JP2017122955A - Production schedule planning support system and support method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To plan a production schedule that satisfies limitations or rules of a production steps.SOLUTION: A production schedule planning support system includes: a knowledge DB that stores predetermined limitations pertaining to a production schedule; a rule learning unit that extracts a first rule from a characteristic of a past schedule serving as a past production schedule having an achievement, and a second rule from the past schedule excluding the characteristic pertaining to the first rule; and a schedule preparation unit that prepares a production schedule of a product with the first rule applied in responce to an order placement of the product, applies the limitation relevant to the first rule to the prepared production schedule of the product, applies the second rule to the prepared production schedule of the product, and applies the limitation relevant to the second rule to the production schedule of the product having the second rule applied.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a production plan planning support system and a support method for supporting production planning of a product for which an order has been placed.

  Product production plans are drafted according to daily orders and production conditions. In the production plan to be drafted, regular orders and temporary orders are collected, and it is necessary to satisfy the delivery date within the range of production capacity in accordance with the stock status and production status at the time of planning. Production planning is a difficult task because there are many restrictions on production planning depending on the product. Production plans are made by skilled workers for products with many restrictions. Therefore, there is a demand for support by a computer for a task in which a skilled person makes a production plan based on experience.

  As a technology that meets this demand, there is Patent Literature 1. As a means for solving the problem of “producing a production plan efficiently by outputting production plan assignment candidates based on empirical rules”, Patent Document 1 describes “production lots to be assigned to production lines at the time of production plan creation”. When selecting, multiple assignment candidates are created and output based on certain rules and empirical rules, each of which is evaluated by summing up the evaluation results based on the empirical rules and the evaluation results predicted by the planning simulation. The output of the allocation candidates is rearranged in the order of the high possibility of being generated.The selected allocation candidates are learned in the rule-of-thumbnail database as the result of re-evaluation of the weight and used for the next generation of allocation candidates. doing.

JP 2012-194712 A

  In patent document 1, it is divided into a plurality of production management units such as production processes and hierarchies in each production process, and in each production management unit, the type and quantity of the ordered product such as the limit of the production amount due to the production facility performance, etc. If there are restrictions that must always be observed regardless of the rules and rules that are required to be protected as much as possible, the creation of a production plan that satisfies these restrictions and rules is not considered.

  The production plan planning support system to be disclosed extracts a first rule from a feature DB of a knowledge DB that stores constraints related to a predetermined production plan and a past plan that is a past production plan with a track record. A rule learning unit that extracts the second rule from the features of the past plan excluding the features related to the product, and, according to the product order, creates the production plan of the product by applying the first rule, and Apply the constraint related to the first rule to the production plan, apply the second rule to the production plan of the created product, and apply the constraint related to the second rule to the production plan of the product to which the second rule is applied Has a planning section.

  The disclosed production planning support system can support the planning of a production plan that satisfies the restrictions and rules in the production management unit as planned by an expert.

It is a block diagram of a production planning support system. It is a structural example of a computer which implements a production planning support system. It is an example of the order inputted into the production planning support system from the order reception system. It is an example of the production plan which a production planning support system outputs. It is an example of a past plan. It is an example of a constraint. It is an example of a rule. It is a process flowchart of a rule learning part. It is a process flowchart of a plan preparation part. It is an example of a learning screen. It is an example of a plan creation screen. It is a block diagram of the production planning support system of Example 2. It is an example of the learning screen of Example 2. It is an example of the plan creation screen of Example 2.

  A production planning support system for planning production of products will be described. The production plan planning support system has a knowledge DB that stores constraints related to a predetermined production plan, and includes a rule learning unit and a processing unit of a plan creation unit.

  The rule learning unit extracts the first rule from the feature of the past plan that is a past production plan with a track record, and extracts the second rule from the feature of the past plan excluding the feature related to the first rule. The rule learning unit sequentially extracts new rules while excluding features related to the extracted rules.

