JP2022035516A - Learned model generation device and production system - Google Patents

Abstract

To provide a learned model generation device which generates a learned model for generating a process plan, and a production system using the generated learned model.SOLUTION: A production system 1 includes a learning processing device 21 and a generation processing device 22. A model generation unit 213 of the learning processing device 21 performs machine learning with explanatory variables of facility operation state data FD, work ability data OD, and production plan data OSD which are stored in a learning data set storage unit 212, and objective variables of process plan data JKD (including personnel assignment data and work instruction data) of a process plan (actual result). A process plan generation unit 223 of the generation processing device 22 generates new process plan data SKD (including personnel assignment data and work instruction data) by using the facility operation state data FD, the work ability data OD, new production plan data NSD, and the generated learned model.SELECTED DRAWING: Figure 8

Description

The present invention relates to a trained model generator and a production system.

Generally, in a production system, a process plan is drawn up based on a production plan for producing a product, which defines the equipment used for production and the staffing of workers who operate the equipment. Then, the worker operates the equipment according to the process plan to produce a fixed number of products every day.

In formulating such a process plan, for example, in Patent Document 1, a priority is set and set to alleviate each of a plurality of restrictions by performing machine learning based on information on a production plan drafted in the past. The technology for formulating a production plan that relaxes the restrictions in the order of priority and satisfies all the restrictions is disclosed. Further, for example, Patent Document 2 discloses a technique for calling attention to a worker who works according to a process plan about a work procedure different for each process.

Japanese Unexamined Patent Publication No. 2018-120342 Japanese Unexamined Patent Publication No. 2018-169933

By the way, when formulating a process plan, a manager who is an experienced and skilled worker often formulates a process plan corresponding to the production plan in consideration of the ability of the equipment and the ability of the worker. In this way, when a skilled worker formulates a process plan based on his / her own experience, equipment suitable for the production of the product is selected, and a worker who is excellent in operating the selected equipment is selected. Can be placed. Furthermore, if it is a skilled worker, it is possible to accurately give a work instruction to the worker for each process.

However, in order to formulate a highly productive process plan, some experience is required, and there is a problem that the human resources who can formulate the process plan are limited. It is desired that anyone with or without experience can formulate a highly productive process plan.

An object of the present invention is to provide a trained model generator for generating a trained model for generating a process plan, and a production system using the generated trained model.

(Trained model generator)
The trained model generator is the equipment operation status data showing the equipment operation status of the equipment that produces the product, and the ability of the worker who uses the equipment to produce the product using the equipment to produce the product. Work capacity data representing work capacity data, production plan data representing a production plan for producing a product, and process plan data representing a past process plan used for producing a product are acquired as training data sets. New production is performed by performing machine learning using the equipment operation status data, work capacity data, and production plan data acquired by the set acquisition unit and the learning data set acquisition unit as explanatory variables, and the process plan data and objective variables. A trained model generation unit for generating a trained model for generating a new process plan data representing a new process plan for a new production plan data representing a plan is provided.

(Production system)
The production system is a production system that is equipped with the above-mentioned trained model generator and produces products, and is a management terminal device that outputs new production plan data and a trained model that acquires equipment operation status data of equipment. Equipment operation status acquisition device to be output to the generation device and management terminal device, work ability acquisition device to acquire the work ability data of the worker and output to the trained model generation device and management terminal device, and new process plan data. The generation processing device includes a generation processing device that generates and outputs the equipment, and the generation processing device includes equipment operation status data output from the equipment operation status acquisition device, work capacity data output from the work capacity acquisition device, and a management terminal. By using the new production plan data output from the apparatus and the trained model generated by the trained model generator, new process plan data is generated and the generated new process plan data is output.

According to these, the equipment operation status data representing the operation status of the equipment, the work capacity data representing the worker's ability, the production plan data, and the past process plan, that is, the process plan data representing the proven process plan are obtained. New process planning data can be generated by machine learning as a training data set. As a result, even if an inexperienced worker becomes a manager and formulates a process plan, it is generated by machine learning using a process plan that has a proven track record as an experienced skilled worker. By using the trained model, it is possible to easily and highly productive process planning.

