Plan generation method and plan generation system

Abstract

A plan generation method executed by a plan generation system that includes a calculation unit and a storage unit. The storage unit holds a reference plan which is a business plan, resource information on resources allocated to a business in the reference plan, and a learning model that outputs a recommendation plan including a change recommended for the reference plan and a probable reason which is an estimated reason for the change in the recommendation plan. The plan generation method includes: a procedure in which the calculation unit generates a recommendation plan on the basis of the reference plan, the resource information, and the learning model; and a procedure in which, when a corrected plan obtained by correcting the recommendation plan and a reason for correction are inputted, the calculation unit retrains the learning model using the corrected plan and the reason for correction as correct answer data of the recommendation plan and the probable reason, respectively.

Description

Plan generation method and plan generation system Incorporation by Reference

[0001] This application claims priority from Japanese Patent Application No. 2024-150979, filed on September 2, 2024, the contents of which are incorporated herein by reference.

[0002] The present invention relates to planning operations such as procurement plans and production plans in a supply chain.

[0003] In modern manufacturing and distribution industries, production planning and other planning tasks require plans that reflect ever-changing conditions, such as worker work schedules. For example, production plans determine the production start date and production volume of a product, taking into account demand forecasts, inventory outlooks, factory production capacity, and other factors. These various plans must be appropriate to the situation. In the case of production planning, creating plans that take into account factors such as worker work schedules can maximize the efficiency of the production line, reduce defect rates, and improve throughput.

[0004] However, at manufacturing sites, information such as worker work schedules changes daily, requiring planners to keep track of and make corrections on a case-by-case basis, which requires a lot of work. For example, consider a use case in which a production plan for a certain product needs to be changed so that production is performed on a production line with many skilled workers. In this case, the planner modifies the production plan based on the skilled workers' work schedules. Japanese Patent Laid-Open Publication No. 2019-46287 (Patent Document 1) is a background technology aimed at reducing such correction work hours. Patent Document 1 discloses a method for generating a plan that learns correction history.

[0005] Patent Document 1: Japanese Patent Application Laid-Open No. 2019-46287

[0006] In Patent Document 1, "information about revisions to a plan is acquired from a predetermined device, and the relationship between the values ​​of predetermined events for plan evaluation and the revised plan is identified using a predetermined algorithm." Patent Document 1, however, generates a plan by learning only the revision history, which can result in learning that does not conform to the planner's intentions. This requires revisions each time the planner reaches the desired plan, resulting in no reduction in labor costs. For example, when a product's production line is changed, there is a reason for the change, such as "to produce the product on a production line where many skilled workers are assigned." However, current methods that learn only the revision history cannot capture such reasons for the change and instead learn information such as "the production line for the product was changed." As a result, when skilled workers are transferred during the next planning phase, the production plan may not be able to be changed to match the skilled workers' work schedules.

[0007] As a result, corrections are required for similar reasons, resulting in a problem in that the amount of work required for corrections by planners is not reduced.

[0008] Therefore, by learning not only the plan differences before and after the correction but also the reasons for the correction, the plan desired by the planner can be generated.

[0009] In order to solve at least one of the above problems, the present invention provides a plan generation method executed by a plan generation system, the plan generation system comprising an arithmetic unit and a memory unit, the memory unit holding a base plan which is a plan for a business, resource information regarding resources allocated to the business in the base plan, a recommended plan including recommended changes to the base plan, and a learning model which outputs an estimated reason which is an estimate of the reason for the change in the recommended plan, the plan generation method including: a first step in which the arithmetic unit generates the recommended plan based on the base plan, the resource information, and the learning model; and a second step in which, when a revised plan which modifies the recommended plan and the reason for the revision are input, the arithmetic unit re-trains the learning model using the revised plan and the reason for the revision as correct data for the recommended plan and the estimated reason, respectively.

[0010] According to one aspect of the present invention, a plan that takes into account the intentions of a planner can be generated, thereby realizing the output of a plan that the planner desires, thereby reducing the number of steps required for planning work.

[0011] Problems, configurations, and effects other than those described above will become apparent from the following description of the embodiments.

[0012] 1 is a block diagram showing the functional configuration of the automatic plan generation system of the present embodiment; FIG. 2 is a flowchart explaining the operation of a calculation unit of the automatic plan generation system of the present embodiment; FIG. 3 is an explanatory diagram of supply chain master information held by the automatic plan generation system of the present embodiment; FIG. 4 is an explanatory diagram of human resource information held by the automatic plan generation system of the present embodiment; FIG. 5 is an explanatory diagram of plan information held by the automatic plan generation system of the present embodiment; FIG. 6 is an explanatory diagram of correction information held by the automatic plan generation system of the present embodiment; FIG. 7 is an explanatory diagram of reason score information held by the automatic plan generation system of the present embodiment; FIG. 8 is an explanatory diagram of reason item relation information held by the automatic plan generation system of the present embodiment; FIG. 9 is an explanatory diagram of information displayed by the display unit of the automatic plan generation system of the present embodiment; FIG. 10 is a flowchart explaining the operation of a generation unit of the automatic plan generation system of the present embodiment; FIG. 11 is a flowchart explaining the operation of a learning unit of the automatic plan generation system of the present embodiment; FIG. 12 is a flowchart explaining the operation of a correction difference calculation unit of the automatic plan generation system of the present embodiment.

[0013] Examples will be described below, but the scope of the present invention is not limited to these examples.

[0014] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0015] FIG. 1 is a block diagram showing the functional configuration of an automatic plan generation system 100 according to this embodiment.