  The plan creation unit creates a production plan of the product by applying the first rule according to the order of the product, applies the constraints related to the first rule to the production plan of the created product, The second rule is applied to the production plan, and the constraints related to the second rule are applied to the production plan of the product to which the second rule is applied. In this way, the plan creation unit applies a rule and a constraint associated with the rule according to the order of the product, and formulates a production plan that represents the production order of the product in the production process.

  In order to make the explanation of the following embodiments of the production planning support system easier to understand, each production process is called a lot, a set, a base, and an assembly.

  FIG. 1 is a configuration diagram of a production planning support system (hereinafter referred to as a support system) 100. The support system 100 creates and outputs a production plan 120 corresponding to the product order input from the order reception system 110. The support system 100 includes a rule learning unit 200, a plan creation unit 300, a past plan 130, and a knowledge DB 140.

  The past plan 130 stores a past production plan with a track record. The knowledge DB 140 stores constraints 150 and rules 160 for making a production plan. The constraints 150 are conditions that must be satisfied in the production plan, such as production capacity and delivery date. The rule 160 is a condition that is required to be satisfied as much as possible in the production plan, such as minimizing the number of times of changing the tip tool of the processing machine.

  The rule learning unit 200 extracts the rule 160 from the past plan 130 and stores it in the knowledge DB 140. A constraint 150 related to production equipment such as production capacity is input in advance by a user (production planner) of the support system 100. Further, the rule 160 determined from the viewpoint of production efficiency without learning the past plan 130 may be input in advance by the user of the support system 100.

  The plan creation unit 300 drafts and outputs a production plan 120 corresponding to an order from the order reception system 110 in accordance with the constraints 150 and rules 160 stored in the knowledge DB 140.

  The order reception system 110 outputs an order indicating the type, number, delivery date, etc. of the order to be produced to the support system 100.

  FIG. 2 is a configuration example of a computer that implements the support system 100. The support system 100 is realized by causing a computer such as a general PC (personal computer) to execute a program.

  The computer includes a CPU 201, a memory 202, an interface 203, a network interface 204, a keyboard 205, a screen 206, a mouse 207, and a hard disk 208.

  The CPU 201 is a central processing unit, and executes a program stored in the memory 202 or transferred from the hard disk 208 to the memory 202 in advance. The memory 202 stores data associated with the execution of the program in addition to the program. The interface 203 connects each part in the computer. The network interface 204 is an interface for communicating with other computers such as the order receiving system 110. The keyboard 205 is operated by a user for inputting commands and data to the computer. A screen 206 is a screen of an output device that displays a production plan and the like by the support system 100. The mouse 207 is operated by the user in the same manner as the keyboard 205, and selects and inputs commands and data to the computer by operating the pointer displayed on the screen 206. The hard disk 208 stores programs and data.

  Each processing unit of the rule learning unit 200 and the plan creation unit 300 of the support system 100 is executed by the CPU 201, and the past plan 130 and the knowledge DB 140 are stored in the memory 202.

  FIG. 3 is an example of an order input from the order reception system 110 to the support system 100. For each order, an order ID 301, a product ID 302 indicating the product type, a product length (depth) 303, a width 304, and a height 305 dimension, a product quantity 306, a weight 307 per product, and , Have delivery date 308. The product length 303, width 304, height 305, and mass 307 per unit are determined to correspond to the model that represents the product type in the case of a standard product. The length 303 and the like may be omitted.

  The order ID 301 is an identifier that uniquely identifies an order. The example shown in FIG. 3 represents “Ord-11”, “Ord-12”, and “Ord-13” as the order ID 302. The product type ID 302 is an identifier for uniquely identifying the product type.

  Each item from the product length 303 to the delivery date 308 has a literal meaning. For example, if the order ID 301 is “Ord-11”, the product ID 302 is “Rcp-4”, the length 303 is “1 m”, the width 304 is “1 m”, the height 305 is “10 cm”, and the quantity 306 is “ 3 ”, the mass 307 represents“ 2.4 kg ”, and the delivery date 308 represents“ 2014/8/25 17:00 ”.