It is a figure which shows the structure of a production system. It is a figure which shows the structure of the equipment possible state acquisition apparatus of FIG. It is a figure for demonstrating data about a machine performance. It is a figure for demonstrating the data about a product. It is a figure which shows the structure of the work capacity acquisition apparatus of FIG. It is a figure for demonstrating work capacity data. It is a figure for demonstrating the skill (ability) of a worker. It is a figure which shows the functional block composition of the trained model generator. It is a figure for demonstrating production planning data. It is a figure for demonstrating process plan data. It is a figure for demonstrating the staffing data, the type data, the production number data, and the work instruction data which make up process plan data. It is a figure for demonstrating new process plan data. It is a figure for demonstrating the staffing by the generated new process plan data. It is a figure for demonstrating the work instruction by the generated new process plan data. It is a figure which shows the functional block composition of the trained model generator in the 1st alternative example. It is a figure which shows the functional block composition of the trained model generator in the 2nd alternative example.

(1. Outline of production system)
The production system is a system that generates a process plan and work instructions using equipment composed of a plurality of processing machines according to the production plan. Specifically, the production system uses a trained model generated by machine learning to automatically generate a process plan (including staffing and work instructions) according to the production plan. Here, the production plan can be exemplified by, for example, planning the number of products to be produced in order to satisfy the delivery date and the number of deliveries ordered from the customer. Further, the process plan can exemplify, for example, planning the number of processing machines used for production and the arrangement of workers (staffing) among a plurality of processing machines in order to satisfy the production plan. Further, the work instruction can exemplify, for example, instructing a point of caution or a change point when operating the processing machine planned by the process plan among a plurality of workers.

In this case, the production system is, for example, a production plan, a production record or equipment capacity at which the equipment can produce a product, and a work record or work at which a worker can produce a product using a processing machine. A trained model is generated by performing machine learning with the ability as an explanatory variable and the process plan and work instructions made by an experienced skilled worker as the objective variable according to the production plan. Then, the production system generates and outputs a new new process plan (including staffing and work instructions) using the input new new production plan and the trained model.

(2. Outline of the configuration of production system 1)
The outline of the configuration of the production system 1 will be described with reference to FIG. The production system 1 includes equipment 10, one arithmetic unit 20 that functions as a trained model generation device, and one management terminal device 30. The equipment 10 includes a plurality of processing machines MC1 and MC2 as equipment. Here, regarding the arrangement of the plurality of processing machines MC1 and MC2 in the equipment 10, it is possible to arrange the plurality of processing machines MC1 and MC2 by the production line method, and the plurality of processing machines MC1 and MC2 are arranged by the cell production method. It is also possible to place it. However, the equipment 10 is not limited to being provided with a plurality of processing machines MC1 and MC2, and may be provided with one processing machine MC1.

The plurality of processing machines MC1 and MC2 are, for example, machine tools for producing various products, distribution machines for transporting products, and the like. Examples of the processing machines MC1 and MC2 as machine tools include a machining center, a lathe, a cutting device for cutting, a grinding machine for grinding, and the like. Here, the product is produced through a plurality of processing steps (processing steps) using a plurality of processing machines MC1 and MC2. Therefore, the plurality of processing machines MC1 and MC2 can perform at least one processing step among the plurality of processing steps (machining steps).

Each of the plurality of processing machines MC1 and MC2 is provided with an equipment operating state acquisition device 11 and a work capacity acquisition device 12. The equipment operation state acquisition device 11 and the work capacity acquisition device 12 are provided so as to be able to communicate with the arithmetic unit 20 and the management terminal device 30 by wire or wirelessly. In this example, the equipment operating state acquisition device 11 and the work capacity acquisition device 12 are provided in each of the plurality of processing machines MC1 and MC2. However, for example, one equipment operation state acquisition device 11 and a work capacity acquisition device 12 may be provided for each of the plurality of processing machines MC1 and MC2, or the equipment operation state acquisition device 11 (or the equipment operation state acquisition device 11) (or each of the plurality of processing machines MC1 and MC2) may be provided. It is also possible to provide one work capacity acquisition device 12 (equipment operating state acquisition device 11) for a plurality of processing machines MC1 and MC2 as well as the work capacity acquisition device 12).

