Machining plan creation support device, machining plan creation support system, machining plan creation support procedure and program

Abstract

A processing plan creation support device (1) comprises a personnel allocation calculator (18) to calculate personnel allocation for an elementary operation corresponding to a production plan described by production plan information provided by a production plan information acquirer (15) based on on-site worker information, action definition information, and action time information for each worker; an action plan information generator (19) to modify a production sequence, initiate the calculation of personnel allocation, and generate the processing plan information describing the personnel allocation and a production sequence to maximize an evaluation value for productivity; and an action time information updater (14) to update the action time information for a worker with a past action log based on image information.which describes a picture of the worker, predicts an elementary action time of a worker without a past action record based on the action time information for a worker with a past action record who has a feature most similar to a feature of the worker without a past action record, and generates the action time information for the worker without a past action record.

Description

Technical area

[0001] The present disclosure relates to a machining plan creation supporting apparatus, a machining plan creation supporting system, a machining plan creation supporting method, and a program. State of the art

[0002] At a production site, workers assigned to items to be manufactured based on a production plan receive instructions to perform actions, for example, with a method for acquiring information on the action time of each of the workers and providing the worker with efficient action instructions that maximize the evaluation value for productivity.

[0003] Patent Literature 1 describes a machining plan creation support device that calculates an action time of an action unit assigned to each work station in a production line, calculates an evaluation index for each work station based on the calculated action time, and outputs the calculated evaluation index. Citation listPatent literature

[0004] Patent Literature 1: Unexamined Japanese Patent Application Publication No. 2018-26071 Overview of the inventionTechnical problem

[0005] The technique described in Patent Literature 1 may not be able to provide efficient action instructions to new workers, assistants, or other workers whose action times are unknown due to a lack of data, so their expected action times are unknown.

[0006] In view of the above circumstances, it is an object of the present disclosure to make it possible to provide efficient action instructions to workers with unknown action times. Solution to the problem

[0007] To achieve the above objective, a machining plan creation support device according to one aspect of the present disclosure includes a production plan information acquirer, a personnel assignment calculator, a machining plan information generator, and an action time information updater. The production plan information acquirer acquires production plan information indicating a production plan for a product. The personnel assignment calculator calculates personnel assignments for an elementary action corresponding to the production plan indicated by the production plan information based on on-site worker information, action definition information, and action time information. The on-site worker information describes a worker who is on-site. The action definition information describes an elementary action included in each process in a production process for the product.The action time information describes the elementary action time required for each worker to perform the elementary action. The processing plan information generator changes a production process, causes the personnel allocation calculator to calculate the personnel allocation, and generates processing plan information that describes the personnel allocation and a production process to maximize an evaluation value for productivity.The action time information updater updates the action time information for a worker with a past action record based on image information describing an image of the worker, predicts an elementary action time of a worker with no past action record based on the action time information for a worker with a past action record who has a feature most similar to a feature of the worker with no past action record, and generates the action time information for the worker with no past action record. Advantageous effects of the invention

[0008] When preparing a machining plan that maximizes an evaluation value for productivity based on the action time of each worker, the apparatus according to the above aspect of the present disclosure predicts the elementary action time of a worker with no past action history based on the action time information of a worker who has a past action history and has a characteristic most similar to the characteristic of the worker with no past action history, and generates action time information describing the elementary action time of the worker with no past action history, so as to make it possible to provide efficient action instructions to workers with unknown action times. Brief description of the drawings Fig. 1 is a schematic diagram of a machining plan creation support device according to Embodiment 1; Fig. 2 is a functional block diagram of a machining plan creation support device according to Embodiment 1; Fig. 3 is a diagram of an example of the action time information in Embodiment 1; Fig. 4 is a diagram of exemplary action definition information in Embodiment 1; Fig. 5 is a functional block diagram of an example of an action time information updater in Embodiment 1; Fig. 6 is a diagram showing an example of personal characteristic information in Embodiment 1; Fig. 7 is a diagram of the personal property information of a new worker X in Embodiment 1; Fig. 8 is a conceptual diagram of a process for predicting an elementary action time of the new worker X in Embodiment 1; Fig. 9 is a graph showing an exemplary probability distribution of the elementary action time of the new worker X in Embodiment 1; Fig. 10 is a flowchart of an example of a machining plan creation support process in Embodiment 1; Fig. 11 is a flowchart of an example of an action time prediction process in Embodiment 1; and Fig. 12 is a diagram of a machining plan creation support device according to Embodiments 1 and 2, showing an exemplary hardware configuration. Description of embodiments

[0009] A machining plan creation support device, a machining plan creation support system, a machining plan creation support method, and a program according to the present embodiment will be described in detail with reference to the drawings. Similar reference numerals denote similar or corresponding components in the drawings. In the following embodiments, an exemplary operation of machining plan creation support for assigning actions included in a production process for manufacturing an article to workers based on a production plan for manufacturing the article will be explained. A manufactured article is an example of a product. Embodiment 1

[0010] The structure of a machining plan creation support system 100 according to Embodiment 1 will be described with reference to Fig. 1. A user (a manager in Fig. 1) of the machining plan creation support system 100 inputs machining plan information, which represents a production plan for an item to be manufactured today, into a machining plan creation support device 1. The production plan information describes the number of units to be produced for each model of the item to be manufactured. The information does not need to specify the production process. The model of the item to be manufactured is an example of a product type.

[0011] A camera 2 captures an image of a worker. The camera may, for example, be a fixed three-dimensional (3D) camera that captures images of the worker's hand while working. The camera 2 transmits image information describing the captured image of the worker to the machining plan creation support device 1.