  FIG. 4 is an example of the production plan 120 output by the support system 100. Production plan 120 includes production plan ID 401, order ID 402, product type ID 403, lot ID 404, set ID 405, base ID 406, assembly ID 407, length 408, width 409, height 410, quantity 411, weight 412 and scheduled completion date and time. 413.

  The plan ID 401 is an identifier that uniquely identifies the plan. The example of FIG. 4 represents production plans of “Pln-11”, “Pln-12”, and “Pln-13” as plan IDs. The order ID 402 is an identifier for uniquely identifying the product order, and corresponds to the order ID 301. The product type ID 403 is an identifier for uniquely identifying the product type, and corresponds to the product type ID 302.

  The lot ID 404 is an identifier that uniquely identifies a lot that is one of the production processes of a product. The set ID 405 is an identifier for uniquely identifying a set that is one of product production processes. The base ID 406 is an identifier that uniquely identifies a base that is one of the production processes of a product. The assembly ID 407 is an identifier that uniquely identifies an assembly process that is one of product production processes.

  The item from length 408 to weight 412 has a literal meaning for the product and corresponds to the item from order length 303 to weight 307. The scheduled completion date and time 413 is the scheduled date and time when the product is completed, and the production plan 120 needs to be set to the date and time before the delivery date 308 of the order.

  For example, the production plan whose plan ID 401 is “Pln-11” is a production plan corresponding to the order ID 402 “Ord-12”, the product ID 403 is “Rcp-4”, the lot ID 404 is “Lo5-1”, and the set ID 405 Is “Ct3-1”, the base ID 406 is “Sb6-0”, the assembly ID 407 is “Ro-4”, the length 408 is “1 m”, the width 409 is “1.2 m”, the height 410 is “50 cm”, The quantity 411 is “2”, the weight 412 is “1.0 kg”, and the scheduled completion date 413 is “2016/1/18 05:20”.

  Production plan planning includes lot ID 404, set ID 405, base ID 406, assembly ID 407 according to product ID 302, length 303, width 304, height 305, quantity 306, weight 307, and delivery date 308 included in the order. Is to assign.

  Since the order of the lot, set, base and assembly is the important information, the lot ID 404, set ID 405, base ID 406 and assembly ID 407 correspond to the number of orders in this example. Let The lot ID 404, set ID 405, base ID 406, and assembly ID 407 are used as objective variables for regression analysis in the processing of the rule learning unit 200 described later.

  FIG. 5 is an example of the past plan 130. The example of FIG. 5 represents production plans with “Pln−1”, “Pln-2”, and “Pln-3” as plan IDs. The past plan 130 is a planning result of a past production plan having almost the same items as the production plan 120. Each item from plan ID 501 to mass 512 of the past plan 130 corresponds to each item from plan ID 401 to weight 512 of the production plan 120. The planner ID 513 is an identifier that uniquely identifies the planner (user). Completion date / time 514 is the completion date / time of the product, and is an item representing the execution result of the past plan 130.

  FIG. 6 shows an example of a constraint 150 that is referred to when the plan creation unit 300 creates a production plan. The constraint 150 has a constraint ID 601, a constraint expression 602, and an application management level 603. The constraint ID 601 is an identifier that uniquely identifies the constraint. The example of FIG. 6 represents the constraints of “Ct−1” and “Ct-2” as the constraint ID 601.

  The constraint expression 602 is an expression representing the content of the constraint. The application management level 603 is a level of application management of constraints to the production plan, and is a constraint that is checked first when the level is higher (the numerical value is larger). For example, the constraint equation 602 of the constraint 150 with the constraint ID 601 of “Ct−1” is “∀x∈lot ID, ys ← input data ID (lot ID = x), Sum (mass (ys)) <200”, The application management level is “6”. That is, the constraint with the constraint ID 601 of “Ct−1” is defined as having an application management level 603 of 6 and a total sum of weights of input data having the same lot ID of less than 200.