The equipment operating state acquisition device 11 acquires the production capacity for producing a product and the operating equipment operating state of each of the processing machines MC1 and MC2. The equipment operation state acquisition device 11 acquires, for example, the machine performance, the type of tool used for machining, the stage change time required for stage change for producing a product, and the like as the equipment operation state for each of the processing machines MC1 and MC2. do. Further, the equipment operating state acquisition device 11 also acquires information on the products processed by the processing machines MC1 and MC2 as the equipment operating state. That is, the equipment operating state acquisition device 11 acquires information about the product, for example, the shape, the material, the processing site, the processing conditions, the processing time, and the like as the equipment operating state. Then, the equipment operation state acquisition device 11 outputs the equipment operation state data representing the acquired equipment operation state to the arithmetic unit 20 and the management terminal device 30.

The work capacity acquisition device 12 acquires the work capacity of a worker who operates each of the processing machines MC1 and MC2 to produce a product. The work ability acquisition device 12 quantified, for example, working hours, operation skills including abnormality handling for each of the processing machines MC1 and MC2, operation qualifications for each of the processing machines MC1 and MC2, and the like for each worker. Acquire working ability. Then, the work ability acquisition device 12 outputs the work ability data representing the acquired work ability to the arithmetic unit 20 and the management terminal device 30.

The arithmetic unit 20 includes equipment operation status data output from the equipment operation status acquisition device 11, work capacity data output from the work capacity acquisition device 12, and production planning data representing a production plan output from the management terminal device 30. Machine learning is performed using process plan data that is stored in the management terminal device 30 and is generated by a skilled worker and represents a proven process plan (including staffing and work instructions). Then, the arithmetic unit 20 generates and outputs a new new process plan (including staffing and work instructions) according to the production plan by using the trained model by machine learning.

As shown in FIG. 1, the arithmetic unit 20 that functions as a trained model generation device is composed of a learning processing device 21 and a generation processing device 22. In this example, the learning processing device 21 and the generation processing device 22 are shown as one device, but the learning processing device 21 and the generation processing device 22 may be configured independently. Further, a part or all of the arithmetic unit 20 may be an embedded system in the equipment 10.

The learning processing device 21 uses equipment operation status data, work capacity data, and production plan data as explanatory variables, and has a proven process plan (including staffing and work instructions) that has been drafted by a skilled worker (manager) in the past. Machine learning is performed using the process plan data representing the above as the objective variable, and a trained model is generated. The generation processing device 22 uses a process plan (including staffing and work instructions) using the trained model generated by the learning processing device 21 and new production plan data representing a new production plan acquired from the management terminal device 30. ) Is generated, and the generated new process plan data is output.

Here, the generation processing device 22 outputs the generated new process plan data to the management terminal device 30, and as shown in FIG. 1, the large common monitor 40 installed in the equipment 10 or each processing machine MC1. , Output to a small individual monitor 50 arranged in MC2. Thereby, the manager who is a skilled worker can confirm the contents of the new process plan (staffing and work instruction) represented by the generated process plan data by using the management terminal device 30. Further, the worker M and the worker N can confirm the new process plan and the work instruction related to themselves through the common monitor 40 or the individual monitor 50.

(3. Details of the configuration of production system 1)
The configuration of the production system 1 will be described in more detail. The production system 1 includes one or more equipment 10, an arithmetic unit 20, that is, a learning processing device 21, a generation processing device 22, a management terminal device 30, a common monitor 40, and an individual monitor 50. The equipment 10 mainly includes processing machines MC1 and MC2, which are machine tools, and equipment operating state acquisition devices 11 and work capacity acquisition devices 12 provided in the respective processing machines MC1 and MC2.

