JP2010257407A - Scheduling device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To create a schedule in which a restrictive condition is taken into account. <P>SOLUTION: A scheduling device includes: a priority matrix setting means for setting a priority matrix showing assignment priority of a tank for each product; and a schedule calculation means for creating a work schedule by performing assignment of work on the basis of the priority matrix using a backward method for avoiding piping contention and contention of a mixer so as to keep the delivery date of a product. A priority matrix representing assignment priority of the tank is set for each connection relation between the tank and piping, piping contention, contention of the mixer, cleaning amount, product brand, and product, the piping contention and contention of the mixer are avoided so as to surely keep the delivery date of a refill request given on the basis of the idea of backward method, and work is assigned on the basis of the priority matrix. Thus, the schedule which avoids the piping contention and contention of the mixer, considers lot concentration, and minimizes a cleaning amount by the change of product brand is created. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a scheduling apparatus that creates a work schedule in a mixing plant capable of producing a variety of products by mixing a plurality of types of raw materials.

  As a conventional scheduling method, there is JP-A-11-53006. In the scheduling method described in this publication, for a plurality of jobs given to a system of predetermined work, the predetermined work for performing the job is divided for each process, and for each process, the process of this job is divided. Information for performing work is taken in, and processing steps of creation of each work schedule and information exchange for creating a schedule between each process are repeated to create a work schedule for the entire system.

  Further, as a mixing plant capable of mixing a plurality of types of raw materials to produce a wide variety of products, for example, there is a lubricating oil mixing plant. In the mixing process of the lubricating oil mixing plant, the raw material is transferred from the base oil tank and mixed by a batch blender or a continuous blender. The mixed lubricating oil is stored in a tank, filled into drums and lorries through a filling line, and shipped as a product.

  Further, as a tank schedule creation support for determining a distribution amount of each crude oil to each tank when distributing a plurality of types of crude oil in a large-scale crude oil tanker (VLCC) to a plurality of tanks, JP-A-11-167852 There is a gazette. In the tank schedule creation support device described in this publication, a pair of calculation target crude oil and a calculation target tank is identified sequentially, and a virtual amount is determined between the pair according to the API degree of the calculation target crude oil and the calculation target tank. If there are unsatisfied constraint conditions in the state where the crude oil movement is no longer possible and the virtual crude oil movement cannot be performed, the virtual crude oil movement between the pair according to the conditions is prohibited, and again From the beginning, a virtual crude oil movement is carried out. In addition, API degree is a unit which shows the specific gravity of the crude oil and petroleum product which American Petroleum Institute (American Petroleum Institute) established.

  Moreover, the piping system described in JP-A-2004-162756 is a piping system that connects an upstream device group and a downstream device group, and includes a plurality of devices in the upstream device group. And a dedicated pipe that is individually connected to each other and a circulation pipe that is connected to each dedicated pipe via a switching device. With such a configuration, when manufacturing a small quantity of products in large quantities, it is possible to individually manufacture each product by operating each system facility without using the circulation piping. By using, it becomes possible to manufacture main varieties and varieties with low production frequency while managing the process.

  Moreover, there exists Unexamined-Japanese-Patent No. 2000-137502 as a scheduling system. The scheduling system described in this publication is a scheduling system in which raw materials having different properties are received and received by a plurality of raw material tanks, and means for setting the receiving priority of raw material tanks by a combination of a plurality of requirements for each raw material. There are means for determining a payout target value that suppresses the property change range within the change limit value, and means for selecting and determining a combination of payout tanks closest to the payout property target value and a discharge flow rate from each tank. With such a configuration, in order to improve the controllability of the production amount of the main product by the manufacturing apparatus, the properties of the raw material dispensed to the manufacturing apparatus can be predicted or controlled quickly and accurately.