  FIG. 7 is an example of a rule 160 that is referred to when the plan creation unit 300 creates a production plan. The rule 160 includes a rule ID 701, an objective variable 702, a main component 703, and regression coefficients (10% regression coefficient 704, 20% regression coefficient 705,...). The rule ID 701 is an identifier that uniquely identifies the rule. The example of FIG. 7 represents the rules “Rul-1” and “Rul-2” as the rule ID 701.

  The objective variable 702 in the present embodiment represents the priority order for production or the order of production. The objective variable 702 is set in advance according to the production process.

  The principal component 703 is a principal component calculation formula for dividing the past plan 130 into categories. The 10% regression coefficient 704 is a regression coefficient for explaining the objective variable 702 using the principal components from the past plan 130 included in 10% of the principal components. The 20% regression coefficient 705 is a regression coefficient for explaining the objective variable 702 based on the principal components from the past plan 130 included in 20% of the principal components. Similarly, the percentile point values are described below.

  For example, in the rule 160 whose rule ID 701 is “Rul−1”, the objective variable 702 is “set ID”, the principal component 703 is “0.7 * delivery time + 0.3 * width”, and the 10% regression coefficient 704 is “123”. .4 ", the 20% regression coefficient 705 is" 234.5 ". This is because 0.7 × delivery time + 0.3 × width can be used to calculate the principal component of the past plan 130, and for each category delimited by the percentile point at that time, the set ID, that is, the order of manufacturing in the set unit is predicted. Regression coefficient is shown. That is, in the rule 160 whose rule ID 701 is “Rul−1”, the principal component 703 based on the delivery date and the width is correlated with the set ID of the objective variable 702, and is similar to the past plan 130 belonging to 0% to 10%. The order is a rule that a set ID calculated based on the regression coefficient of “123.4” and the principal component should be assigned.

  The processing of the rule learning unit 200 will be described with reference to FIGS.

  FIG. 10 is an example of a learning screen 1001 used for the processing of the rule learning unit 200. The learning screen 1001 includes a “past plan call” button 1002, a “rule candidate display” button 1003, a “learn” button 1004, a “learning end” button 1005, a past plan display area 1006, a rule candidate display area 1007, and a constraint input area. Having 1008.

  A “past plan call” button 1002 is a button for the user to instruct the learning screen 1001 to call the past plan 130. A “rule candidate display” button 1003 is a button for the user to instruct the rule learning unit 200 to search for and display rule candidates. The “learn” button 1004 is a button for the user to instruct the rule learning unit 200 to learn rule candidates. The “learning end” button 1005 is a button for the user to instruct the rule learning unit 200 to end the learning process.

  The past plan display area 1006 is an area for displaying the past plan 130. When the user clicks a “past plan call” button 1002 and selects an appropriate past plan, the past plan 130 is displayed in the past plan display area 1006. The rule candidate display area 1007 is an area for displaying rule candidates. When the user clicks a “rule candidate display” button 1003, the rule 160 is displayed as a candidate in the rule candidate display area 1007. The constraint input / display area 1008 is an area for inputting / displaying the constraint 150.

  FIG. 8 is a process flowchart of the rule learning unit 200. The rule learning unit 200 derives a rule 160 from the past plan 130 that the user has planned in the past.

  The rule learning unit 200 calls the past plan 130 in response to a click on the “past plan call” button 1002 by the user, displays the past plan 130 in the past plan display area 1006 of the learning screen 1001, and displays the past plan display area 1006. In response to the selection by the user from the inside, the past plan 130 to be learned is specified (S801).

  The rule learning unit 200 calls the constraint 150 from the knowledge DB 140 and displays it in the constraint input / display area 1008. Further, the rule learning unit 200 inputs / displays a new constraint from the constraint input / display area 1008 and stores it as the constraint 150 in the knowledge DB 140 (S802).