[0012] The machining plan creation support device 1 calculates personnel assignments for actions that correspond to the production plan based on the input production plan information and the obtained image information. The machining plan creation support device 1 calculates the productivity evaluation value based on the calculated personnel assignment. The machining plan creation support device 1 changes the production flow, executes personnel assignments, and calculates the personnel assignment and production flow that maximizes the productivity evaluation value. The machining plan creation support device 1 outputs the machining plan information describing the personnel assignment and production flow that maximizes the productivity evaluation value.

[0013] The functional components of the machining plan creation support device 1 are described with reference to Fig. 2. The machining plan creation support device 1 includes an on-site worker information generator 11 that receives image information from the camera, identifies a worker who is on-site based on the image information, and generates on-site worker information describing the on-site worker; an on-site worker information storage 12 that stores the on-site worker information; an action time information storage 13 that stores action time information describing the action time for each worker; an action time information updater 14 that receives the image information from the camera 2 and updates the action time information based on the image information; a production plan information acquirer 15 that receives input of production plan information describing today's production plan from the user; and a production plan information storage 16.which stores the production plan information, and an action definition information memory 17 which stores action definition information describing the elementary actions included in the processes for each model.

[0014] The machining plan creation support device 1 further includes a personnel assignment calculator 18 that calculates personnel assignments for actions corresponding to the production plan described by the production plan information based on the on-site worker information, the action time information, and the action definition information; a machining plan information generator 19 that calculates personnel assignments and a production flow that maximize the evaluation value for productivity and generates the machining plan information; and a machining plan information output unit 20 that outputs the machining plan information.

[0015] The on-site worker information generator 11 acquires the name of a worker working on-site, or more specifically, the name of a worker who is on-site, by identifying a person in images described by the image information obtained from the camera 2. The on-site worker information generator 11 generates on-site worker information indicating the name of the worker who is on-site and stores the information in the on-site worker information storage 12. A person in the images can be identified, for example, by reading a worker-specific augmented reality (AR) marker or by facial recognition technology.

[0016] The action time information memory 13 stores the action time information describing the action time of each worker. The action time information is stored with reference to Fig. 3. In the example of Fig. 3, the action time information includes data required for an elementary action for each combination of the worker, the model, and the elementary action. The time required for the elementary action is hereinafter referred to as an elementary action time. The elementary action time described by the action time information is not a constant but a continuous probability distribution, with the elementary action time indicated by the horizontal axis and the number of times indicated by the vertical axis. Fig. 3 shows the probability distributions of the elementary action numbers of a combination of a worker 1, a model M1 and an elementary action E01, a combination of the worker 1, the model M1 and an elementary action E02, a combination of a worker 2, the model M1 and the elementary action E01, and a combination of the worker 2, the model M1 and the elementary action E02.

[0017] For each combination of a worker, a model, and an elementary action, action time information describing an elementary action time based on a history is stored for a worker with a past action history, while action time information describing a predicted elementary action time is stored for a worker without a past action history (e.g., a new worker or an assistant). A method for predicting an elementary action time is described later. Each new action history acquired for a combination of a worker, a model, and an elementary action results in an update of the action information.

[0018] Reference is again made to Fig. 2. The personnel allocation calculator 18 performs personnel allocation that maximizes the productivity evaluation value for the production plan (the model and the number of units) described by the production plan information based on the on-site worker information, the action time information, and the action definition information. Personnel allocation refers to the allocation of a worker to an elementary action included in each process for producing a manufactured item. The productivity evaluation value is, for example, the value obtained by dividing the total number of units produced by the total on-site worker time (units / hour). In this case, productivity is higher when the number of units produced divided by the total on-site worker time (units / hour) is larger.

[0019] The action definition information is provided with reference to Fig. 4. In the example of Fig. 4, the action definition information specifies the types of production processes that comprise the elementary actions for each model. The production process for model M1 comprises four processes, or setup: process 1, process 2, and process 3. The setup includes an elementary action E01..., an elementary action E0n0. Process 1 includes an elementary action E11..., an elementary action E1n1. Process 2 includes an elementary action E21..., an elementary action E2n2. Process 3 includes an elementary action E31..., an elementary action E3n3. For example, process 1 is pre-assembly, process 2 is property testing, and process 3 is final assembly. For example, the elementary action E01 in the setup is a tool exchange, and the elementary action E0n0 is an entry operation.For example, the elementary action E11 in process 1 is a tool exchange, and the elementary action E1n1 is screwing. The elementary action E21 in process 2 is a tool exchange, and the elementary action E2n2 is an inspector's acceptance test. The elementary action E31 in process 3 is a tool exchange, and the elementary action E3n3 is a transport.

[0020] Reference is again made to Fig. 2. The machining plan information generator 19 changes the production flow for the model and causes the personnel allocation calculator 18 to again perform the personnel allocation that maximizes the evaluation value. If the maximum evaluation value calculated by the personnel allocation calculator 18 exceeds the previous calculation result, the optimal solution for the personnel allocation and the work flow is updated. The machining plan information generator 19 repeats the above processing, compares all the maximum evaluation values ​​calculated by the personnel allocation calculator 18, and generates machining plan information using the personnel allocation and the production flow that maximizes the evaluation value as an optimal solution.

[0021] The machining plan information output unit 20 outputs the machining plan information generated by the machining plan information generator 19. The machining plan information can be output, for example, on a screen or using sound. Alternatively, the machining plan information can be sent to a user terminal used by the user (the manager or the worker).