As shown in FIG. 2, the equipment operation state acquisition device 11 includes an interface 111 and an equipment operation state database 112. The interface 111 is communicably connected to the processing machines MC1 and MC2, acquires the equipment operation state data FD representing the operation state of the processing machines MC1 and MC2, and outputs the equipment operation state data FD to the equipment operation state database 112. Further, the interface 111 outputs the equipment operation status data FD stored in the equipment operation status database 112 to the learning processing device 21 and the generation processing device 22 of the arithmetic unit 20 and the management terminal device 30.

The equipment operation status database 112 stores and stores the equipment operation status data FD acquired from each of the processing machines MC1 and MC2 via the interface 111. In this example, the equipment operation status data FD includes various data related to machine performance and various data related to the product to be processed, as described above.

Specifically, as various data related to machine performance, for example, when the processing machines MC1 and MC2 are machine tool cutting devices, the number, stroke, and spindle size of each processing machine MC1 and MC2 are shown in FIG. And the capacity including the spindle rotation speed, the number of holders, the type and the tool including the recommended cutting feed, the shape of the product that can be clamped, the jig including the clamper type and the clamper quantity, the setup change time and the like can be included. Further, as shown in FIG. 4, various data regarding the product to be processed can include, for example, the shape, material, processing site, processing conditions, processing time, and the like of the product.

As shown in FIG. 5, the work capacity acquisition device 12 includes an interface 121 and a work capacity database 122. The interface 121 is communicably connected to the processing machines MC1 and MC2, acquires work capacity data OD representing the work capacity of the workers M and N who operate the processing machines MC1 and MC2, and outputs the work capacity data OD to the work capacity database 122. do. Further, the interface 111 outputs the work capacity data OD stored in the work capacity database 122 to the learning processing device 21 and the generation processing device 22 of the arithmetic unit 20 and the management terminal device 30.

The work capacity database 122 accumulates and stores the work capacity data OD acquired for each worker (for example, the workers M and N shown in FIG. 1). In this example, as shown in FIG. 6, the work capacity data is, for example, the number of workers who should be involved in the production of the product, whether or not the NC device can be operated, which can perform abnormality response (abnormal response capability), correction input, and the like. It can include abilities (skills) including (operation ability), processing time required to process a predetermined production quantity (for example, working hours (working hours), etc.), physical condition, and the like.

Here, regarding the skill (ability) of the worker, as shown in FIG. 7, for example, when each of the products A, B, and C, which are products, is produced through the process X and the process Y, the work in the process X is performed. The skill level indicating the degree of skill for each work content of XA and work XB, and work YA, work YB, and work YC of process Y is quantified and expressed. The skill level indicates that the larger the numerical value is, the more proficient the processing machines MC1 and MC2 to be used are used, that is, the higher the skill (ability) is.

(4. Functional block configuration of production system 1)
The functional block configuration of the production system 1 will be described with reference to FIG. As described above, the production system 1 includes an equipment operating state acquisition device 11, a work capacity acquisition device 12, an arithmetic unit 20, and a management terminal device 30.

The arithmetic unit 20 includes a learning processing device 21 and a generation processing device 22. The learning processing device 21 manages the equipment operation status data FD stored in the equipment operation status database 112 of the equipment operation status acquisition device 11 and the work capacity data OD stored in the work capacity database 122 of the work capacity acquisition device 12. New process plan data SKD (personnel) based on the past production plan data OSD input by the manager via the terminal device 30 and the process plan data JKD planned and proven by the manager (skilled worker). Generate a trained model to generate (including placement and work instructions). The learning processing device 21 includes a learning data set acquisition unit 211, a learning data set storage unit 212, and a model generation unit 213.

The learning data set acquisition unit 211 acquires a learning data set for performing machine learning. Specifically, the learning data set acquisition unit 211 acquires the equipment operation state data FD stored in the equipment operation state database 112 of the equipment operation state acquisition device 11 as a learning data set. In this example, as described above, the equipment operating state data FD includes various data related to mechanical performance (see FIG. 3) and data related to the product to be processed (see FIG. 4).