  In response to the user clicking the “learn” button 1004, the rule learning unit 200 substitutes the past plan 130 specified as the learning target in fs (S803). Here, fs is a set of past plans that are processed by the rule learning unit 200. The initial state of fs has items of plan ID 501 to completion date and time 514 of the past plan 130 specified in S801, and the number of items decreases as the processing proceeds (for each loop processing from S804 to S807 below). .

  The rule learning unit 200 determines whether or not there are a predetermined number of items of the past plan 130 in fs (S804). If there are more than a predetermined number of items in fs, the process proceeds to S805. If so, the process ends. Here, the presence / absence of the predetermined number or more items is determined to be less than the predetermined number when the number of items becomes one, that is, when the number of items is less than two. Alternatively, if the ratio of the number of items remaining in fs to the number of items of the past plan 130 (plan ID 501 to 14 of completion date 514 in the example of FIG. 5) is less than 10%, for example, it is determined that the number is less than a predetermined number. May be.

  The rule learning unit 200 calculates principal components by general principal component analysis (for example, Shimauchi, other edition, algorithm dictionary, Kyoritsu publication (1994)) based on the values of each item of the past plan 130 in fs. (S805). The principal component is calculated from the value of the item remaining in fs, and the first principal component is extracted. The principal component calculation formula is stored in rule 160 as principal component 703.

  The rule learning unit 200 divides the past plan 130 specified as a learning target based on the principal component into categories, and calculates a regression coefficient that explains the objective variable for each category (S806). Here, the objective variable is a value representing a production order such as a lot ID 504, a set ID 505, a base ID 506, and an assembly ID 507 included in the past plan 130. Based on the extracted principal component values, the rule learning unit 200 calculates the percentile points as shown in FIG. 7 and categorizes the values into 10% regression coefficient 704 and 20% regression coefficient 705. Record. Next, the rule learning unit 200 calculates and records a regression coefficient that improves the objective variable based on the principal component calculated in Step 805 from the past plan 130 belonging to each category. For example, a general single regression analysis method may be used for calculating the regression coefficient. As shown in FIG. 7, the principal component calculation formula calculated in step 805 and the set of regression coefficients for each category calculated in step S806 are collectively referred to as a rule.

  The rule learning unit 200 displays the created rule in the rule candidate display area 1007. Here, you may stop once for the confirmation of the rule by a user. In this case, the user clicks on a rule in the rule candidate display area 1007 to instruct the adoption or non-adoption of the rule. Note that the user may check the rules every time the rules are created, or may check the rules collectively at the end. When confirming collectively at the end, after the determination that the number of items in fs is less than the predetermined number in S804 and before the user clicks the “end learning” button 1005, the user can select the rule candidate display area. Click on the rule in 1007 to instruct whether to adopt or reject the rule.

  The rule learning unit 200 stores the extracted first principal component item removed from fs in fs (S807). That is, the component having the principal component as the basis (projection length dropped from fs on the axis having the principal component as the basis) is subtracted from the explanatory variable of fs.

  The process of the plan creation unit 300 will be described with reference to FIGS. 9 and 11.

  FIG. 11 shows an example of a plan creation screen 1101 used by the plan creation unit 300 for processing. The plan creation screen 1101 includes an “Read Order Information” button 1102, a “Create Plan” button 1103, a “Modify Plan” button 1104, an “End Plan Creation” button 1105, an order information display area 1106, a rule display area 1107, and a plan display. A region 1108 is included.

  A “read order information” button 1102 is a button for the user to instruct the plan creation unit 300 to read an order from the order reception system 110. The “plan creation” button 1103 is a button for the user to instruct the plan creation unit 300 to create a production plan. A “plan correction” button 1104 is a button for the user to instruct the plan creation unit 300 to correct the production plan. A “plan creation end” button 1105 is a button used by the user to instruct the plan creation unit 300 to end the production plan creation. The order information display area 1106 is an area for displaying an order read from the order receiving system 110 by the plan creation unit 300. The rule display area 1107 is an area where the plan creation unit 300 displays the rule 160. The plan display area 1108 is an area for displaying the production plan 120 created by the plan creation unit 300.