[0022] The action time information updater 14 identifies the model and person in the images described by the image information obtained from the camera 2 and measures the elementary action time. The action time information updater 14 updates the action time information stored in the action time information memory 13 based on the measured elementary action time.

[0023] The functional components of the action time information updater 14 are described with reference to Fig. 5 described.The action time information updater 14 includes a detector 141 that detects a worker and a model of the object to be manufactured handled by the worker from the images described by the image information, an action time measuring module 142 that measures the elementary action time of the worker, and an updater 143 that updates the action time information based on the elementary action time measured by the action time measuring module 142, a human feature information storage 144 that stores human feature information indicating the characteristics of each worker with a past action record, and an action time prediction module 145 that predicts the elementary action time based on the human feature information when a worker detected by the detector 141 has no past action record (for example, a new worker or an assistant).

[0024] The detector 141 detects the worker and the model of the object to be manufactured being handled by the worker from the images described by the image information obtained from the camera 2. The detector 141 refers to the personal characteristic information stored in the personal characteristic information storage 141 and determines whether the detected worker has a past action history. If the detected worker has a past action history, the detector 141 sends information describing the detected worker and the detected model, as well as the image information, to the action time measurement module 142.If the detected worker is not a worker with a past action history (e.g., a new worker or an assistant), the detector 141 sends information describing the detected worker and the detected model and the image information to the action time prediction module 145.

[0025] When the action time measurement module 142 receives information describing the worker and the model as detected by the detector 141 and the image information, it identifies the elementary action performed by the worker based on the images described by the image information and measures the worker's elementary action time. The elementary action can be identified, for example, using a behavior classifier. The behavior classifier can classify a series of action images determined by artificial intelligence (AI) based on images of the worker during the action and acquire the elementary action time. If the elementary action can be identified with a tool, an AR marker specific to the tool can be read to identify the elementary action.The action time measuring module 142 sends to the updater 143 information indicating the worker and the model as obtained from the detector 141 and information indicating the calculated elementary action time.

[0026] The updater 143 updates, based on the information indicating the worker and the model obtained from the action time measuring module 142 and information indicating the elementary action time, the action time information describing a combination of the corresponding worker, the model, and the elementary action included in the action information stored in the action time information storage 13.

[0027] The personal characteristic information is provided with reference to Fig. 6. The personal attribute information store stores the worker's attributes and the values ​​for the attributes. In the example of Fig. 6 The characteristics of the worker include name, gender, age, years of service, physical constitution, evaluation for action A, evaluation for action B and evaluation for action C.

[0028] The value for gender is 1 for males and 2 for females. The value for age is 1 for 10 to 19 years old, 2 for twenties, 3 for thirties, 4 for forties, 5 for fifties, and 6 for sixties and older. The value for years of service is 1 for ten years or less, 2 for 11 to 20 years, 3 for 21 to 30 years, 4 for 31 to 40 years, and 5 for 41 years or more. The value for physical constitution is 1 for thin, 2 for standard, and 3 for fat. The evaluation for action A is an evaluation of the speed at which action A can be completed. Finally, the evaluation for action B is the evaluation of the speed at which action B can be completed, and the evaluation for action C is the evaluation of the speed at which action C can be completed.Actions A, B, and C include, for example, picking up components, installing, screwing, inserting electronic components, joining a connection, soldering, and transporting. The scores for the evaluation of action A, the evaluation of action B, and the evaluation of action C are 3 for fast, 2 for standard, and 1 for slow, respectively. The evaluation scores for the evaluation of action A, the evaluation of action B, and the evaluation of action C are determined based on predetermined rules.

[0029] The values ​​for the worker’s characteristics in the personal characteristic information do not have to be the same as in Fig. 6 may be limited. For example, age may be more finely classified in units of five years, or it may be an exact age value. Years of service may be the value of continuous years of service. Physical constitution may be the body mass index (BMI), which is calculated based on height and weight. The characteristics may include the evaluation value for the speed with which an action other than actions A, B, and C is completed. The personal characteristic information includes the characteristics and values ​​for the characteristics of a worker with no past action record (for example, a new worker or an assistant).

[0030] It will be back on Fig. 5. The action time prediction module 155 determines, based on the personal characteristic information, a worker with a past action history who is similar to a worker without a past action history (for example, a new worker or an assistant) using a calculation formula. The action time prediction module 145 refers to the action information about a worker with a past action history who is similar to a worker without a past action history (for example, a new worker or an assistant) and predicts the elementary action time of the worker without a past action history.

[0031] The elementary action time of a new worker X, who is a worker without a past action record, is predicted in the example given below with reference to the Fig. 7 and Fig. 9. The action time prediction module 145 executes steps 1 to 4 to predict the elementary action time of the new worker X. In step 1, an individual (a worker with a past action record) whose characteristics are most similar to the characteristics of the new worker X is identified. In step 2, action time information for the individual is read from the action time information store 13. In step 3, the skill level is inversely calculated. In step 4, the probability distribution of the elementary action time in the elementary action for the model for the new worker X is calculated.