Japanese Patent Laid-Open No. 11-53006 Japanese Patent Laid-Open No. 11-167652 JP 2004-162756 A JP 2000-137502 A

  In a mixing plant capable of producing a wide variety of products, drafting a production plan that takes into account piping competition at the inlet and outlet sides of the tank, reduction of cleaning costs by switching when changing items, and improvement of mixing process capability by consolidating mixed lots Is required. Production planning in a conventional mixed plant has been scheduled based on the knowledge and experience of field operators, so the operational efficiency depends on the judgment and ingenuity of the operator, and improvement in operational efficiency is required. It was. Conventionally, a scheduling device in which a commercial software package is customized is mainly used, and it is difficult to reflect various constraint conditions of the mixing plant in the conventional software.

  In addition, the schedule system described in Patent Document 5 has a problem that it is not possible to optimize constraints such as costs, although it is possible to set the acceptance priority order and create a schedule.

  The present invention has been made to solve such problems, and creates a schedule that takes into account the constraints in a chemical plant composed of a plurality of tanks and piping such as a mixing plant and a filling plant. It is an object of the present invention to provide a scheduling device capable of performing the above.

  A scheduling device according to the present invention is a scheduling device that creates a work schedule in a mixing plant having a plurality of tanks, a mixing device, and a pipe connecting the tanks and the mixing device, and indicates the priority of tank allocation for each product. By assigning work based on the priority matrix using the priority matrix setting means to set the degree matrix and the backward method to avoid the competition of piping and mixing equipment so as to meet the delivery date of the product And a schedule calculation means for creating a work schedule.

  Such a scheduling device sets a priority matrix that expresses tank allocation priority for each tank and piping connection relationship, piping contention, mixing device contention, cleaning amount, product brand, and product. In the scheduling device, based on the idea of the backward method, it is possible to avoid the competition between the piping and the mixing device so as to satisfy the delivery date of the given filling request, and to assign the work based on the priority matrix. This makes it possible to create a schedule that minimizes the amount of washing due to product brand change in consideration of lot aggregation while avoiding competition among a plurality of pipes and a plurality of mixing devices.

  In addition, it is possible to obtain a solution close to the optimum according to the allowable calculation time by using the simulated annealing method, and it is preferable to create a work assignment order list. Thereby, cost reduction can be aimed at.

  The priority matrix setting means expresses the priority matrix with three kinds of integers. The priority matrix is a first integer that represents a situation that is always assigned, and a first integer that represents a situation that can be assigned. It is preferably represented by an integer of 2 and a third integer that represents a situation that cannot be assigned. For example, the priority matrix can be expressed as an integer from 1 to 3, and expressed as “3: a situation that is always assigned”, “2: a situation that can be assigned”, and “1: a situation that cannot be assigned”. Therefore, it is possible to assign work reflecting these priority matrices.

  In addition, the scheduling calculation means assigns the work so that the penalty value for not satisfying the constraint condition in the plant is minimized, and the penalty value increases the cost caused by not satisfying the constraint condition. It is preferable to set according to. Here, the penalty value is the cost for exceeding the maximum intermediate inventory amount of the tank, the cost for exceeding the specified number of mixing (equipment capacity), the cost for replenishment of the shortage of the mixing amount, and work occurs on holidays Cost for product inspection, lack of product inspection work, cost for exceeding the production capacity of mixing equipment, penalty cost for violating competition restrictions of mixing work, violation of filling capacity constraints It is preferable to set the cost according to the cost for exceeding, the penalty cost for violating the filling requirement constraint, and the like.

  In addition, it is preferable that the scheduling calculation unit fixes the selected work and creates a reschedule.

  Moreover, as a system configuration for realizing the scheduling calculation means, a configuration including a shared database, a Gantt chart output system, and a database change system can be mentioned. As a result, it is possible to flexibly cope with various changes of brands and changes in tank connection relations.

  In addition, the scheduling calculation means distinguishes the mixing amount and mixing timing in the tank into a general-purpose tank and a dedicated tank, and considers the restrictions on the minimum mixing amount and the maximum mixing amount of the tank, so that the cleaning amount is minimized. It is preferable to create a work schedule for mixing, inspection and cleaning.