Further, the learning data set acquisition unit 211 acquires the work ability data OD stored in the work ability database 122 of the work ability acquisition device 12 as a learning data set. In this example, the work capacity data OD is acquired for each of the plurality of workers.

Further, the learning data set acquisition unit 211 acquires the production plan data OSD representing the past production plan as the learning data set via the management terminal device 30. As shown in FIG. 9, the production plan data OSD is, for example, the daily target production quantity for each of the products A, B, C, D, E, F, and G, which are the products represented by the monthly plan table. It represents.

Further, the learning data set acquisition unit 211 learns the process plan data JKD representing the process plan (actual result) created by the manager (skilled worker) based on the production plan in the past through the management terminal device 30. Get as a data set for. Specifically, the process plan data JKD is represented as shown in FIG. 10 in order to realize the production plan (see FIG. 9).

Then, in the process plan data JKD, when the products AG are produced by using the processing machines MC1 and MC2, the manager (skilled worker) produces each product AG by which processing machine, when and how much. It is planned to allocate workers who can operate the processing machine to be used. Therefore, as shown in FIG. 11, the process plan data JKD representing the process plan (actual result) includes the personnel allocation data indicating the allocation of workers when the products AG are actually produced, and the product (that is, the product (that is, the product). It includes type data indicating the types of products AG) and production number data indicating the production numbers of each of the products AG. Further, as shown in FIG. 11, the process plan data JKD contains work instructions for producing products AG, for example, work precautions different for each product and processing conditions obtained by past experience. Work instruction data representing the correction value of the above is included.

Therefore, the learning data set acquisition unit 211 acquires the process plan data JKD including the personnel allocation data, the type data, and the production number data as the learning data set via the management terminal device 30.

The learning data set storage unit 212 associates the equipment operation status data FD, the work capacity data OD, and the production plan data OSD acquired by the learning data set acquisition unit 211 with the process plan data JKD of the process plan (actual). Store as a training data set.

The model generation unit 213 performs machine learning, for example, machine learning with a teacher, using the learning data set stored in the learning data set storage unit 212. Specifically, the model generation unit 213 uses the equipment operation status data FD, the work capacity data OD, and the production plan data OSD stored in the learning data set storage unit 212 as explanatory variables, and processes the process plan (actual). Machine learning is performed using the planning data JKD (more specifically, staffing data, type data, production number data, and work instruction data) as objective variables. Then, the model generation unit 213 generates a trained model for generating a new process plan (new process plan data SKD described later) corresponding to the new production plan (new production plan data NSD described later).

The generation processing device 22 generates and outputs new process plan data SKD representing a new process plan based on, for example, a new production plan data NSD representing a new production plan acquired from the management terminal device 30. The generation processing device 22 includes a model storage unit 221, a generation data acquisition unit 222, a process plan data generation unit 223, and an output unit 224.

The model storage unit 221 stores the trained model generated by the model generation unit 213. In this example, the model storage unit 221 (learned model storage unit) is provided in the generation processing device 22. However, it is also possible to provide the trained model storage unit in the learning processing device 21.

The generation data acquisition unit 222 acquires the equipment operation status data FD, the work capacity data OD, and the new production plan data NSD from each of the equipment operation status acquisition device 11, the work capacity acquisition device 12, and the management terminal device 30. Here, the generation data acquisition unit 222 acquires the equipment operation state data FD, the work capacity data OD, and the new production plan data NSD by the same processing as the learning data set acquisition unit 211.

In this example, the generation data acquisition unit 222 will be described as a separate element from the learning data set acquisition unit 211. However, the function of the learning data set acquisition unit 211 to acquire the equipment operation status data FD, the work capacity data OD, and the production plan data OSD can also be used together with the generation data acquisition unit 222. That is, a part of the functions of the learning processing device 21 is also used as a part of the functions of the generation processing device 22.