  FIG. 9 is a process flowchart of the plan creation unit 300. The plan creation unit 300 determines the manufacturing order in each production process such as lot, set, base, and assembly for a plurality of orders as shown in FIG.

In response to the click of the order information reading button 1102 by the user, the plan creation unit 300 inputs an order from the order reception system 110 and displays the input order in the order information display area 1106 (S901). In response to the user clicking the plan creation button 1103, the plan creation unit 300 extracts one rule from the rules 160 stored in the knowledge DB 140 (S902). If there is a rule 160, the process proceeds to S903, and if there is no rule 160, the process proceeds to S907. The plan creation unit 300 calculates the principal component score of the order according to the principal component calculation formula of the extracted rule 160 (S903). Here, the principal component score of the order is calculated using the calculation formula of the principal component 703 included in the rule 160. Then, orders are classified into categories based on the principal component score.

  The plan creation unit 300 predicts the objective variable (order) of each order according to the regression coefficient corresponding to the extracted rule 160, and rearranges the orders in the predicted order (S904).

  The plan creation unit 300 checks whether there is a constraint 150 having a variable having a high contribution rate to the principal component 703 of the extracted rule 160 in the constraint equation 602 (S905). If there is a constraint 150, the process proceeds to S906, and if not, the process returns to S902.

  The plan creation unit 300 applies all the constraints up to the application management level 603 of the constraints 150 (S906). Here, as described above, the constraint 150 having a high (large numerical value) application management level 603 is checked first. Here, the application management level 603 of the constraint 150 is referred to, and the constraint 150 having the same or higher application management level is applied. Here, if the constraint 150 is not satisfied, a value that satisfies the constraint 150 is searched by moving the order item value up or down. The search may try the value of the item of the order one by one, or a technique such as general constraint programming may be used. In the case of using constraint programming, the value of the order item may be a constraint variable, and all the other constraints 150 up to the application management level 603 of the constraint 150 may be applied to search for the value of the constraint variable.

  Here, the plan creation unit 300 may be stopped once to confirm the plan. In this case, the user clicks on the production plan in the plan display area 1108 to instruct the adoption or non-adoption of the production plan. In addition, the user may confirm for every production plan, and may confirm collectively at the end. When checking all together at the end, the user clicks the production plan in the plan display area 1108 after the extraction of each rule 150 in S902 and before the user clicks the end button 1105, and adopts the production plan. Or instruct non-recruitment. Alternatively, the user may directly change each value of the production plan in the plan display area 1108.

  If the unapplied constraint 150 remains, the plan creation unit 300 applies the remaining constraint 150 as an unapplied constraint (S907).

  According to the production planning support system of the present embodiment, it is possible to support the planning of a production plan that satisfies the restrictions and rules in the production management unit as planned by a skilled person.

  When a plurality of users share a production plan, different rules may be adopted for each user for the same order. In such a case, one of different rules may be selectively employed or different rules may be used in combination. The present embodiment is a production planning support system corresponding to such a situation.

  FIG. 12 is a configuration diagram of the production plan planning support system 100 of the present embodiment. The production planning support system 100 of the present embodiment is different from the production planning support system 100 of the first embodiment in that rule 2 (1201) is stored in the knowledge DB 140. Note that rules 3 and 4 may be included.

  FIG. 13 is an example of a learning screen 1001 of the present embodiment. The learning screen 1001 of this embodiment has a learning target variable selection area 1301 and a rule number input area 1302 in addition to the screen configuration of the learning screen 1001 of the first embodiment shown in FIG. In S801 of the rule learning unit 200 (FIG. 8), the user can select an objective variable to be learned, and learn by adding a rule number as a partial rule 2 (1201). According to the rule number, the rule learning unit 200 stores the rule number 2 (1201) in the knowledge DB 140. A rule number may be added to the constraint 150 corresponding to the rule number and stored in the knowledge DB 140.