[0016] Focusing on functions, the automatic plan generation system 100 has a storage unit 110, a display unit 120, a calculation unit 130, and a communication unit 140. Information required by the storage unit 110, the display unit 120, the calculation unit 130, and the communication unit 140 can be acquired via a communication network 200 or the like from, for example, an enterprise resource planning (ERP) system, a database that stores data equivalent thereto, or a simple storage device.

[0017] The storage unit 110 includes an input information storage unit 111, a plan storage unit 112, a correction information storage unit 113, a reason score storage unit 114, and a reason item relation storage unit 115. The storage unit 110 is configured by a storage device such as a hard disk drive or a flash memory device, for example.

[0018] The display unit 120 includes a plan display unit 121 , a revised plan receiving unit 122 , an estimated reason display unit 123 , and a revision reason receiving unit 124 .

[0019] The calculation unit 130 includes a generation unit 131 , a learning unit 132 , and a correction difference calculation unit 133 .

[0020] The communication unit 140 is an interface device for communicating with an external device, the user terminal device 300, via the communication network 200. The communication unit 140 may use either wired communication or wireless communication.

[0021] 2 is a flowchart illustrating the operation of the calculation unit 130 of the automatic plan generation system 100 of this embodiment. Each step in FIG. 2 will be described below.

[0022] (FIG. 2: Step S101: Acquire Master Plan and Input Information) Step S101 in FIG. 2 is a process of acquiring the master plan 1211, supply chain master information 11110, and human resources information 11120 used for plan generation performed by the generation unit 131. The master plan 1211, supply chain master information 11110, and human resources information 11120 are input via the communication network 200.

[0023] FIG. 5 is an explanatory diagram of the plan information 1120 held by the automatic plan generation system 100 of this embodiment.

[0024] The plan information 1120 holds the base plan 1211 acquired in step S101, as well as a recommended plan 1212 and a revised plan 1221, which will be described later. Here, the base plan 1211 will be described.

[0025] The master plan 1211 is a plan input from another planning system to the automatic plan generation system 100, among the plans held as the plan information 1120. As shown in FIG. 5 , the plan information 1120 holds a plan ID 1121, a plan date 1122, a scheduled date 1123, an item ID 1124, a planned quantity 1125, a performance value 1126, a target plan 1127, a plan type 1128, and attribute information 1129.

[0026] The plan ID 1121 is a primary key that uniquely identifies the plan. The plan date 1122 is the date when the plan was created. The scheduled date 1123 is the date when the plan will be executed. The item ID 1124 identifies the item that is the subject of the plan. The planned quantity 1125 is the planned quantity (for example, in the case of a production plan, the planned production quantity of the subject item). The actual value 1126 is the actual value for the planned quantity 1125 (for example, in the case of a production plan, the quantity of the item that was actually produced according to the plan).

[0027] The target plan 1127 indicates the target of the plan, such as a production plan, procurement plan, or sales plan. The plan type 1128 indicates whether the plan is the master plan 1211, a recommended plan 1212 (described later), or a revised plan 1221. The attribute information 1129 indicates plan target information, such as the line name for a production plan, the supplier name for a procurement plan, or the buyer name for a sales plan. The master plan 1211 is stored in the plan storage unit 112. At this time, the plan type 1128 is input as "master plan."

[0028] 3 and human resource information 11120 are information for generating a recommended plan 1212 and an estimated reason score 1231 from a base plan 1211 using the plan automatic generation system 100. The supply chain master information 11110 and the human resource information 11120 are stored in the input information storage unit 111.

[0029] FIG. 3 is an explanatory diagram of the supply chain master information 11110 held by the automatic plan generation system 100 of this embodiment.

[0030] The supply chain master information 11110 holds, for example, a base ID 11111, a base name 11112, an item ID 11113, an item name 11114, a shipping source 11115, and a shipping destination 11116. This information can be obtained from a system such as an ERP.

[0031] The base ID 11111 and base name 11112 identify the bases of the businesses that make up the supply chain. The item ID 11113 and item name 11114 identify the items handled at each base. The transport source 11115 and transport destination 11116 indicate the transport source and destination of the goods when the business is a transporter.

[0032] FIG. 4 is an explanatory diagram of the human resource information 11120 held by the automatic plan generation system 100 of this embodiment.

[0033] The human resources information 11120 holds, for example, an employee ID 11121 , years of service 11122 , qualification A held 11123 , evaluation 11124 , skill level classification 11125 , person in charge 11126 , and overtime hours worked in the current month 11127 .

[0034] The information on employee ID 11121, years of service 11122, qualification A holding 11123, and evaluation 11124 can be obtained from a system (not shown) such as an attendance management system. The employee ID 11121 identifies personnel (e.g., employees of each business) registered in the human resources information 11120. The years of service 11122 indicates the length of service of each employee. The qualification A holding 11123 indicates whether each employee holds a certain qualification (in this example, "qualification A"). While this example indicates whether only one qualification is held, the human resources information 11120 may also include information indicating whether each of multiple qualifications is held. The evaluation 11124 indicates the evaluation of each employee. For example, the evaluation 11124 is an evaluation value by a user of the system, and may also be an evaluation value by another person (such as each employee's supervisor) regarding each employee's overall ability or work attitude.

[0035] The proficiency level classification 11125 is an index showing whether each employee is a beginner or an expert. The proficiency level classification 11125 may be determined from information on the years of service 11122, whether qualification A is held 11123, and the evaluation 11124, or may be acquired from an external human resources management system (not shown), etc. For example, the years of service 11122 is set to "1" if it is five years or more, "0" if it is less than five years, "1" if it is qualification A held 11123, and "0" if it is not held, and the evaluation 11124 is input as a real number ranging from "0" to "1."

