setting data to be newly input based on the information from the storage unit 110.

[0091] The storage unit 110 is mainly realized by the auxiliary storage unit 103. The storage unit 110 repeatedly obtains the setting data input to the input unit 11 and stores it in association with the information used by the learning unit 16. Figure 13 Recipe library shown.

[0092] The recipe library is a database that stores records that associate setting data, target values when the setting data is input, environmental information indicating the environment in which objects 21 and 22 are processed based on the setting data, the first and second models applied to the setting data, and error information. The target values in the recipe library can be included as part of the error information, or can be input to the input unit 11 as part of the setting data or as data separate from the setting data. The storage unit 110 is an example of an accumulation unit that associates multiple setting data with target values, as well as a third error between the target value and the third processing result of the processing device that processes the object by applying the first and second models to the setting data.

[0093] The providing unit 111 is implemented by at least one of the processor 101 and the output unit 105. The providing unit 111 can also read the recipe library from the storage unit 110 according to the user's request and provide the recipe library to the user. The user can also refer to the provided recipe library to determine the setting data corresponding to the user's desired target value. In addition, when the setting data corresponding to the user's desired target value is associated with error information indicating a large error, the user can input the setting data into the input unit 11 after slightly changing the setting data. In addition, when the record containing the user's desired target value is not included in the recipe library, the user can refer to multiple records to infer appropriate setting data. In addition, the providing unit 111 can also provide information of the recipe library to the determining unit 112.

[0094] The determination unit 112 is primarily implemented by the processor 101. When a user specifies a target value via the input unit 11, the determination unit 112 determines, based on the recipe library, the determination data corresponding to the setting data corresponding to the specified target value. Specifically, the determination unit 112 may also determine, as the determination data, the setting data associated with the target value specified by the user in the recipe library. Furthermore, when the setting data associated with the target value specified by the user in the recipe library is also associated with error information indicating a large error, or when a record containing the target value specified by the user is not included in the recipe library, the determination unit 112 may also determine, using an estimation method such as linear interpolation of multiple records, the determination data corresponding to the setting data that should be input corresponding to the target value. The determination unit 112 is an example of a determination unit that determines the determination data corresponding to the setting data corresponding to the specified target value based on multiple setting data and a third processing result of a processing device that processes an object by applying the first model and the second model to each setting data.

[0095] As described above, using the recipe library facilitates the determination of new setting data. The input unit 11 is an example of input means for accepting input of a new target value when processing a new object, and the determination unit 112 is an example of determination means for determining specific data corresponding to the setting data corresponding to the new target value. The processing device processes the new object based on the results of applying the first and second models to the specific data determined corresponding to the new target value.

[0096] Implementation method 4.

[0097] Next, the fourth embodiment will be described with the focus on the differences from the third embodiment. In addition, the same reference numerals are used for the same or equivalent structures as those in the third embodiment. The difference between this embodiment and the third embodiment is that Figure 14 As shown, the determination unit 112 generates an estimation model for estimating setting data.

[0098] The determination unit 112 of this embodiment includes: an estimation model generation unit 1121 that generates an estimation model 43 for estimating appropriate setting data based on the target value; and a target value acquisition unit 1122 that acquires the target value input to the input unit 11. Figure 14 In FIG, the thicker dotted line shows the flow of information when the estimation model is used.

[0099] The estimation model generation unit 1121 reads Figure 13 The formula library shown in FIG4 is used to generate the estimation model 43 based on the formula library. Figure 15 As shown, the estimation model 43 is a model that determines, based on known target values contained in the recipe library, setting data for achieving the target value with a small error as determination data. Furthermore, for unknown target values, determination data for achieving the target value with a small error is also determined. Here, a small error refers to an error that is smaller than a predetermined threshold.

[0100] For example, even if the length of the machining time in cutting is doubled, the amount of cutting depends on the shape of the blade and is therefore not necessarily doubled. In this way, when the relationship between the set data and the target value is not obvious to the user, an estimation model is used. Figure 16 A simple example of an estimation model is shown in FIG. The estimation model generation unit 1121 generates an estimation model for obtaining, as determination data, setting data that minimizes the error from the target value, by, for example, performing regression analysis or supervised learning with the setting data as the target variable and the target value, environmental information, the first model 41, the second model, and error information as explanatory variables.

[0101] The determination unit 112 applies the estimation model 43 to the target value obtained by the target value obtaining unit 1122, determines the determination data as the setting data for achieving the target value, and inputs the determination data to the input unit 11. If the user inputs the target value without being aware of the selection of the setting data, the object 23 is processed so as to achieve the target value.