The process plan data generation unit 223 acquires the trained model stored in the model storage unit 221. Further, the process plan data generation unit 223 acquires the equipment operation state data FD, the work capacity data OD, and the new production plan data NSD acquired by the generation data acquisition unit 222. As a result, the process plan data generation unit 223 uses the equipment operation status data FD, the work capacity data OD, the new production plan data NSD, and the trained model, as shown in FIG. 12, the workers M, N, O. Generates new process plan data SKD representing a new process plan for producing products, that is, products A to G by the processing machines MC1 and MC2. Here, in the new process plan data SKD generated by the process plan data generation unit 223, as in the process plan data JKD of the past process plan (actual), personnel allocation data, type data, production number data and work instructions Contains data.

The output unit 224 outputs the new process plan data SKD generated by the process plan data generation unit 223 to the common monitor 40 or the individual monitor 50. As a result, the common monitor 40 or the individual monitor 50 can display the generated process plan as shown in FIG. 12 based on the generated new process plan data SKD. As a result, each worker can easily grasp the process for producing the products AG.

Further, the generated new process plan data SKD includes personnel allocation data and work instruction data. Here, the staffing data is mainly data that is highly related to the work capacity data OD. Therefore, as shown in FIG. 13, the common monitor 40 or the individual monitor 50 is an operator who has little experience in operating the processing machine MC1 used in the process X, for example, when the product A is produced using the processing machine MC1. The state in which O and the worker M who is excellent in the operation of the processing machine MC1 are arranged as a support is displayed. Further, as shown in FIG. 13, the common monitor 40 or the individual monitor 50 can display, for example, the staffing data in which only the worker N who is excellent in the operation of the processing machine MC2 used in the process Y is assigned.

Further, as shown in FIG. 14, the individual monitor 50 provided in the processing machine MC1 is operated by the processing machine MC1 for the worker M and the worker O who work in the work XA of the process X for producing the product A, for example. Work instructions for correcting roughing conditions can be displayed. As a result, when the worker M and the worker O work in the process X, the product A is appropriately rough-processed by applying the correction value instructed to the operating state of the processing machine MC1. Can be done.

Similarly, the individual monitor 50 provided in the processing machine MC2 displays, for example, a work instruction for correcting the finishing processing conditions by the processing machine MC2 to the worker N working in the process Y for producing the product A. be able to. As a result, the worker N can appropriately finish the product A by applying the correction value instructed to the operating state of the processing machine MC2 when working in the process Y.

As can be understood from the above description, the production system 1 can include an arithmetic unit 20, that is, a learning processing device 21 and a generation processing device 22 that constitute the trained model generation device. As a result, the equipment operating status data FD representing the operating status of the processing machines MC1 and MC2, the working capacity data OD representing the capacities of the workers M, N, and O, the production plan data OSD, and the past process plan, That is, new process plan data SKD can be generated by machine learning using the process plan data JKD representing a proven process plan as a learning data set. As a result, even if an inexperienced worker becomes a manager and formulates a process plan, it is generated by machine learning using a process plan that has a proven track record as an experienced skilled worker. By using the trained model, it is possible to easily and highly productive process planning.

(5. First alternative example)
The management terminal device 30 has "working hours (working hours)" (see FIG. 6) for each worker obtained from the work capacity acquisition device 12 and production results for each worker (for example, the number of products produced). ) Can be stored as process execution ability data PD. Then, as shown in FIG. 15, the management terminal device 30 can also output the process execution ability data PD to the learning processing device 21 and the generation processing device 22.

As a result, when the learning processing device 21 includes the process execution ability data PD as the learning data set, the learning processing device 21 can perform machine learning by including the process execution ability data PD in the explanatory variables. Further, when the process execution ability data PD is acquired, the generation processing device 22 determines the worker based on the acquired process execution ability data PD, and generates the personnel allocation data included in the new process plan data SKD. be able to.

Therefore, in the case of the first alternative example, for example, it is possible to assign a highly productive worker to each process. As a result, in the equipment 10, as shown in FIG. 13, a plurality of workers can be placed in the right place in the right place, and as a result, the product can be produced with high production efficiency.