  In addition, it further includes piping data storage means for storing piping data, which is data relating to piping connection relations, piping competition between multiple brands, and flow velocity in the pipe, and the schedule calculation means is pipe competition between multiple brands. It is preferable to create a work schedule in consideration of the flow rate.

  Moreover, it is preferable to further include brand data storage means for storing brand data that is data relating to a plurality of brands, and the schedule calculation means preferably additionally updates the work schedule according to addition or change of brand data.

  According to the scheduling apparatus of the present invention, it is possible to create a schedule that takes into account the constraints in the mixing plant.

It is a schematic block diagram of a lubricating oil mixing plant. It is the schematic of the scheduling apparatus which concerns on embodiment of this invention. It is a figure which shows an example of the data list memorize | stored in the memory | storage part. It is a flowchart which shows the process sequence of priority allocation. It is a flowchart which shows the preparation procedure of the filling job allocation order list | wrist using SA method. It is a flowchart which shows the preparation procedure of _new production plan _Rnew . It is a figure which shows the priority matrix expressing the allocation priority of the product with respect to a tank. It is a flowchart illustrating a procedure for allocating a filling job f _i during filling job allocation order list L _f. It is a flowchart which shows the allocation procedure of a mixed job. It is a production plan Gantt chart which shows an example of the production plan created by the scheduling apparatus. It is a table | surface which shows the performance of a mixer. It is a figure which shows tank equipment data. It is a table | surface which shows an example of washing | cleaning cost. It is a table | surface which shows an example of washing | cleaning cost. It is a figure which shows the allocation priority matrix which concerns on a present Example. It is a figure which shows the allocation priority matrix which concerns on a present Example.

  Preferred embodiments of the present invention will be described below with reference to the drawings. In the description of the drawings, the same or corresponding elements are denoted by the same reference numerals, and redundant description is omitted. A scheduling apparatus according to an embodiment of the present invention supports creation of a work schedule in a mixing plant capable of manufacturing a wide variety of products. Hereinafter, a scheduling device for creating a work schedule in the lubricating oil mixing plant will be described.

FIG. 1 is a schematic configuration diagram of a lubricating oil mixing plant. A lubricating oil mixing plant 1 shown in FIG. 1 includes a raw material tank T _{A to} T _C for storing raw material oil, a mixing device M _k (continuous processing device M ₁ , batch blender M ₂ ) for mixing a plurality of types of raw material oil, and mixing. A product tank T _k (T _{1 to} T ₄ ) for storing lubricating oil (product) mixed by the apparatus M _k , a container filled with lubricating oil (for example, drum 5, lorry 6, P / L can, 4 L can, carrier ship) Etc.) is provided. These tanks T _{A to} T _C , the mixing device M _k , the tank T _k , and the filling device 4 are connected by a plurality of pipes L _{1 to} L ₆ . In addition, a pump for transferring an internal fluid (lubricating oil, cleaning oil) is connected to the pipes L _{1 to} L ₆ .

  FIG. 2 is a schematic diagram of the scheduling apparatus according to the embodiment of the present invention. The scheduling device 10 shown in FIG. 2 includes an input unit 11 for inputting various data such as job (filling request) data, an electronic control unit 20 for creating a schedule based on the input data, and processing performed by the electronic control unit 20. An output unit 12 for outputting the result is provided.

  The input unit 11 includes an input unit (for example, a keyboard) operated by an operator (user) and a connection unit that enables connection with various recording media.

  The output unit 12 outputs a schedule created by the electronic control unit 20, and includes an image display device and a printing device.

  The electronic control unit 20 includes a CPU that performs arithmetic processing, a ROM and a RAM that are storage units, an input signal circuit, an output signal circuit, a power supply circuit, and the like. In the storage unit of the electronic control unit 20, a job DB 31, a product tank DB 32, a product DB 33, and a constraint condition DB 34 are constructed. In the electronic control unit 20, a work setting unit 21, a priority matrix setting unit 22, and a schedule calculation unit 23 are constructed by executing a program stored in the storage unit.