  FIG. 14 is an example of the plan creation screen 1101 of the present embodiment. The plan creation screen 1101 of this embodiment has a plan target objective variable selection area 1401 and an application rule number input area 1402 in addition to the screen configuration of the plan creation screen 1101 of the first embodiment shown in FIG. In S901 of the plan creation unit 300 (FIG. 9), the user can select a target variable to be planned, add a rule number as an applied rule, and instruct the plan creation. In accordance with the rule number, the plan creation unit 300 reads the rule 160 or rule 2 (1201) of the number and the constraint 150 from the knowledge DB 140, and executes the processing from S902. Further, S907 of the plan creation unit 300 (FIG. 9) may be executed after all the rules 160 are applied.

  According to the production plan planning support system of the present embodiment, it is possible to support the planning of a production plan that satisfies at least one of the restrictions in the production management unit and a plurality of different rules as planned by an expert.

  100 ... Production planning support system, 110 ... Order reception system, 120 ... Production plan, 130 ... Past plan, 140 ... Knowledge DB, 150 ... Restriction, 160 ... Rule, 200 ... Rule learning part, 300 ... Planning part, 1001 ... Learning screen, 1101 ... Plan creation screen, 1201 ... Rule 2.

Claims (8)

A knowledge DB that stores constraints related to a predetermined production plan;
A rule learning unit that extracts a first rule from a feature of a past plan that is a past production plan with a track record, and extracts a second rule from the feature of the past plan excluding the feature according to the first rule; And, according to a product order, applying the first rule to create a production plan for the product, applying the constraint associated with the first rule to the produced production plan for the product, A production plan formulation comprising: a plan creation unit that applies the second rule to a production plan and applies the constraint related to the second rule to a production plan of the product to which the second rule is applied Support system.   The rule learning unit calculates a principal component by a principal component analysis for each item value of the past plan representing a feature of the past plan, and an objective variable representing a production order of the products in a process of producing the product 2. The production planning according to claim 1, wherein the regression coefficient is calculated, and the calculation formula for calculating the principal component and the calculated regression coefficient are the first rule or the second rule. Support system.   The production planning support system according to claim 2, wherein the rule learning unit calculates the regression coefficient corresponding to a predetermined percentile point.   The plan creation unit calculates a principal component score of the order according to the calculation formula of the principal component of the first rule, and predicts the objective variable of the order according to the regression coefficient of the first rule. Rearranging the orders in the order indicated by the predicted objective variable, creating a production plan for the product, and when there is the constraint relating to the first rule, the production plan for the product relates to the first rule 4. The production planning support system according to claim 3, wherein the restriction is applied. A production plan planning support method in a production plan planning support system having a knowledge DB for storing constraints relating to a predetermined production plan, wherein the production plan planning support system includes:
Extract the first rule from the features of the past plan, which is a past production plan with a track record,
Extracting a second rule from the features of the past plan excluding the features related to the first rule;
Depending on the product order, the first rule is applied to create a production plan for the product,
Applying the constraints associated with the first rule to the production plan of the created product;
Applying the second rule to the production plan of the product;
A production plan planning support method, wherein the restriction related to the second rule is applied to a production plan of the product to which the second rule is applied. The production plan planning support system calculates a principal component by principal component analysis for each item value of the past plan representing the characteristics of the past plan,
Calculating a regression coefficient that describes an objective variable that represents the production order of the product in the process of producing the product;
6. The production planning support method according to claim 5, wherein the calculation formula for calculating the principal component and the calculated regression coefficient are the first rule or the second rule.   7. The production planning support method according to claim 6, wherein the production planning support system calculates the regression coefficient corresponding to a predetermined percentile point. The production planning support system calculates the principal component score of the order according to the calculation formula of the principal component of the first rule,
Predicting the objective variable of the order according to the regression coefficient of the first rule;
Create a production plan for the product in which the orders are rearranged in the order represented by the predicted objective variable,
8. The production plan planning support method according to claim 7, wherein when there is the restriction related to the first rule, the restriction related to the first rule is applied to the production plan of the product.

Images