[0032] Step 1 is described first. The attribute values ​​of the new worker X are denoted by a01, a02, ..., a0N. In the example of Fig. 7, the characteristic values ​​of the new worker X are 1 for gender (male), 2 for age (20 years), 1 for years of service (ten years or less), and 3 for physical constitution (bold). The evaluation value is 2 for the evaluation of action A, 1 for the evaluation of action B, and 3 for the evaluation of action C. The characteristic values ​​of a worker i (i = 1, 2, ..., N) are denoted by ai1, ai2, ..., aiN. A similarity degree Si between the characteristics of the new worker X and worker i is expressed by Formula 1 below. Si=∑j=1nkj(aij−a0j)2

[0033] The action time prediction module 145 determines that the worker i with the attributes with the smallest similarity degree Si is a worker who has the characteristics most similar to the characteristics of the new worker X. The coefficient kj (j = 1, ..., n) (0 ≤ kj ≤ 1) is a weighting coefficient for each characteristic included in the person characteristic information. A smaller value of kj represents the characteristic that is more likely to affect the similarity degree.

[0034] When the new worker X actually performs an action and the action time measurement module 142 measures the elementary action time, the action time prediction module 145 updates the coefficient kj (j = 1, ..., n) at a predetermined time. The coefficient kj (j = 1, ..., n) is updated in the manner described later.

[0035] For example, the action time prediction module 145 outputs as the initial values ​​of the coefficients kj (j = 1, ..., n), k1 = 1.0 (gender), k2 = 0.5 (age), k3 = 0.5 (years of service), k4 = 0.7 (diagonal), k5 = 0.1 (evaluation of action A), k6 = 0.1 (evaluation of action B), and k7 = 0.1 (evaluation of action C). For these initial values, the coefficient kj can be set low for each feature likely to have a high correlation in predicting the degree of similarity.

[0036] Although Formula 1 uses the least squares method to calculate the degree of similarity, the degree of similarity can be calculated using a different method. For example, multiple regression analysis can be used.

[0037] In step 2, the action time prediction module 145 reads from the action time information storage 13 the action time information for worker i with the smallest feature similarity degree Si. The elementary action time of the elementary action to the model for worker i selected in step 2 is the probability distribution resulting from worker i having a skill level acquired through experience in the elementary action. The new worker X does not have such experience. In step 3, the skill level is therefore back-calculated by predicting the elementary action time of the new worker X. The skill level is typically expressed using Formula 2 below. y(x)=cpa(x)

[0038] In the above formula, y(x) is the elementary action time at the x-th production, and x ≥ 0, c is the elementary action time at the first production and c ≥ 0, and p is the skill rate, and p = y(2x) / y(x). a(x) = logx / log2.

[0039] The action time prediction module 145 acquires the probability distribution of the elementary action time of the new worker X by dividing the probability distribution of the elementary action time at the current time (action number x) of the worker i with the highest degree of feature similarity to the new worker X by the ability rate p. The ability rate p and the action number x for each worker are stored in the individual feature information storage 144. As each worker experiences more actions, the ability rate p, which is described by the ability level information, is updated accordingly.

[0040] In step 4, a probability distribution graph of the elementary action time in the elementary action to the model is calculated for the new worker X. The action time prediction module 145 repeats steps 1 to 4 in the same way to calculate the probability distribution graph of the elementary action time for each new worker X for all models and all elementary actions.

[0041] Fig. Figure 8 is a conceptual diagram of a process for predicting the elementary action time of the new worker X. In the example of Fig. 8, the worker i with the smallest feature similarity degree Si to the new worker X is a worker 4. In other words, the characteristics of worker 4 are most similar to the characteristics of the new worker X among the workers 1 to 4. The action time prediction module 145 acquires the probability distribution of the elementary action time of the new worker X by dividing the probability distribution of the elementary action time at the current time of worker 4 with the highest similarity degree to the new worker X by the ability rate p.

[0042] Fig. Figure 9 is a graph showing an example probability distribution of the elementary action time of the new worker X as predicted by the action time prediction module 145. In the example of Fig. 9, the probability distribution gives an elementary action time for the combination of the new worker X, the model M1 and the elementary action E01.

[0043] The action time prediction module 145 stores the action information describing the predicted probability distribution for each elementary action time of the new worker X in the action time information storage 13. When the new worker X performs actions a predetermined number of times, the action time prediction module 145 stores the action information describing the probability distribution for each elementary action time of the action performed by the new worker X in the action time information storage 13. The predetermined number of times may be, for example, 50 times. The action information describing the probability distribution for each elementary action time of the action performed by the new worker X is stored in the action time information storage 13 to achieve the new worker X becoming a worker with an action record.

[0044] The coefficient kj (j = 1, ..., n) is updated in the manner described in detail below. The action time prediction module 145 generates action time information for the new worker X. The action time prediction module 145 compares the pre-predicted elementary action time with the elementary action time based on the protocol, evaluates the accuracy, and updates the coefficient kj (j = 1, ..., n), which is used to predict the degree of similarity.

[0045] The probability density function predicted as an elementary action time of the new worker X is denoted by f0(t), and the probability density function acquired as an elementary action time of the new person X based on the log is denoted by fX(t), where t (0 ≤ t) is a real number representing the elementary action time. The action time prediction module 145 first evaluates the accuracy of the prediction. The action time prediction module 145 calculates an evaluation value V for evaluating the prediction accuracy based on the difference between the prediction and the log. The evaluation value V is calculated using Formula 3 below. V=∫0∞|fX(t)−f0(t)|dt

[0046] The action time prediction module 145 stores the evaluation value V as a history. The evaluation value V is smaller when the prediction accuracy is higher. If the predicted elementary action time and the elementary action time based on the log completely agree, V = 0.