The constraint DB 31 stores data related to constraints in the lubricating oil mixing plant. The data relating to the constraint conditions include the filling delivery date constraint conditions, the constraint conditions related to the usable mixing apparatus, the daily mixing frequency limit (batch blender M ₂ ), the daily mixing volume limit (continuous processing apparatus M ₁ ), and the tank competing constraints T _k inlet pipe L _2, competing constraints of the filling device 4 inlet side pipe L _3, L _4, have data on and inspection work constraints.

  FIG. 3 is a diagram illustrating an example of a data list stored in the storage unit. The job DB 32 is a database that stores data relating to jobs (filling requests), and stores request IDs, (product) oil types, (product) request amounts, and delivery times.

  The product tank DB 33 is a database that stores data related to the product tank, and stores a tank ID (tank number), (product) oil type, tank capacity, connection piping data, minimum lot, and maximum lot. The connection piping data is information regarding piping connected to the tank, and is data regarding piping connection destination, piping connection source, piping branching, and the like. The product tank DB 32 stores data related to competing piping. The data such as the pipe connection destination, the connection source, and the pipe branching are obtained by quantifying the data shown in FIG.

  The product DB 34 is a database that stores data related to products (brands), and stores data such as product ID, (product) oil type, mixing time, inspection time, filling time, and cleaning time.

  The electronic control unit 20 executes a program stored in the storage unit, thereby constructing a schedule calculation unit. The schedule calculation unit 21 functions as priority matrix setting means and schedule calculation means of the present invention.

  The priority assignment executed by the schedule calculation unit 21 of the electronic control unit 20 is an algorithm combining a meta strategy and a rule base based on the concept of the backward method. By using the backward method of assigning jobs in reverse direction from the delivery date, it is possible to guarantee the delivery date.

First, an order list of jobs to be processed: L _f = {f ₁ , f ₂ ,..., F _n } (1) is created, and job assignment is executed in order on this order list (each job is executed). Decide which tank to play). FIG. 4 is a flowchart showing a procedure for creating a schedule. Hereinafter, the step may be abbreviated as S. When creating a schedule according to the order list, job creation (S1), filling plan creation (S2), and mixed plan creation (S3) are executed. In other words, a filling job assignment order list creation step (S1), a filling job assignment step (S2), and a mixed job assignment step (S3) are performed.

  In step 1, a filling job allocation order list is created. In step 2, priority is assigned to the tanks in order of the order list. In step 3, a mixing job is assigned to each tank so that filling can be performed. Then, these S1 to S3 are repeated to optimize the objective function. That is, optimization is performed by creating a production plan that minimizes the objective function indicating the mixing cost and the cleaning cost.

  Next, creation of a filling job allocation order list will be described. Here, a filling job allocation order list is created using a simulated annealing method (hereinafter referred to as “SA method”), which is a kind of metaheuristics. By using the SA method, a nearly optimal solution can be obtained according to an allowable calculation time.

FIG. 5 is a flowchart showing a procedure for creating a filling job allocation order list using the SA method. In the creation of the filling job allocation order list using the SA method, the processing of step 11 to step 22 is executed. In step 11, parameters used in the SA method (T ₀ : initial temperature, N ₀ : maximum number of repetitions with the same objective function value, N _S : number of lists generated at the same temperature, δ: temperature update rate) are set. . Let N _count ← 0.