[0036] If the average of each element (years of service 11122, holding qualification A 11123, evaluation 11124) exceeds "0.5", the person is considered to be an expert, and the proficiency category 11125 is set to "1". Conversely, if the average is less than "0.5", the person is considered to be a beginner, and the proficiency category 11125 is set to "0". Take as an example a case where an employee has been with the company for 8 years and 2 months, holds qualification A, and has an evaluation of "0.7". Taking into account that the employee has been with the company for more than 5 years, so it is set to "1", that the employee holds qualification A, so it is set to "1", and that the evaluation is "0.9", the proficiency category 11125 is calculated by taking the average of these, which is (1 + 1 + 0.7) / 3 = 0.9, so it is 0.5 or more, and the proficiency category 11125 of this person in question can be determined to be "1".

[0037] The person in charge 11126 indicates the scope of work that each employee is responsible for. For example, the person in charge 11126 is input with the line that the employee is responsible for in the case of a production plan, the supplier that the employee is responsible for in the case of a procurement plan, or the buyer that the employee is responsible for in the case of a sales plan. The person in charge 11126 can also be acquired from a system such as a human resources master.

[0038] The current month overtime hours 11127 indicates the overtime hours worked by each employee in the current month. For example, the current month overtime hours 11127 indicates the total overtime hours worked from the beginning of the month to the current time, including the current time.

[0039] (FIG. 2: Step S102: Generating a recommendation plan and an estimated reason score) Step S102 in FIG. 2 is a process performed by the generation unit 131. The generation unit 131 generates a recommendation plan 1212 and an estimated reason score 1231 from the master plan 1211, supply chain master information 11110, and human resources information 11120 acquired in step S101. A specific generation flow of the recommendation plan 1212 and the estimated reason score 1231 is shown in FIG. 10.

[0040] 10 is a flowchart illustrating the operation of the generation unit 131 of the automatic plan generation system 100 of this embodiment. Each step in FIG. 10 will be described below.

[0041] (FIG. 10: Step S1311: Read Master Plan and Input Information) In step S1311 of FIG. 10, the generation unit 131 acquires the master plan 1211, supply chain master information 11110, human resources information 11120, etc. from the plan storage unit 112 and the input information storage unit 111.

[0042] 5 , the master plan 1211 is obtained by referring to the master plan date 1122 and the target plan 1127. For example, in the case of a production plan, a plan in which the master plan date 1122 is the execution date of this system and the target plan 1127 is the production plan is obtained from the plan storage unit 112.

[0043] In addition, in the case of the supply chain master information 11110 in Figure 3, information filtered using the base ID 11111 and item ID 11113 is read from the input information storage unit 111, and in the case of the human resources information 11120 in Figure 4, information filtered using the employee ID 11121 is read from the input information storage unit 111.

[0044] (FIG. 10: Step S1312: Generate a recommendation plan and an inference reason score from the base plan and input information using the model generated by the learning unit) Step S1312 in FIG. 10 is a process performed by the generation unit 131. The generation unit 131 generates a recommendation plan 1212 and an inference reason score 1231 based on the base plan 1211, supply chain master information 11110, and human resources information 11120 received in step S1311, and the learning model generated by the learning unit 132.

[0045] The recommended plan 1212 is a plan obtained by modifying the base plan 1211 based on the supply chain master information 11110 and human resource information 11120. The recommended plan 1212 has the information shown in FIG. 5 , just like the base plan 1211. "Recommended plan" is entered as the plan type 1128 for the recommended plan among the plans included in the plan information 1120. Hereinafter, the various components of the display unit 120 in FIG. 9 will be described in conjunction with each other.

[0046] FIG. 9 is an explanatory diagram of information displayed by the display unit 120 of the automatic plan generation system 100 of this embodiment.

[0047] The estimated reason score 1231 displayed in the estimated reason display unit 123 in Fig. 9 is information indicating the reason for changing from the base plan 1211 to the recommended plan 1212, and may include, for example, at least one of information describing the reason itself and a score indicating the importance of the reason. The estimated reason score 1231 is held as reason score information 1140 in the reason score storage unit 114. Similarly, a modification reason score 1241, which will be described later, is also held as reason score information 1140. Here, the reason score information 1140 will be described with reference to Fig. 7.

[0048] FIG. 7 is an explanatory diagram of the reason score information 1140 held by the automatic plan generation system 100 of this embodiment.

[0049] The reason score information 1140 includes, for example, a plan ID 1141, an item 1142, a reason score 1143, and a reason score type 1144. The plan ID 1141 identifies the generated plan (for example, a recommended plan or a revised plan). The item 1142 is information indicating the reason for changing the plan, such as consideration of proficiency, reducing the number of changeovers, or leveling out overtime hours. The reason score 1143 is a score representing the importance of the item 1142. The reason score type 1144 is information indicating whether the reason score is the estimated reason score 1231 or the revised reason score 1241.

[0050] That is, among the reason score information 1140, information in which the plan ID 1141 is the identification information of the recommendation plan 1212 displayed in the plan display section 121 shown in Figure 9 and the reason score type 1144 is "estimated reason score" is displayed in the estimated reason display section 123 as the estimated reason score 1231.

[0051] 9 or the value of the estimated reason score 1231 corrected thereby is held as information in the reason score information 1140, in which the plan ID 1141 is identification information of the correction plan 1221 displayed in the correction plan receiving unit 122 shown in Fig. 9 and the reason score type 1144 is "correction reason score." The correction reason score 1241, the details of which will be described later, is the reason score 1143 of the item 1142 that the user (e.g., the planner) considered important when correcting the recommendation plan 1212 generated by the automatic plan generation system 100.

[0052] The estimated reason score 1231 is generated based on the correction reason score 1241. An example of the generation method is shown below.