[0102] The estimated model corresponds to an example of a model for estimating determination data corresponding to setting data corresponding to a specified target value, which is used to suppress a third error between a third processing result of a processing device that processes an object using the first model and the second model and a target value. The determination unit 112 corresponds to an example of determination means that generates the estimated model based on the accumulated data, applies the estimated model to the new target value, and thereby determines determination data corresponding to the setting data corresponding to the new target value.

[0103] Next, refer to Figures 17-18 , the estimation model application process performed by the processing device 10 will be described. Figure 6 That is, the estimated model application process is executed after the second model 42 is learned.

[0104] exist Figure 17 In the estimation model application process shown in FIG. 1 , the determination unit 112 performs the estimation model generation process (step S41). In the estimation model generation process, as shown in FIG. Figure 18 As shown, the determination unit 112 fixes the first model 41 and the second model 42 (step S411) and reads data from the recipe library of the storage unit 110 (step S412). Here, fixing the first model 41 and the second model 42 in step S411 means that neither the first model 41 nor the second model 42 is newly learned. Therefore, even if different first models 41 or second models 42 are registered in the recipe library depending on conditions such as setting data or environmental information, records representing different first models 41 and second models 42 may be read in step S412.

[0105] Next, the estimation model generation unit 1121 generates the estimation model 43 through learning based on the data read in step S412 (step S413). The estimation model generation unit 1121 then determines whether the error when the estimation model 43 generated in step S413 is applied to the new target value is within a predetermined range (step S414). The estimation model generation unit 1121 can obtain the error in step S414 using the target value newly input by the user, or it can obtain it through cross-validation of the data read in step S412.

[0106] When it is determined that the error is not within the range (step S414: No), the estimation model generation unit 1121 repeats the processing after step S413 and continues to learn the estimation model. Specifically, in the learning of the estimation model 43 of the previous step S413, the estimation model generation unit 1121 starts the repetition of the processing from the time when the repetition of the predetermined processing is rounded up. In addition, when not all data are read from the recipe library in step S412, when the judgment in step S414 is negative (step S414: No), it returns to step S412, and the determination unit 112 may also read new data. When it is determined in step S414 that the error is within the range (step S414: Yes), the processing of the processing device 10 returns from the estimation model generation processing to the Figure 17 The estimated model is applied to process.

[0107] return Figure 17 , following the estimation model generation process of step S41, the target value acquisition unit 1122 acquires the new target value input to the input unit 11 (step S42), and the determination unit 112 applies the estimation model 43 to the new target value, thereby determining the determination data (step S43).

[0108] Next, the processing unit 12 processes the object 23 using the first model 41 and the second model 42 based on the determined data (step S44). A new record of the processing of the object 23 is then registered in the recipe library of the storage unit 110 (step S45). This registration can be performed by the learning unit 16 or by the processing unit 12.

[0109] As described above, the determination unit 112 allows the user to execute processing of the object 23 based on appropriate determination data simply by inputting a target value. Furthermore, by enriching the recipe library of the storage unit 110 with additional information, the determination unit 112 can be expected to determine more appropriate determination data.

[0110] Implementation method 5.

[0111] Next, Embodiment 5 will be described focusing on the differences from Embodiment 4. The same reference numerals are used for the same or equivalent configurations as those in Embodiment 4. In this embodiment, learning of the second model 42 and learning of the estimated model 43 are repeated alternately.

[0112] exist Figure 19 The flow of the model improvement process performed by the processing device 10 of this embodiment is shown in FIG. Figure 6 The model application process shown (step S51) is performed. Thus, the second model 42 is initialized.

[0113] Next, data is stored in the recipe library (step S52 ). Specifically, based on the newly input setting data, the object 22 is processed using the second model 42 a plurality of times, and information on the processed object 22 is stored in the storage unit 110 .

[0114] Then, execute Figure 17 The estimation model 43 is initialized by applying the estimation model shown in FIG.

[0115] Next, data is accumulated in the recipe library (step S54). Specifically, the estimation model 43 is applied to the newly input target value to determine the specific data. Based on the determined specific data, the object 23 is processed multiple times using the second model 42. Then, information on the processing of the object 23 is accumulated in the storage unit 110.

[0116] Next, the learning unit 16 learns the second model 42 based on the data accumulated in step S54 (step S55). This improves the second model 42. The estimated model generation unit 1121 then generates the estimated model 43 (step S56). This improves the estimated model 43. The processing device 10 then repeats the processes from step S54 onward.