Next, in step 12, numbers are assigned from 1 in order of request with the earliest delivery date, a filling job assignment order list Lf is created, filling jobs are assigned and mixed jobs are assigned in order of the order list Lf according to the procedure described later, and production is performed. Make a plan R. Further, the optimum production plan R at the present time is set as the optimum production plan R _opt . Furthermore, setting the temperature parameter t with t = _{T s.}

Next, in step 13, a new optimum production plan R _opt is created by executing steps 31 to 33 shown in FIG. FIG. 6 is a flowchart showing a procedure for creating a _new production plan _Rnew . In step 31, randomly selecting two jobs from the filling job allocation order list L _f. In step 32, the two jobs selected in step 31 are replaced, and the replaced job is set as a new filling job allocation order list _Lfnew . In step 33, according to the new filling job assignment order list L _fnew set in step 32, filling job assignment and mixed job assignment are performed according to the processing procedure described later, and a production plan R _new is created.

Returning to FIG. 5, the process of step 14 is executed. Here, the objective function value for the production plan R is expressed as y (R). It is determined whether or not y (R _new ) ≦ y (R). If y (R _new ) is better than y (R), it is determined that y (R _new ) ≦ y (R). When it is determined that y (R _new ) ≦ y (R), the process proceeds to step 15, and when it is not determined that y (R _new ) ≦ y (R), the process proceeds to step 16. .

In step 15, R ← R _new and L _f ← L _fnew are set, a new production plan R _new and a filling job allocation order list L _fnew are unconditionally accepted, and the process proceeds to step 18. In step 16, R ← R _new and L _f ← L _fnew are accepted with probability e ^{−Δ / t} and the process proceeds to step 17. In step 17, it is determined whether or not the _new production plan _Rnew and the filling job allocation order list _Lfnew have been accepted. If the new production plan R _new and the filling job allocation order list L _fnew are accepted, the process proceeds to step 18, and if the new production plan R _new and the filling job assignment order list L _fnew is not accepted, Proceed to step 19.

In step 18, the number of times the same objective function value has continued is _Ncount = 0. In step _19, 1 is added to _{N count.}

In the subsequent step 20, if y (R) ≦ y (R _opt ), R _opt ← R. In step 21, if N _count ≧ N ₀ , the process ends. Otherwise, return to Step 13.

In step 22, repeat the step 13 to step 21 _{N S} times. Return to step 13 as t = δ · t.

  Next, a procedure for assigning filling jobs to tanks that have been assigned in the order of assignment will be described. First, the priority matrix will be described. In a conventional lubricating oil mixing plant, an operator determines a tank to be used at the time of manufacturing each product in consideration of later production plans. A matrix that reflects such job allocation criteria in a numerical value is set as a priority matrix.

  FIG. 7 is a diagram illustrating a priority matrix that represents the priority of product allocation to tanks. FIG. 7 shows an example of a priority matrix when there are five types of product oil types (P1 to P5) and the number of tank bases is four (T1 to T4). Here, the ease of allocation of the filling job is expressed using integers in three stages of 1 to 3. The meanings of the numerical values are “3: always assigned”, “2: occasionally assigned”, and “1: not assigned”. For example, in the case of product oil P1, for tanks T1 and T4, “1: not allocated”, for tank T2, “2: occasionally allocated”, and for tank T2, “3: “Always assigned” is set. In this case, the product oil P1 is most easily allocated to the tank T2, and then represents that the product oil P1 is easily allocated to the tank T3.

  Next, a filling job allocation method using a priority matrix will be described. The cleaning cost is determined when the tank to which the filling job is assigned is determined. Therefore, the optimal filling process plan is created by determining the tank to be operated in consideration of the cleaning cost required when the tank is allocated and the priority of the tank to be allocated.

Next, the filling job allocation order list _{L f = {f 1, f} 2, ..., f i, ..., f n-1, f n} will be described To allocate the filling job _{f i} in. When the product filling job f _i and I _fi, allocation algorithm is as follows. Figure 8 is a flowchart illustrating a procedure for allocating a filling job f _i during filling job allocation order list L _f.

First, in step 41, the priority for the product _{I fi} tank numbers k and ^P _{k Ifi,} the period of performing the assignment of _{f i} and _{D i.} Further, sorting is performed in the order of the tanks with the highest priority for the product I _fi .