(6. Second example)
The management terminal device 30 has abnormalities or failures in the "working hours (working hours)" (see FIG. 6) for each worker obtained from the work ability acquisition device 12 and the processing machines MC1 and MC2 constituting the equipment 10. When it occurs, it can be stored as improvement ability data YD for each worker in association with the return time until the worker returns the processing machines MC1 and MC2 to the normal production state. Then, as shown in FIG. 16, the management terminal device 30 can also output the improvement capability data YD to the learning processing device 21 and the generation processing device 22.

As a result, when the improvement ability data YD is included as the learning data set, the learning processing device 21 can perform machine learning by including the improvement ability data YD in the explanatory variables. Further, when the improvement capacity data YD is acquired, the generation processing apparatus 22 can generate the staffing data included in the new process plan data SKD based on the acquired improvement capacity data.

Therefore, in the case of the second alternative example, for example, as shown in FIG. 13, it is possible to assign a worker having a high improvement ability to the processing machine MC1 in which an abnormality or a failure is likely to occur. As a result, even if an abnormality or failure occurs in the equipment 10, it is possible to shorten the time during which production is stopped, and as a result, the product can be produced with high production efficiency.

(7. Third alternative example)
As described in the second alternative example, when an abnormality or failure occurs in the processing machines MC1 and MC2, the production number fluctuates due to the stoppage of the processing machines MC1 and MC2 even for a highly productive worker. there is a possibility. Further, for example, when the depth of machine learning by the learning processing device 21 is shallow, the accuracy of the generated trained model becomes low, and as a result, the new process planning data SKD (process planning) generated by the generation processing device 22 The accuracy may be low. Further, if the past process plan data JKD used by the learning process device 21 as the training data set is incorrect, the accuracy of the generated trained model becomes low, and as a result, the new process plan generated by the generation process device 22. The accuracy of the data SKD (process plan) may be low.

Therefore, as shown in FIG. 1, the management terminal device 30 adjusts the new process plan data SKD generated by the generation processing device 22, and more specifically, the adjustment unit 31 that adjusts the personnel allocation data and the work instruction data. Can have. In this case, the manager (skilled worker) changes the staffing based on the generated staffing data, for example, based on the past experience by using the adjusting unit 31 of the management terminal device 30. Alternatively, the correction value based on the work instruction data shown in FIG. 14 can be changed.

Then, the learning processing device 21 uses the new process plan data SKD adjusted by using the adjustment unit 31 of the management terminal device 30 as the process plan data JKD representing a new process plan (actual result), and the generation unit 213 is new. Machine learning can be performed by including the process plan data JKD (adjusted new process plan data SKD) as an explanatory variable. Even in this case, the trained model generated by the generation unit 213 is stored in the model storage unit 221 of the generation processing device 22. That is, the generation unit 213 of the learning processing device 21 can generate a new trained model, and the model storage unit 221 of the generation processing device 22 can update and store the new trained model.

As a result, in the third alternative example, the learning processing device 21 can generate a trained model with higher accuracy. Then, the generation processing apparatus 22 can generate highly accurate new process planning data SKD, that is, personnel allocation data and work instruction data by using a more accurate trained model.

1 ... Production system, 10 ... Equipment, 11 ... Equipment operation status acquisition device, 12 ... Work capacity acquisition device, 20 ... Arithmetic device (learned model generation device), 21 ... Learning processing device, 22 ... Generation processing device, 30 ... Management terminal device, 31 ... Adjustment unit, 40 ... Common monitor, 50 ... Individual monitor, 211 ... Learning data set acquisition unit, 212 ... Learning data set storage unit, 213 ... Model generation unit (learned model generation unit) , 221 ... Model storage unit (learned model storage unit) 222 ... Generation data acquisition unit 223 ... Process plan data generation unit 224 ... Output unit, MC1, MC2 ... Processing machine, FD ... Equipment operation status data, OD ... Work capacity data, PD ... Process execution capacity data, YD ... Improvement capacity data, OSD ... Production plan data, NSD ... New production plan data, JKD ... Process plan data, SKD ... New process plan data