[0053] First, only the correction reason scores are extracted from the reason score type 1144. Next, the average of the reason scores 1143 is calculated for each item listed in the item 1142. For example, suppose that when only records in which "overtime leveling" is entered in the item 1142 are extracted, there are two records. Next, if the reason scores 1143 of the two records are assumed to be 3 and 5, the average is 4, which is used as the importance of the estimated reason score for overtime leveling. Similarly, the importance is calculated for each item, such as the importance of the estimated reason score for taking proficiency into consideration and the importance of the estimated reason score for reducing the number of setup changes.

[0054] The method for generating the recommendation plan 1212 is described below.

[0055] The generation unit 131 selects the item 1142 with the highest importance of the estimated reason score 1231, modifies the base plan 1211, and outputs a recommended plan 1212. For example, assume that the item "leveling overtime hours" has the highest importance. In the base plan 1211 of FIG. 9 , the overtime hours for Line A on Day 1 are -0.5 hours and the overtime hours on Day 2 are 2.0 hours. In this case, the automatic plan generation system 100 generates a recommended plan 1212 by moving 25 of the 75 units of item B produced on Day 2 to Day 1 so that the overtime hours on Day 1 and Day 2 are both 1.0 hour. As a result, the number of setup changes on Day 1 increases by one, but the overtime hours on Day 1 and Day 2 are 1.0 hour. This generates a recommended plan 1212 with modifications that prioritize leveling overtime hours over reducing the number of setup changes.

[0056] (FIG. 10: Step S1313: Store recommendation plan and estimated reason score in plan storage unit and reason score storage unit) Step S1313 in FIG. 10 is processing performed by the generation unit 131. The plan storage unit 112 stores the recommendation plan 1212 generated in step S1312 as plan information 1120. At this time, "recommended plan" is input into the plan type 1128. The reason score storage unit 114 assigns a new ID to the recommendation plan generated in step S1312 and stores the ID in the plan ID 1141, stores the generated estimated reason score item in item 1142, and stores the reason score in the reason score 1143. At this time, "estimated reason score" is input into the reason score type 1144.

[0057] (FIG. 2: Step S103: Displaying the Base Plan and the Recommended Plan to the User) Step S103 in FIG. 2 is a process performed by the plan display unit 121. The plan display unit 121 displays the base plan 1211 acquired in step S101 and the recommended plan 1212 generated in step S102 to the user.

[0058] For display in the master plan 1211, plans whose plan type 1128 is master plan are extracted from the plan information 1120 in the plan storage unit 112. Furthermore, filtering is performed using the plan date 1122, the scheduled date 1123, and the target plan 1127 to identify the plan to be displayed. Figure 9 shows a case where the target plan 1127 is a production plan.

[0059] Line information is obtained from attribute information 1129. Product name information is obtained from item name 11114 of a record in which the item ID 1124 in the plan information 1120 matches the item ID 11113 in the supply chain master information 11110. The production volume for each day shown in the master plan 1211 can be obtained from the planned quantity 1125 of a record in which the scheduled date 1123 in the plan information 1120 shown in FIG. 5 corresponds to each day of the master plan 1211.

[0060] In the above example, the target plan 1127 is a production plan, but if it is a sales plan, information such as the base name 11112 of a record in which the attribute information 1129 of the plan information 1120 (FIG. 5) matches the base ID 11111 of the supply chain master information 11110 (FIG. 3) is searched for and displayed. Similarly, if it is a transportation plan, information on the transportation source 11115 and transportation destination 11116 of the supply chain master information 11110 (FIG. 3) is searched for and displayed.

[0061] The recommended plan 1212 is displayed by extracting plans whose plan type 1128 is "recommended plan" from the plan information 1120. The extraction method is the same as that for the base plan 1211 described above.

[0062] (FIG. 2: Step S104: Displaying Estimated Reason Score to User) Step S104 in FIG. 2 is a process performed by the estimated reason display unit 123. The estimated reason display unit 123 displays the estimated reason score generated in step S102 to the user.

[0063] As shown in Fig. 9, the estimated reason score is displayed to the user in the estimated reason score 1231 of the estimated reason display section 123 of the display unit 120. As the estimated reason item, items 1142 of the reason score information 1140 (Fig. 7) such as leveling out overtime hours, taking proficiency into consideration, reducing the number of setup changes, etc. are displayed to the user. As the importance, the reason score 1143 of the reason score information 1140 (Fig. 7) is displayed. In Fig. 9, the importance is displayed using numeric radio buttons, but this can be replaced with a slider display, a display in natural language, etc.

[0064] (FIG. 2: Step S105: The revised plan receiving unit receives a revised plan from the user and stores it in the plan storage unit) Step S105 in FIG. 2 is a process performed by the revised plan receiving unit 122 and the plan storage unit 112. The revised plan receiving unit 122 receives a revised plan 1221 from the user who has confirmed the base plan 1211 and the recommended plan 1212 displayed in step S103. Specifically, the revised plan receiving unit 122 of the display unit 120 in FIG. 9 receives the revised plan 1221 from the user. Furthermore, the received revised plan 1221 is stored in the plan storage unit 112 as plan information 1120.

[0065] 5, the revised plan 1221 holds, for example, a plan ID 1121, a planned date 1122, a scheduled date 1123, an item ID 1124, a planned quantity 1125, an actual result value 1126, a target plan 1127, a plan type 1128, and attribute information 1129. A new ID number that does not overlap with other records is input into the plan ID 1121. The date of receipt of the revised plan 1221 is input into the plan date 1122. The target plan date of the revised plan 1221 is input into the scheduled date 1123. "Revised plan" is input into the plan type 1128. Plan target information, such as the line name for a production plan, the supplier name for a procurement plan, or the buyer name for a sales plan, is input into the attribute information 1129.