[0047] The action time prediction module 145 then calculates a similarity degree Vi to the elementary action time of another worker i (i = 1, ..., N) using Formula 4 below, where fi(t) is the probability density function of the elementary action time based on the log of the other worker i. Vi=∫0∞|fX(t)−fi(t)|dt

[0048] The worker with the smallest elementary action time similarity degree Vi (a person with the highest elementary action time similarity to the new worker X) is a worker Y. The action time prediction module 145 uses Formula 1 to quantify the feature similarity degree to calculate the feature similarity degree Si of the other worker i (i = 1, ..., N). The action time prediction module 145 calculates and updates the coefficients k1, ..., kn that satisfy Formula 5. SY(=∑j=1nkj(aij−aYj)2)→min SY≦S1,SY≦S2,…,SY≦SN

[0049] In Formula 5, 0 ≤ kj ≤ 1 with (j = 1, ..., n). The coefficients k1, ..., kn that satisfy Formula 5 can be calculated by, for example, setting each coefficient to three decimal places or smaller values ​​and repeating the search a finite number of times to obtain the optimal solution. The coefficient kj (j = 1, ..., n) is updated to the value that has the smallest feature similarity degree SY to the worker Y who has the smallest elementary action time similarity degree Vi. This results in the person who has the highest elementary action time similarity to the new worker X also having the highest feature similarity to the new worker X, thereby improving the prediction accuracy of the elementary action time in subsequent actions.

[0050] The machining plan creation support device 1 updates the personal feature information stored in the personal feature information storage 144 and the action definition information stored in the action definition information storage 17 at predetermined timings.

[0051] For example, in the personal characteristic information, the values ​​of the characteristics that change, such as age, years of service, physical constitution, evaluation of action A, evaluation of action B, and evaluation of action C, are updated based on the changes. If the machining plan creation support device 1 stores information describing a worker's birthday and employment date as information associated with the personal characteristic information, the age and years of service can be automatically updated. The machining plan creation support device 1 can also acquire information describing the worker's physical constitution and automatically update the physical constitution, for example, by establishing a link with an in-house health management system that manages the worker's health.

[0052] The action definition information is edited and updated by the user when the configuration of an elementary action included in each process in the manufacturing process for each model changes, for example, when the production of a new model starts or when the design of an object to be manufactured changes.

[0053] In the structure where the user updates the action definition information, the user may forget to update the information or may not notice changes in the configuration of the elementary action. Therefore, the machining plan creation support device 1 can notify the user when the configuration of an elementary action included in each process in the manufacturing process for each model may have changed. For example, the camera 2 in the machining plan creation support system 100 captures images of a worker during work. The action time information updater 14 in the machining plan creation support device 1 identifies the worker, the model, and the elementary action in the images described by the image information obtained from the camera 2.When the order of elementary actions performed by the worker on the model, as identified by the action time information updater 14, differs from the order of elementary actions performed on the model as described by the action definition information, the machining plan creation support device 1 notifies the user of a possible change in the elementary actions on the model.

[0054] The notification may be provided to the user, for example, by displaying a message on the screen, issuing a voice message, or sending a message to a user terminal used by the user. An editing screen for editing the action definition information may be displayed, along with a message to receive an edit from the user. The editing screen for the action definition information may display a candidate elementary action to be changed based on the sequence of elementary actions actually performed by the worker on the identified model.

[0055] A machining plan creation support process executed by the machining plan creation support device 1 will be described with reference to Fig. 10. The Fig. The machining plan creation support process shown in Figure 10 starts when, for example, the machining plan creation support device 1 is turned on. The production plan information acquirer 15 in the machining plan creation support device 1 determines whether production plan information describing a machining plan for today is input (step S11). If no production plan information is input (No in step S11), processing proceeds to step S19. If production plan information is input (Yes in step S11), the production plan information acquirer 15 stores the input production plan information in the production plan information storage 16.The personnel allocation calculator 18 executes the personnel allocation that maximizes the evaluation value for productivity for the production plan (the model and the number of units) described by the production plan information, based on the on-site worker information, the action time information, and the action definition information (step S12).

[0056] In the example in Fig. Figure 3 shows the action time information, which stores the elementary action time for each combination of the worker, the model, and the elementary action. The elementary action time described by the action time information is not a constant but a continuous probability distribution, with the elementary action time indicated by the horizontal axis and the number of times indicated by the vertical axis. Fig. 3 shows the probability distributions of the elementary action times of a combination of a worker 1, a model M1 and an elementary action E01, a combination of the worker 1, the model M1 and an elementary action E02, a combination of a worker 2, the model M1 and the elementary action E01, and a combination of the worker 2, the model M1 and the elementary action E02.

[0057] In the example of Fig. Figure 4 shows the action definition information for the types of production processes that contain the elementary actions for each model. The production process for model 1 comprises four processes, or setups: process 1, process 2, and process 3. The setup includes an elementary action E01, ..., an elementary action E0n0. Process 1 includes an elementary action E11, ..., an elementary action E1n1. Process 2 includes an elementary action E21, ..., an elementary action E2n2. Process 3 includes an elementary action E31, ..., and an elementary action E3n3.

[0058] It will be back on Fig. 10. The machining plan information generator 19 determines whether the maximum evaluation value calculated by the personnel allocation calculator 18 exceeds the previously calculated result (step S13). If the calculation result does not exceed the previous calculation result, or if no previous calculation result is stored (No in step S13), the processing proceeds to step S15. If the calculation result exceeds the previous calculation result (Yes in step S13), the machining plan information generator 19 updates the optimal solution for the personnel allocation and production flow (step S14) and determines whether the personnel allocation calculator 18 has calculated all the maximum evaluation values ​​(step S15).