[0117] Therefore, if Figure 20 As shown schematically, a combination of the second model 42 and the estimated model 43 that minimizes the error from a specific target value is searched. Figure 20 In the figure, point P2 represents the combination of the initial values of second model 42 and estimated model 43. This point P2 moves as indicated by the arrow along the horizontal axis due to the improvement of second model 42, and further moves as indicated by the arrow along the vertical axis due to the improvement of estimated model 43. By repeating this movement, it is expected that second model 42 and estimated model 43 will approach point P3 corresponding to the combination with the minimum error.

[0118] As mentioned above, although embodiment of this disclosure was described, this disclosure is not limited to the said embodiment.

[0119] For example, the processing system 1000 is described as being the same as the processing device 10, but the present invention is not limited thereto. Figure 21 As shown, the processing system 1000 can also be composed of a terminal 50 having an input unit 11, a processing unit 12, a model information acquisition unit 13, an environmental information acquisition unit 14, an error information acquisition unit 15, and a learning unit 16, and a processing device 10 as a machine tool. Here, the terminal 50 is a UI (User Interface) terminal for operating the processing device 10, for example, an industrial PC (Personal Computer). Figure 21In the example of , the processing device 10 also processes the object based on the results of applying the first model 41 and the second model 42 to the setting data.

[0120] In addition, if Figure 22 As shown, the recipe library may be stored in an external storage device 110 a instead of being stored in the storage unit 110 of the processing device 10 .

[0121] The above embodiments may be combined arbitrarily. For example, Figure 21 The terminal 50 shown includes the characteristic information acquisition unit 18 and the model generation unit 19 of the second embodiment, and the storage unit 110 , the providing unit 111 , and the specifying unit 112 of the third and fourth embodiments.

[0122] Furthermore, the example in which the first model 41 and the second model 42 are used to obtain the correction value of the setting data based on the setting data has been described, but the present invention is not limited to this. Figure 23 As shown, in the case where a simple process is predetermined in which a control output value different from the setting data is output to the processing unit 17 of the processing object according to the setting data, the first model 41 can also output a control output value appropriate for the simple process according to the setting data. Here, the setting data is, for example, the moving speed of the tool in the cutting process, and the control output value is the current value flowing into the motors of the main shaft that rotates the tool and the moving axis that moves the worktable. The second model 42 can obtain the output value of the first model 42 and output a correction value of the output value, or it can obtain the setting data and output a correction value of the output of the first model 41. Figure 23 In the example of , the model generating unit 19 that generates the first model 41 can use the learning data including the history of the control output value. In addition, the learning unit 16 that learns the second model 42 can also use the learning data including the history of the control output value.

[0123] The first model 41 and the second model 42 may have inputs and outputs that are included in a process that starts with the setting data input by the user and ends with the processing of the object based on the setting data. The processing device 10 may process the object based on the results of applying the first model 41 and the second model 42 to the setting data.

[0124] The functions of the processing system 1000 in the above-described embodiment can be realized by dedicated hardware, or can be realized by a general computer system.

[0125] For example, the program P1 can be distributed by storing it in a computer-readable recording medium such as a floppy disk, CD-ROM (Compact Disk Read-Only Memory), DVD (Digital Versatile Disk), or MO (Magneto-Optical disk), and the program P1 can be installed in a computer, thereby forming a device that performs the above-mentioned processing.

[0126] Alternatively, the program P1 may be stored in advance in a disk device of a server device on a communication network such as the Internet, and downloaded to a computer by being superimposed on a carrier wave, for example.

[0127] Furthermore, the above-described processing can also be realized by executing and starting the program P1 while transferring it via a network represented by the Internet.

[0128] Furthermore, the above-described processing can be realized by executing all or part of the program P1 on a server device, and by causing a computer to execute the program P1 while transmitting and receiving information related to the processing via a communication network.

[0129] When the OS (Operating System) shares the above functions or when the OS and applications cooperate to realize the above functions, only the parts other than the OS may be stored in a medium and distributed, or may be downloaded to a computer.

[0130] Furthermore, the means for realizing the functions of the processing system 1000 are not limited to software, and a part or all of the functions may be realized by dedicated hardware or circuits.

[0131] The present disclosure is capable of various embodiments and variations without departing from the broad spirit and scope of the present disclosure. Furthermore, the aforementioned embodiments are intended to illustrate the present disclosure and do not limit its scope. That is, the scope of the present disclosure is defined by the claims, not by the embodiments. Furthermore, variations implemented within the meaning of the claims and their equivalents are considered to be within the scope of the present disclosure.

[0132] Industrial applicability

[0133] The present disclosure is suitable for reducing the variation in performance of devices installed at FA sites.