[0066] Furthermore, the modification information storage unit 113 stores modification information 1130 .

[0067] FIG. 6 is an explanatory diagram of the revision information 1130 held by the automatic plan generation system 100 of this embodiment.

[0068] As shown in Figure 6, the revision information storage unit 113 of the plan automatic generation system 100 holds revision information 1130 consisting of a revision ID 1131, a revision date and time 1132, a pre-change plan ID 1133, and a post-change plan ID 1134. An ID that uniquely identifies the revision information 1130 is newly assigned and input as the revision ID 1131. The date and time of assignment is recorded as the revision date and time 1132. The plan ID of the recommended plan 1212 is extracted from the plan ID 1121 in the plan storage unit 112 and input as the pre-change plan ID 1133. The plan ID of the revised plan 1221 is extracted from the plan ID 1121 in the plan storage unit 112 and input as the post-change plan ID 1134.

[0069] (FIG. 2: Step S106: Revision reason acceptance unit accepts revision reason score from user and stores it in reason score storage unit) Step S106 in FIG. 2 is a process performed by the revision reason acceptance unit 124 and the reason score storage unit 114. The revision reason acceptance unit 124 accepts an intention to revise the revision plan 1221 from the user who input the revision plan 1221 in step S105. Specifically, the revision reason acceptance unit 124 of the display unit 120 in FIG. 9 accepts the revision reason score 1241 from the user. Furthermore, the accepted revision reason score 1241 is stored in the reason score storage unit 114.

[0070] 7 , the modification reason score 1241 holds a plan ID 1141, an item 1142, a reason score 1143, and a reason score type 1144. In the plan ID 1141, the plan ID 1121 of the modified plan 1221 evaluated by the modification reason score 1241 is input. In the reason score type 1144, "modification reason score" is input.

[0071] The method of receiving and storing the items 1142 and reason scores 1143 will be described below.

[0072] The method of accepting the items 1142 and the reason score 1143 is to accept input of the importance of the revision reason score 1241 as a positive or negative integer centered around 0, in the same way as the revision reason accepting unit 124 (FIG. 9), in order to accept the degree of revision. For example, if one wishes to increase the importance of the reason "taking proficiency into consideration" the most, one would enter "+3" as the importance of taking proficiency into consideration, and if one wishes to decrease the importance of the reason "leveling overtime hours," one would enter a negative number such as "-2."

[0073] 9, input is accepted using numeric radio buttons, but this may be replaced with input using a slider or input in natural language using generation AI, etc. It is also possible to accept new revision reasons from the user in free text.

[0074] The item 1142 and the reason score 1143 are stored by taking the sum of the importance of the correction reason score 1241 received by the correction reason receiving unit 124 ( FIG. 9 ) and the importance of the estimated reason score, and storing the result as the item 1142 and the reason score 1143 in the reason score storage unit 114. For example, consider a case where the importance of the reason for proficiency consideration in the estimated reason score is "0," and the importance of proficiency consideration in the correction reason score 1241 received from the user is "+3." In this case, the reason score 1143 of the reason score information 1140 stored in the reason score storage unit 114 is stored as "3," since 0 + 3 = 3.

[0075] In the above example, the user inputs an increase or decrease in the importance of the estimated reason score, and the result of addition or subtraction based on that is stored as the correction reason score. However, this input method is just one example, and the user may directly input the changed importance (i.e., the correction reason score itself to be stored in the reason score storage unit 114).

[0076] (FIG. 2: Step S107: Re-learning using recommendation plan, revision plan, estimated reason score, and revision reason score as input) Step S107 in FIG. 2 is processing performed by the learning unit 132 and the revision difference calculation unit 133. The revision difference calculation unit 133 calculates the difference between the revision plan 1221 received from the user in step S105 and the recommendation plan 1212 generated in step S102, and the difference between the revision reason score 1241 received from the user in step S106 and the estimated reason score 1231 generated in step S102. A specific revision difference calculation flow will be described later.

[0077] The learning unit 132 generates a learning model from the difference between the revised plan 1221 and the recommended plan 1212 obtained from the revision difference calculation unit 133 and the difference between the revision reason score 1241 and the estimation reason score 1231. A specific learning flow will now be described (see FIG. 11 ). The base plan 1211, supply chain master information 11110, and human resource information 11120 may be used for re-learning.

[0078] 11 is a flowchart illustrating the operation of the learning unit 132 of the automatic plan generation system 100 of this embodiment. Each step in FIG. 11 will be described below.

[0079] (FIG. 11: Step S1321: Acquire recommendation plan, revision plan, estimated reason score, and revision reason score from plan storage unit and reason score storage unit) Step S1321 in FIG. 11 is processing performed by the plan storage unit 112 and the reason score storage unit 114. The plan storage unit 112 acquires the recommendation plan 1212 and the revision plan 1221. Specifically, it extracts records whose plan type 1128 is the recommendation plan and records whose plan type is the revision plan from the plan information 1120 in FIG. 5. The reason score storage unit 114 acquires the estimated reason score 1231 and the revision reason score 1241. Specifically, it extracts records whose reason score type 1144 is the estimated reason score and records whose reason score type is the revision reason score 1241 from the reason score information 1140 in FIG. 7.

[0080] (FIG. 11: Step S1322: The difference between the recommendation plan and the correction plan, and the difference between the estimated reason score and the correction reason score are obtained from the correction difference calculation unit) Step S1322 in FIG. 11 is processing performed by the correction difference calculation unit 133. The correction difference calculation unit 133 obtains the difference between the recommendation plan 1212 and the correction plan 1221, and the difference between the estimated reason score 1231 and the correction reason score 1241, according to the flow shown in FIG. 12, and passes them to the learning unit 132. A specific correction difference calculation flow will be described later with reference to FIG. 12.