[0059] If all maximum evaluation values ​​have not yet been calculated (No in step S15), the machining plan information generator 19 changes the production plan for the model (step S15). Processing returns to step S12, and the personnel allocation calculator 18 performs the personnel allocation that maximizes the evaluation value. Steps S12 to S16 are repeated. If the personnel allocation calculator 18 has calculated all maximum evaluation values ​​(Yes in step S15), the machining plan information generator 19 compares the maximum evaluation values ​​calculated by the personnel allocation calculator 18 and generates machining plan information (step S17) using the personnel allocation and the production flow that maximizes the evaluation value as an optimal solution.

[0060] The machining plan information output unit 20 outputs the machining plan information generated by the machining plan information generator 19 (step S18). If the machining plan creation support device 1 is not turned off (No in step S19), the processing returns to step S11. Steps S11 to S19 are repeated. If the machining plan creation support device 1 is turned off (Yes in step S19), the processing ends.

[0061] The action time prediction process executed by the machining plan creation support device 1 will be described with reference to Fig. 11. The Fig. The action time prediction process shown in Figure 11 starts when, for example, no past action records of workers are detected by the detector 141 in the machining plan creation support device 1. The action time prediction module 145 in the machining plan creation support device 1 identifies a person (a worker with a past action record) with the characteristics most similar to the characteristics of the new worker X.

[0062] In the example of Fig. 8, the person with characteristics most similar to the characteristics of the new worker X is a worker 4.

[0063] Reference is again made to Fig. 11. The action time prediction module 145 reads the action time information for the identified person from the action time information storage 13 (step S22). The action time prediction module 145 calculates the skill level of the identified person (step S23). The probability distribution of the elementary action time of the new worker X is calculated by dividing the probability distribution of the elementary action time of the identified person at the current time (number of actions x) by the skill level (step S24). The process then ends.

[0064] In the example of Fig. 9 describes the probability distribution of an elementary action time for the combination of the new worker X, the model M1 and the elementary action E01.

[0065] When preparing a machining plan that maximizes the evaluation value for productivity based on each worker's action time, the machining plan creation support device 1 predicts the elementary action time of a worker without a past action history based on the action time information for a worker with a past action history that has characteristics most similar to the characteristics of the worker without the past action history, and generates action time information describing the elementary action time of the worker without the past action history, in order to provide efficient action instructions to workers with unknown action times. The elementary action time is used in the probability distribution to account for varying elementary action times. Embodiment 2

[0066] In Embodiment 2, the action time prediction module 145 uses a prediction process different from the prediction process used in Embodiment 1. The other components are the same as in Embodiment 1. The elementary action time of a worker is expressed using an exponential function model with a negative exponent using Formula 6 below. y(x)=p−x+c

[0067] In the above formula, y(x) is the elementary action time at the x-th production and x ≥ 0, c is the elementary action time at the first production and c ≥ 0, and p is the skill rate, where y(x) is an elementary action time of a worker performing the x-th production and p is the skill rate of the worker to minimize the result of Formula 7. D(p)=∑i=1N(y(x)−T(x))2= ∑i=1N(p−x+c−T(x))2

[0068] The number of decimal places in the skill rate p can be determined based on the accuracy desired by the user. The calculation time generally increases as the number of decimal places increases. As in Embodiment 1, the action time prediction module 145 acquires the probability distribution of the elementary action time of the worker who has no past action record by dividing the probability distribution of the elementary action time at the current time (action number x) of the worker with the highest degree of feature similarity to the worker who has no past action record by the skill rate p.

[0069] The machining plan creation support device 1 according to Embodiment 2 increases the prediction accuracy by using the skill rate that minimizes the difference between the elementary action time calculated using the skill rate and the elementary action time based on the log to predict the elementary action time of a worker with no past action log.

[0070] In Embodiments 1 and 2, the on-site worker information generator 11 identifies a person in images described by the image information to acquire the name of the worker working on-site, or specifically, the name of a worker who is on-site, and generates worker information describing the name of the worker who is on-site. However, the information may be generated in any other way. For example, the on-site worker information generator 11 may acquire the name of a worker working on-site, or specifically, the name of a worker who is on-site, by identifying the person in a still image of the worker, and generate the on-site worker information describing the name of the on-site worker.The machining plan creation support device 1 can acquire on-site worker information in cooperation with an in-house management system that manages start and end times of each worker.

[0071] In Embodiments 1 and 2 described above, the machining plan creation support device 1 includes the action time information storage 13, the on-site worker information storage 12, and the action definition information storage 17. These storages may be included in an external device or system. The action time information storage 13 and the action time information update module 14 may be included in an action time information update device different from the machining plan creation support device 1. Similarly, the on-site worker information storage 12 and the on-site worker information generator 11 may be included in an on-site worker information generation device different from the machining plan creation support device 1.

[0072] In Embodiments 1 and 2 described above, the production plan information acquirer 15 in the machining plan creation support device 1 receives an input of production plan information describing the production plan for today's manufactured items. However, this information may be input in any other way. The user can input production plan information describing the production plan for today's manufactured items for which the machining plan is to be created into the machining plan creation support device 1.When the user inputs the date on which the machining plan is to be created (designated date) into the machining plan creation support device 1 in cooperation with an in-house production plan system that manages the production plan, the production plan information acquirer 15 can acquire production plan information describing the production plan on the designated date from the production plan system. The machining plan can be prepared in units other than one day, such as in units of half a day or two days.