[0134] Description of labels

[0135] 10, 10a, 10b: processing device; 11: input unit; 12: processing unit; 13: model information acquisition unit; 14: environmental information acquisition unit; 15: error information acquisition unit; 16: learning unit; 17: processing unit; 18: characteristic information acquisition unit; 19: model generation unit; 20-23: object; 30: measuring device; 41: first model; 42: second model; 43: estimated model; 50: terminal; 100: factory; 100a: production site; 101: processor; 102: main storage unit; 103: auxiliary storage unit; 104: input unit; 105: output unit; 106: communication unit; 107: internal bus; 110: storage unit; 110a: storage device; 111: providing unit; 112: determining unit; 1000: processing system; 1121: estimation model generating unit; 1122: target value acquiring unit; P1: program; P2, P3: points.

Claims

1. A processing system, wherein: The processing system has: an input unit into which setting data is input; a processing device installed in the factory and processing the object according to the setting data; a model information acquisition unit that acquires model information indicating a first model applied to the setting data to suppress a first error included in a first processing result of the processing device due to inherent characteristics of the processing device before installation in the factory; an environmental information acquisition unit that acquires factory environmental information indicating an environment of the processing equipment installed in the factory; an error information acquisition unit that acquires first error information indicating a second error between a second processing result of the processing device that processes the object in the factory using the first model based on the setting data input in the factory and a target value; and a learning unit that learns a second model for suppressing the second error based on the factory environment information and the first error information; The processing device processes the object based on a result of applying the first model and the second model to the setting data.

2. The processing system according to claim 1, wherein The information indicating the inherent characteristics is excluded from the information used by the learning unit to learn the second model.

3. The processing system according to claim 1 or 2, wherein: The processing system further includes a determination unit that determines determination data corresponding to the setting data corresponding to the designated target value based on the plurality of setting data and a third processing result of the processing device that processes the object by applying the first model and the second model to each of the setting data. The input unit receives input of a new target value when processing a new object. The determining unit determines the determination data corresponding to the setting data corresponding to the new target value, The processing device processes the new object based on a result of applying the first model and the second model to the determination data determined corresponding to the new target value.

4. The processing system according to claim 3, wherein: The processing system further includes an accumulation unit that associates the plurality of setting data with the target value and a third error between the third processing result and the target value and accumulates the data as accumulation data. The determination unit generates an estimation model for estimating the determination data that suppresses the third error and is equivalent to the setting data corresponding to the specified target value based on the accumulated data, and applies the estimation model to the new target value, thereby determining the determination data equivalent to the setting data corresponding to the new target value.

5. The processing system according to claim 4, wherein: The learning of the second model by the learning unit and the generation of the estimated model by the determination unit are repeated alternately.

6. The processing system according to any one of claims 1 to 5, wherein: The first model is a model for obtaining a value obtained by correcting the setting data based on the setting data.

7. The processing system according to any one of claims 1 to 6, wherein: The second model is a model for obtaining a value obtained by correcting the output value of the first model based on the output value and the plant environment information.

8. The processing system according to any one of claims 1 to 7, wherein: The processing system also features: a characteristic information acquisition unit that acquires characteristic information indicating the inherent characteristic; and a model generating unit, which generates the first model, The environmental information acquisition unit acquires production environmental information indicating the environment of a production site where the processing device is produced. The error information acquisition unit acquires second error information indicating the first error. The model generation unit generates the first model based on the characteristic information, the production environment information, and the second error information.

9. A processing method, wherein: This processing method includes the following processing: The input unit accepts setting data, The model information acquisition unit acquires model information representing a first model, the first model being applied to the setting data to suppress a first error included in a first processing result of a processing device installed in a factory and processing an object according to the setting data due to inherent characteristics before the processing device is installed in the factory. An environmental information acquisition unit acquires environmental information indicating an environment of the processing device installed in the factory. an error information acquisition unit that acquires error information indicating a second error between a second processing result of the processing device that processes the object in the factory using the first model based on the setting data input in the factory and a target value; The learning unit learns a second model for suppressing the second error based on the environmental information and the error information. The processing device processes the object based on a result of applying the first model and the second model to the setting data.

10. A program, wherein This program causes the computer to execute the following processing: Accepting setting data, obtaining model information representing a first model, the first model being applied to the setting data to suppress a first error included in a first processing result of a processing device installed in a factory and processing an object according to the setting data due to inherent characteristics before the processing device is installed in the factory; acquiring environmental information indicating the environment of the processing equipment installed in the factory, obtaining error information indicating a second error between a second processing result of the processing device that processes the object in the factory using the first model based on the setting data input in the factory and a target value; Based on the environmental information and the error information, a second model is learned for suppressing the second error. The object is processed based on a result of applying the first model and the second model to the setting data.

Images