[0081] (FIG. 11: Step S1323: Generate a learning model by learning from the difference between the recommendation plan and the revision plan and the difference between the estimated reason score and the revision reason score) Step S1323 in FIG. 11 is processing performed by the learning unit 132. The learning unit 132 can be omitted when calculating the estimated reason score based on the reason score information 1140 received in the revision reason score 1241, as in step S1312 described in this embodiment, and calculating the importance to be displayed in the recommendation plan 1212 and the estimated reason score 1231.

[0082] On the other hand, as a means other than the method of calculating the importance described in this embodiment, a technique such as machine learning may be applied. In this case, the reason item relation information 1150 corresponding to each item of the base plan 1211 and the correction reason score 1241 presented to the user is used as input, the recommended plan 1212 and the estimated reason score and importance displayed in the estimated reason score 1231 are used as output, and the output, the corrected plan 1221 corrected by the user, and the correction reason score 1241 entered by the user are used as correct labels for learning, thereby generating a learning model.

[0083] FIG. 8 is an explanatory diagram of the reason item relation information 1150 held by the automatic plan generation system 100 of this embodiment.

[0084] 8 , the reason item relation storage unit 115 of the automatic plan generation system 100 holds reason item relation information 1150 consisting of items 1151 and input information 1152. The items 1151 correspond to the items of the reasons included in the revision reason score 1241. The input information 1152 indicates information held by the automatic plan generation system 100 that corresponds to each item of the reason included in the revision reason score 1241.

[0085] For example, item 1151 stores overtime leveling, proficiency consideration, reduction in the number of setup changes, etc. Of these, input information 1152 corresponding to overtime leveling stores the current month overtime hours 11127 of the human resources information 11120. This indicates that overtime leveling is related to the current month overtime hours 11127 of the human resources information 11120.

[0086] When machine learning is used as described above, in step S1312, a trained model is used to input information indicated by input information 1152 of reason item relationship information 1150 corresponding to each item of base plan 1211 and correction reason score 1241, and the estimated reason score and importance to be displayed in recommendation plan 1212 and estimated reason score 1231 are output, and are displayed in recommendation plan 1212 and estimated reason score 1231, respectively.

[0087] 12 is a flowchart illustrating the operation of the correction difference calculation unit 133 of the automatic plan generation system 100 of this embodiment. Each step in FIG. 12 will be described below.

[0088] (FIG. 12: Step S1331: In response to a call from the learning unit, the recommended plan, the revised plan, and the estimated reason score, and the revision reason score are acquired from the plan memory unit and the reason score memory unit) Step S1331 in FIG. 12 is a process performed by the plan memory unit 112 and the reason score memory unit 114. The plan memory unit 112 filters and acquires the recommended plan 1212 and the revised plan 1221, for which a difference is to be calculated, using the planned date 1122, the scheduled date 1123, and the target plan 1127 of the plan memory unit 112. The reason score memory unit 114 filters and acquires the estimated reason score 1231 and the revision reason score 1241, for which a difference is to be calculated, using the plan ID 1141, the item 1142, the reason score 1143, and the reason score type 1144 from the reason score memory unit 114.

[0089] (FIG. 12: Step S1332: Calculating the difference between the recommendation plan and the revised plan, and the difference between the estimated reason score and the revision reason score) Step S1332 in FIG. 12 is processing performed by the revision difference calculation unit 133. The revision difference calculation unit 133 calculates the difference between the recommendation plan 1212 and the revised plan 1221 acquired in step S1331, and the difference between the estimated reason score 1231 and the revision reason score 1241. Specifically, the difference is found by subtracting one numerical value from the other.

[0090] In calculating the difference between the recommendation plan 1212 and the revised plan 1221, the difference in the planned amount 1125 of the record extracted in step S1331 is calculated from the plan storage unit 112. In calculating the difference between the estimated reason score 1231 and the revision reason score 1241, the difference in the reason score 1143 of the record extracted in step S1331 is calculated from the reason score storage unit 114.

[0091] (FIG. 12: Step S1333: Output the difference between the recommendation plan and the revised plan, and the difference between the inferred reason score and the revision reason score to the learning unit) Step S1333 in FIG. 12 is processing performed by the revision difference calculation unit 133. The revision difference calculation unit 133 outputs the difference between the recommendation plan 1212 and the revised plan 1221, calculated in step S1332, and the difference between the inferred reason score 1231 and the revision reason score 1241, to the learning unit 132.

[0092] Here, examples of the above processing, including the base plan 1211 obtained in step S101 of Figure 2, the recommended plan 1212 and estimated reason score 1231 generated in step S102, the revised plan 1221 received from the user in step S105, and the revised reason score 1241 received from the user in step S106, will be described with reference to Figure 9.

[0093] The plan display section 121 of the display unit 120 displays a base plan 1211 and a recommended plan 1212. The base plan 1211 shown in FIG. 9 is an example of the base plan acquired in step S101 of FIG. 2 and shows a planned production volume for each item on each production line of the factory by date. The example of FIG. 9 shows plans for the daily production volumes of items A, B, C, and D on lines A and B from the 1st to the 4th. An actual factory may have more production lines, may produce more items, and may have production volumes planned for the 5th and subsequent days. However, for the sake of explanation, the above-mentioned plans for two lines, four items, and four days are shown here.

[0094] According to the master plan 1211 in Figure 9, line A is scheduled to produce 25 units of item A per day. This corresponds to one record of the plan information 1120 shown in Figure 5. That is, the plan corresponds to a record in which the scheduled date 1123 is "1st" (e.g., February 1, 2020), the item ID 1124 is "Item A" (i.e., item A), the planned quantity 1125 is "25", the target plan 1127 is "production plan", the plan type 1128 is "master plan", and the attribute information 1129 is "line A".