Claims (15)

Equipment operating status data showing the equipment operating status of the equipment that produces the product, work that represents the ability of the worker who uses the equipment to produce the product using the equipment to produce the product. A training data set for acquiring capacity data, production plan data representing a production plan for producing the product, and process plan data representing a past process plan used for producing the product as a training data set. Acquisition department and
New production is performed by performing machine learning using the equipment operating state data, the work capacity data, and the production planning data acquired by the learning data set acquisition unit as explanatory variables, and the process planning data and the objective variable. A trained model generator that generates a trained model for generating new process plan data that represents a new process plan for new production plan data that represents a plan.
A trained model generator. The equipment operation status data is
The trained model generator according to claim 1, which includes data on the mechanical performance of the processing machine constituting the facility and data on the product. The work capacity data is
The invention according to claim 1 or 2, which includes data on the abnormality handling ability of the worker corresponding to an abnormality occurring in the processing machine constituting the equipment and data on the operation ability of the worker operating the processing machine. Trained model generator. The data relating to the abnormality handling ability and the data relating to the operating ability are data expressed numerically based on the past work contents for producing the product by the worker, according to claim 3. The trained model generator described. The work capacity data is
The trained model generator according to claim 2 or 3, further comprising data on the processing time required to process a predetermined production quantity and data on the physical condition of the worker. The process plan data and the new process plan data include the personnel allocation data that determines the allocation of the worker to the equipment, and the work instruction data that represents the work instruction to the worker who operates the equipment. -The trained model generator according to any one of 5. The trained model generation device according to any one of claims 1-6, comprising a trained model storage unit that stores the trained model generated by the trained model generation unit. A production system comprising the trained model generator according to any one of claims 1-7 and producing the product.
The management terminal device that outputs the new production plan data, and
The equipment operation state acquisition device that acquires the equipment operation state data of the equipment and outputs the learned model generation device and the management terminal device, and the equipment operation state acquisition device.
A work ability acquisition device that acquires the work ability data of the worker and outputs the learned model generation device and the management terminal device.
It is equipped with a generation processing device that generates and outputs the new process plan data.
The generation processing device is
The equipment operation state data output from the equipment operation state acquisition device, the work capacity data output from the work capacity acquisition device, the new production plan data output from the management terminal device, and the learning. A production system that generates the new process plan data and outputs the generated new process plan data by using the trained model generated by the completed model generator. The management terminal device is
The data regarding the processing time required for the worker to process a predetermined production number, which is included in the work capacity data acquired from the work capacity acquisition device, and the production in which the product is produced by the worker. Memorize process execution ability data associated with actual results,
The trained model generation unit of the trained model generation device is
The production system according to claim 8, wherein the trained model is generated by performing machine learning by including the process execution ability data acquired from the management terminal device as the learning data set in the explanatory variables. The management terminal device is
The data regarding the processing time required for the worker to process a predetermined production number, which is included in the work capacity data acquired from the work capacity acquisition device, and the worker when an abnormality occurs in the equipment. Stores the improvement capacity data associated with the return time until the equipment is returned to the normal production state.
The trained model generation unit of the trained model generation device is
The production system according to claim 8 or 9, wherein the trained model is generated by performing machine learning by including the improvement ability data acquired from the management terminal device as the learning data set in the explanatory variables. .. The management terminal device is
The production system according to any one of claims 8-10, comprising an adjusting unit for adjusting the new process planning data generated by the generation processing apparatus. The production system according to any one of claims 8-11, wherein the product is produced through a plurality of processing steps. The equipment has a plurality of processing machines and has a plurality of processing machines.
The production system according to any one of claims 8-12, wherein the processing machine performs at least one processing step among a plurality of processing steps. The processing machine is a machine tool and
The production system according to claim 13, wherein the product is produced through processing by the machine tool. The production system according to claim 14, wherein the processing is cutting or grinding.

Images