[0073] Although Embodiments 1 and 2 each describe an example of a process for assisting in the creation of a processing plan for assigning actions included in a production process for manufactured items to workers, the process is not limited to operations related to the manufacture of items. The manufactured items may be any products that involve predetermined actions included in the manufacturing process.

[0074] The hardware configuration of the machining plan creation support device 1 will be described with reference to Fig. 12. As described in Fig. 12, the machining plan creation support device 1 includes a temporary memory 101, a memory 102, a calculator 103, an input unit 104, a transmitter-receiver 105, and a display 106. The temporary memory 101, the memory 102, the input unit 104, the transmitter-receiver 105, and the display 106 are connected to the calculator 103 through a bus.

[0075] The computer 103 is, for example, a central processing unit (CPU). The computer 103 executes the processing of the on-site worker information generator 11, the action time information updater 14, the personnel assignment computer 18, and the machining plan information generator 19 based on the control program stored in the memory 102.

[0076] Temporary memory 101 is, for example, random access memory (RAM). Temporary memory 101 loads the control program stored in memory 102 and is used as a work area for computer 103.

[0077] The memory 102 is a non-volatile memory, such as a flash memory, a hard disk drive, a digital versatile disc random access memory (DVD-RAM), and a rewritable digital versatile disc (DVD-RW). The memory 102 prestores programs for causing the computer 103 to execute the processing of the machining plan creation support device 1, provides the computer 103 with data stored in the program as instructed by the computer 103, and stores data provided by the computer 103. The on-site worker information memory 12, the action time information memory 13, the production plan information memory 16, and the action definition information memory 17 are included in the memory 102.

[0078] The input unit 104 includes an input device, such as a keyboard, a pointing device, and a voice input device, and an interface connecting the input device to the bus. Information input by the user is provided to the computer 103 via the input unit 104. The input unit 104 serves as the machining plan information acquirer 15. In the structure in which the action time information updater 14 displays an editing screen for editing the action definition information along with a notification and receives an edit from the user, the input unit 104 serves as the action time information updater 14.

[0079] The transceiver 105 is a network terminator or a wireless communication device connected to the network, and a serial interface or a local area network (LAN) interface connected to the network terminator or wireless communication device. In the structure where the machining plan information output unit 20 sends machining plan information to a user terminal used by users (the manager or the worker), the transceiver 105 serves as the machining plan information input unit 20.

[0080] The display 106 is, for example, a liquid crystal display (LCD) or an organic electroluminescence (EL) display device. In the structure where the machining plan information output unit 20 displays machining plan information on a screen, the transmitter-receiver 105 serves as the machining plan information output unit 20. In the structure where the action time information updater 14 displays an editing screen for editing the action definition information along with a notification and receives an edit from the user, the display 106 serves as the action time information updater 14.

[0081] The processing for the on-site worker information generator 11, the on-site worker information storage 12, the action time information storage 13, the action time information updater 14, the production plan information acquirer 15, the production plan information storage 16, the action definition information storage 17, the personnel allocation calculator 18, the machining plan information generator 19 and the machining plan information output unit 20, which are in the Fig. 2 are executed by the control program using, for example, the temporary memory 101, the calculator 103, the memory 102, the input unit 104, the transmitter-receiver 105, and the display 106 as resources.

[0082] The hardware configuration and flowchart described above are only examples and can be changed or modified as required.

[0083] The main components that perform the processing of the machining plan creation support device 1, including the calculator 103, the temporary storage 101, the memory 102, the input unit 104, the transmitter-receiver 105, and the display, can be executed by a general computer system rather than a dedicated system. For example, a computer program executable to implement the above operation can be stored in a non-transitory computer-readable recording medium such as a flexible disk, a compact disk read-only memory (CD-ROM), or a DVD-ROM for distribution. The computer program can be installed in a computer to provide the machining plan creation support device 1 that performs the above processing.The computer program may be stored in a storage device included in a server or a communication network, typically the Internet, and may be downloaded by a general computer system to implement the machining plan creation support device 1.

[0084] When the functions of the machining plan creation support device 1 are partially implemented by the operating system (OS) and application programs or by the cooperation between the OS and the application programs, parts executable by the application programs can be easily stored in a non-volatile recording medium or a storage device.

[0085] The computer program can be superimposed on a carrier wave for delivery over a communications network. For example, the computer program can be placed in a bulletin board system (BBS) on a communications network for delivery over the communications network. The above processing can be performed by activating the computer program and executing the program in the same way as other application programs under the control of the operating system.

[0086] The foregoing describes several exemplary embodiments for illustrative purposes. Although the foregoing discussion has set forth specific embodiments, those skilled in the art will recognize that changes in form and detail may be made without departing from the broader spirit and scope of the invention. Accordingly, the specification and drawings are to be regarded in an illustrative rather than a restrictive sense. This detailed description, therefore, is not to be interpreted in a limiting sense, and the scope of the invention is defined only by the appended claims, which are to include the full range of equivalence permitted by the claims.