[0095] Here, by referring to the human resources information 11120 shown in FIG. 4, the number of employees in charge of line A and the proficiency level of each employee can be identified, and based on this information, the overtime hours required to produce 25 units of item A in one day on line A can be calculated, and this overtime hours (for example, -0.5 hours) is displayed and included in the master plan 1211.

[0096] 9, line A produces 75 units of item B, 50 units of item B, and 50 units of item B on the 2nd, 3rd, and 4th days, respectively, resulting in overtime hours of 2.0 hours, 1.0 hours, and 1.0 hours. On the other hand, line B produces 60 units of item C, 40 units of item D, 70 units of item D, and 30 units of item B on the 1st, 2nd, 3rd, and 4th days, respectively, resulting in overtime hours of 1.4 hours, 0.0 hours, 1.0 hours, and -1.0 hours.

[0097] 9 is an example of a recommendation plan generated in step S102 based on the base plan 1211 and input information (including, for example, the human resources information 11120 shown in FIG. 3). Here, the changes made to the recommendation plan 1212 from the base plan 1211 will be described.

[0098] First, 25 of the 75 units of item B on line A on day 2 are changed to be produced on line A on day 1. As a result, the overtime hours on line A are leveled out to 1.0 hour on each day from the 1st to the 4th. However, because line A produces two items, item A and item B, on the same day, the number of setup changes increases.

[0099] Similarly, 10 of the 60 units of item C produced on line B on the 1st are now produced on line B on the 2nd. Furthermore, 20 of the 70 units of item D produced on line B on the 3rd are now produced on line B on the 4th. As a result, the overtime hours on line B are leveled out to 1.0 hour on each of the 1st through 4th days. Instead, line B produces two items, item C and item D, on the 2nd, and two items, item B and item D, on the 4th, which increases the number of setup changes.

[0100] The estimated reason display section 123 displays, as the reason for the change, that priority was placed on leveling out overtime hours, and not on employee proficiency and reducing the number of setup changes.

[0101] The modification plan receiving unit 122 displays a modification plan 1221. The modification plan 1221 shown in Fig. 9 is an example of the modification plan received from the user in step S105. Here, the parts of the modification plan 1221 that have been modified from the recommended plan 1212 will be described.

[0102] First, all 25 units of item A on line A for one day are revised to be produced on line B for one day. Furthermore, 25 of the 30 units of item B on line B for four days are added to the production volume of line A for one day, and 5 units are added to the production volume of line A for four days, 25 of the 50 units of item C on line B for one day are added to the production volume of line B for two days, and 25 of the 40 units of item D on line B for two days are revised to be added to the production volume of line B for four days.

[0103] This revised plan 1221 is an example in which the user has revised the plan by placing the greatest importance on employee proficiency, because item A tends to have a high defect rate, while many highly skilled employees are assigned to line B, and by moving production of item A to line B and instead moving production of other items to line A. As a result, the importance of leveling overtime hours has decreased relatively, and as a result, overtime hours on line A for the four days have increased to 1.2 hours, while overtime hours on line B for the four days have decreased to 0.5 hours. Such revision reasons are input into the revision reason receiving unit 124, as described above.

[0104] If the learning unit 132 were to simply re-learn using the corrected plan 1221 as the correct data for the recommendation plan 1212 without taking into account the inference reason and the reason for correction, it would learn that item A should be produced on line B, and it is thought that thereafter a recommendation plan for producing item A on line B would be more likely to be generated. In contrast, in the present invention, by re-learning using the corrected plan 1221 and the correction reason score 1241 as the correct data for the recommendation plan 1212 and the inference reason score 1231, respectively, it is expected that a recommendation plan for producing item A on a production line with many skilled workers will be more likely to be generated. This makes it easier to generate a plan that takes into account the intentions of the planner, and reduces the man-hours required for planning work.

[0105] Although the above embodiment has mainly described production planning, the present invention is not limited to production planning and can be applied to any business plan that constitutes a supply chain, such as procurement planning, inventory planning, transportation planning, sales planning, etc. Furthermore, the present invention can also be applied to any business plan, such as a railway operation plan, regardless of its relevance to the supply chain.

[0106] The employees registered in the human resources information 11120 and the production lines for which each employee is responsible are examples of resources such as personnel and equipment that are allocated to a business plan. For example, in a transportation plan, transportation vehicles and their drivers correspond to resources, and in a railway operation plan, railway vehicles and their drivers correspond to resources.

[0107] In plans such as master plans, resource allocation and output volume are planned for each period (for example, each date). The output volume may be the production volume for each item in the case of a production plan, the transport volume in the case of a transportation plan, or the number of passengers to be transported in the case of a railway operation plan.

[0108] The system according to the embodiment of the present invention may be configured as follows, for example.

[0109] (1) A plan generation method executed by a plan generation system (e.g., an automatic plan generation system 100), the plan generation system including a calculation unit (e.g., a calculation unit 130) and a storage unit (e.g., a storage unit 110), the storage unit storing a master plan (e.g., a master plan 1211 included in plan information 1120) that is a plan for a business, resource information (e.g., human resource information 11120) regarding resources allocated to the business in the master plan, a recommendation plan (e.g., a recommendation plan 1212) that includes recommended changes to the master plan, and an estimation reason (e.g., an estimation reason score 123) that is an estimation of a reason for the change in the recommendation plan. and a learning model (e.g., a learning model used by the generation unit 131) that outputs the reference plan, the resource information, and the learning model. The plan generation method includes: a first procedure (e.g., step S102) in which the calculation unit generates the recommendation plan based on the reference plan, the resource information, and the learning model; and a second procedure (e.g., steps S105 to S107) in which, when a revised plan (e.g., revised plan 1221) obtained by correcting the recommendation plan and a reason for the revision (e.g., revision reason score 1241) are input, the calculation unit re-trains the learning model using the revised plan and the reason for the revision as correct data for the recommendation plan and the reason for the revision, respectively.

[0110] This allows the generation of a plan based on the intentions of the planner, thereby realizing the output of the plan that the planner desires, and reducing the man-hours required for planning work.

[0111] (2) A plan generation method according to (1) above, wherein the reason for estimation includes one or more reason items and a score indicating the importance of each item, and the reason for correction includes the amount of correction to the importance of each item included in the reason for estimation, or the importance after correction.

[0112] This makes it easier to understand and input the intent of the plan.

[0113] (3) The plan generation method described in (2) above, wherein the resources include personnel (e.g., employees of a factory, etc.) and equipment (e.g., a production line), the master plan includes the allocation of the resources for each period in a business (e.g., a manufacturing industry) that constitutes a supply chain and the output volume of the business (e.g., the planned production volume for each item), the resource information includes information indicating the proficiency of the personnel (e.g., proficiency category 11125), information regarding the working hours of the personnel (e.g., overtime hours for the current month 11127), and information indicating the equipment for which the personnel is responsible (e.g., responsibility 11126), and the reasons for estimation and the reasons for correction include at least one of reasons related to the proficiency of the personnel (e.g., consideration of proficiency), reasons related to the working hours of the personnel (e.g., leveling out overtime hours), and reasons related to the utilization efficiency of the equipment (e.g., reducing the number of setup changes).

[0114] This makes it easier to understand and input the intent of a plan when generating a plan for businesses that make up a supply chain.

[0115] (4) In the plan generation method described in (3) above, the information indicating the proficiency of the personnel includes at least one of information based on the personnel's years of service (e.g., years of service 11122), qualifications held (e.g., qualification A held 11123), and evaluations from others (e.g., evaluations 11124), and information obtained from a personnel management system.

[0116] This allows for appropriate proficiency to be acquired.

[0117] (5) In the plan generation method described in (1) above, the estimation reasons and the correction reasons include reasons written in natural language (e.g., free description).

[0118] This increases the degree of freedom in inputting the reason.

[0119] The present invention is not limited to the above-described embodiments, but includes various modifications. For example, the above-described embodiments have been described in detail to provide a better understanding of the present invention, and the present invention is not necessarily limited to those including all of the described configurations. Furthermore, it is possible to replace part of the configuration of one embodiment with the configuration of another embodiment, or to add the configuration of another embodiment to the configuration of one embodiment. Furthermore, it is possible to add, delete, or replace part of the configuration of each embodiment with other configurations.

[0120] Furthermore, the above-described configurations, functions, processing units, processing means, etc. may be partially or entirely implemented in hardware, for example, by designing them as integrated circuits. The above-described configurations, functions, etc. may also be implemented in software, with a processor interpreting and executing a program that implements each function. Information such as programs, tables, and files that implement each function can be stored in a storage device, such as a memory, a hard disk drive, or an SSD (Solid State Drive), or in a computer-readable non-transitory data storage medium, such as an IC card, an SD card, or a DVD.

[0121] In addition, the control lines and information lines shown are those that are considered necessary for the explanation, and not all control lines and information lines in the product are necessarily shown. In reality, it can be considered that almost all components are interconnected.

Claims

1. A plan generation method executed by a plan generation system, the plan generation system comprising a calculation unit and a memory unit, the memory unit holding a base plan which is a plan for a business, resource information on resources to be allocated to the business in the base plan, a recommended plan which includes changes recommended for the base plan, and a learning model which outputs inference reasons which are inferences of reasons for the changes in the recommended plan, the plan generation method comprising: a first step in which the calculation unit generates the recommended plan based on the base plan, the resource information, and the learning model; and a second step in which, when a revised plan which has revised the recommended plan and the reason for the revision are input, the calculation unit re-trains the learning model using the revised plan and the reason for the revision as correct data for the recommended plan and the inference reason, respectively.

2. A plan generation method as described in claim 1, wherein the estimated reason includes one or more reason items and a score indicating the importance of each item, and the reason for correction includes the amount of correction to the importance of each item included in the estimated reason, or the importance after correction.

3. A plan generation method as described in claim 2, wherein the resources include personnel and equipment, the master plan includes the allocation of the resources for each period in the businesses that make up the supply chain and the amount of output of the businesses, the resource information includes information indicating the proficiency of the personnel, information regarding the working hours of the personnel, and information indicating the equipment for which the personnel are responsible, and the reasons for estimation and reasons for correction include at least one of reasons related to the proficiency of the personnel, reasons related to the working hours of the personnel, and reasons related to the utilization efficiency of the equipment.

4. A plan generation method as described in claim 3, characterized in that the information indicating the proficiency of the personnel includes at least one of information based on the personnel's years of service, qualifications held, and evaluations from others, and information obtained from a personnel management system.

5. A plan generation method according to claim 1, wherein the reason for estimation and the reason for modification include reasons written in natural language.

6. A plan generation system comprising: a calculation unit; and a memory unit, wherein the memory unit holds a base plan which is a plan for a business; resource information regarding resources to be allocated to the business in the base plan; a recommended plan which includes changes recommended for the base plan; and a learning model which outputs inferred reasons for the changes in the recommended plan; wherein the calculation unit generates the recommended plan based on the base plan, the resource information, and the learning model, and when a revised plan which amends the recommended plan and the reason for the revision are input, the calculation unit re-trains the learning model using the revised plan and the reason for the revision as correct data for the recommended plan and the inferred reason.

Images