[0087] This application claims priority to Japanese Patent Application No. 2022-079193, filed on May 13, 2022, the entire disclosure of which is hereby incorporated by reference. List of reference symbols 1 machining plan creation support device 2 cameras 11 On-site worker information generators 12 on-site worker information stores 13 Action time information memory 14 Action time information updater 15 production plan information acquirers 16 Production plan information storage 17 Action definition information storage 18 personnel allocation calculator 19 Machining plan information generator 20 Machining plan information output unit 100 Machining plan creation support system 101 temporary storage 102 memory 103 computers 104 Input unit 105 transmitter-receiver 106 Advertisement 141 Detector 142 Action time measurement module 143 updaters 144 Personal characteristic information storage 145 Action time prediction module QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] JP 2022 - 079 193

[0087]

Claims

[1] A machining plan creation support device comprising: a production plan information acquirer to acquire production plan information describing a production plan for a product; a personnel allocation calculator for calculating a personnel allocation for an elementary action corresponding to the production plan described by the production plan information based on on-site worker information, action definition information, and action time information, wherein the on-site worker information describes a worker who is on-site, the action definition information describes an elementary action included in each process in a production process for the product, the action time information describes an elementary action time required for each worker to perform the elementary action; a processing plan information generator to change a production flow, cause the personnel allocation calculator to calculate the personnel allocation, and generate processing plan information describing the personnel allocation and a production flow to maximize an evaluation value for productivity; and an action time information updater to update the action time information for a worker with a past action record based on image information describing an image of the worker, predict an elementary action time of a worker with no past action record based on the action time information for a worker with a past action record who has a feature most similar to a feature of the worker with no past action record, and generate the action time information for the worker with no past action record. [2] The machining plan creation support device according to claim 1, wherein the action time information updater comprises: a detector for detecting from the image described by the image information a worker and a type of product handled by the worker, an action time measuring module to measure an elementary action time of the worker based on the image information, an updater for updating, when the worker detected by the detector has a past action record, the action time information of the worker with the past action record based on the worker and the type of product handled by the worker detected by the detector and the elementary action time measured by the action time measuring module, and an action time prediction module for, when the worker detected by the detector has no past action history, identifying a worker with a past action history who has a feature most similar to a feature of the worker without a past action history based on worker feature information describing a feature of each worker, predicting an elementary action time of the worker without a past action history based on the action time information of the identified worker, and generating the action time information describing the elementary action time of the worker without a past action history. [3] The machining plan creation support device according to claim 2, wherein the elementary action time described by the action time information is a probability distribution of each elementary action time of each worker. [4] The machining plan creation support device according to claim 3, wherein the action time prediction module acquires the probability distribution of each elementary action time of a worker with no past action history by dividing the probability distribution of each elementary action time at a current time of a worker with a past action history who has a feature most similar to a feature of the worker with no past action history by a skill rate of the worker with the past action history who has a feature most similar to a feature of the worker with no past action history. [5] The machining plan creation support device according to any one of claims 2 to 4, wherein, when the worker with no past action history performs actions a predetermined number of times, the action time prediction module determines the worker with no past action history as a worker with a past action history and generates the action information based on the worker and the type of product handled by the worker detected by the detector and the elementary action time measured by the action time measuring module. [6] A machining plan creation support device according to any one of claims 1 to 5, further comprising: an on-site worker information generator to identify the on-site worker based on the image information and generate the on-site worker information. [7] Machining plan creation support system, comprising: a camera to take a picture of a worker; and a machining plan creation support device connected to the camera, the machining plan creation support device comprising: a production plan information acquirer to acquire production plan information describing a production plan for a product; a personnel allocation calculator for calculating a personnel allocation for an elementary action corresponding to the production plan described by the production plan information based on on-site worker information, action definition information, and action time information, wherein the on-site worker information describes a worker who is on-site, the action definition information describes an elementary action included in each process in a production process for the product, the action time information describes an elementary action time required for each worker to perform the elementary action; a processing plan information generator to change a production flow, cause the personnel allocation calculator to calculate the personnel allocation, and generate processing plan information describing the personnel allocation and a production flow to maximize an evaluation value for productivity; and an action time information updater to update the action time information for a worker with a past action record based on image information captured by the camera that describes an image of the worker, predict an elementary action time of a worker with no past action record based on the action time information for a worker with a past action record that has a feature most similar to a feature of the worker with no past action record, and generate the action time information for the worker with no past action record. [8] A machining plan creation support method implementable with a machining plan creation support device, the method comprising: Acquiring production plan information that describes a production plan for a product; Calculating a personnel allocation for an elementary action corresponding to the production plan described by the production plan information based on on-site worker information, action definition information and action time information, wherein the on-site worker information describes a worker who is on-site, the action definition information describes an elementary action included in each process in a production process for the product, the action time information describes an elementary action time required for each worker to perform the elementary action; Changing a production process, initiating the calculation of personnel allocation, and generating processing plan information describing the personnel allocation and a production process in order to maximize an evaluation value for productivity; and Updating the action time information for a worker with a past action history based on image information describing an image of the worker, predicting an elementary action time of a worker with no past action history based on the action time information for a worker with a past action history who has a feature most similar to a feature of the worker with no past action history, and generating the action time information for the worker with no past action history. [9] A program for causing a computer to perform operations which include: Calculating a personnel assignment for an elementary action corresponding to the production plan described by acquired production plan information based on on-site worker information, action definition information, and action time information, wherein the on-site worker information describes a worker who is on-site, the action definition information describes an elementary action included in each process in a production process for the product, the action time information describes an elementary action time required for each worker to perform the elementary action; Changing a production process, initiating the calculation of personnel allocation, and generating processing plan information describing the personnel allocation and a production process in order to maximize an evaluation value for productivity; and Updating the action time information for a worker with a past action history based on image information describing an image of the worker, predicting an elementary action time of a worker with no past action history based on the action time information for a worker with a past action history who has a feature most similar to a feature of the worker with no past action history, and generating the action time information for the worker with no past action history.

Images