Next, in step 42, and q = 1, attempts to assignment to fill job _{f i} to q-th tank _{T kq.} Consider the following constraints when assigning. The constraint, "the other filling requests to Di-life of the tank T _kq is not assigned", "the Di-life of the tank T _kq, or filling a request Di + 1 period other varieties is not associated (after There is a restriction condition for assigning mixed jobs in In step 43, it is determined whether a constraint condition is satisfied.

When the allocation to the tank T _kq satisfies the constraint condition, the process proceeds to step 44, where the maximum priority P ^max _Ifi <P ^kq _Ifi for the product I _fi is set. Further, the tank T _k is set as the current optimum allocation tank k _opt ← k _q , and the process proceeds to step 46. On the other hand, if the allocation to the tank T _kq does not satisfy the constraint condition, 1 is added to q and step 42 is repeated.

In step 46, a temporary filling job is _assigned to the tank T _kopt . At that time, a penalty g required for allocation is calculated using g _min = C _jkq , _Ifi (2). Here, the current minimum penalty value is g _min ← g. jkp indicates the front oil of tank k, and C _jkq and _Ifi indicate the switching costs from j _kq to product type I _fi .

In step 46, q = q + 1 and temporary allocation is performed. At that time, a penalty g is calculated using g = C _jkq , _Ifi + A · (P ^max _Ifi− P ^kq _Ifi ) (3). Note that A in the formula (3) indicates a penalty coefficient with respect to the priority being reduced by one. Here, if g <g _min is satisfied, the current minimum penalty is gmin ← g. In _addition, the _{k opt} ← _{k q,} accept the _{k q} as the optimal allocation tank.

Next, a procedure for assigning mixed jobs will be described. FIG. 9 is a flowchart showing the mixed job allocation procedure. First, in step 51, the plan start period D _f and the plan end period _De are read. At this time, the tank number to which the mixed job is assigned is expressed by k, and k = 1.

Subsequently, in step 52, and d = _{D f,} the period x and x = d-1 for performing allocation of mixed jobs.

Next, at step 53, it determines whether or not if the filling job f _i Product I _fi is entered schedule of d-life of the tank T _k, constraint conditions shown below are satisfied. Mixing no mixing scheduled appointment x-life of the tank T _k0 that between the tank pipe competition, which is connected to the tank number k - The constraints tank T _k tank number k mixing device M _k each other there is that the use number of times L _Mk of the device M _k does not exceed the maximum number of mixed lot of the day of the mixing device M _k.

That is, this is a case _where the number of use L _Mk of the mixing device M _k does not exceed the maximum number of mixing lots per day, and the tank T _k and the tank T _k0 that competes with each other for the piping between the mixing device M _k and the tank. It is determined that the constraint condition is satisfied (does not violate the constraint condition) when there is no mixture plan in the schedule of the x period.

If the constraint is satisfied, formulates mixed schedule B _Ifi filling job f _i the same amount in the x-life of tank numbers k, rewrites the pre-oil data item d period after the tank number k to I _fi. Further, L _Mk = L _Mk +1 and d = d + 1 are set, and the process proceeds to step 54.

  On the other hand, if the constraint condition is not satisfied, step 53 is repeated with x = x-1. If there is no filling schedule in d period, d = d + 1 and step 53 is repeated. If d = De, k = k + 1 and the process returns to step 52.

In step 55, with reference to the schedule of d-life, the loading amount into the filling job f _i of the filling job f _i if there is B _i of the same product oil as product oil I _fi mixed jobs created in step 56 Add together. Further, d = d + 1. If there is a different product oil filling job, the process returns to step 53. If there is no filling job, step 54 is repeated. If d = De, k = k + 1 and return to step 52. When k is equal to the total number of tanks K ₀ , the allocation of the mixed job is finished.

  FIG. 10 is a production plan Gantt chart showing an example of the production plan created by the scheduling device. The data in the column shows the oil type in the upper part and the amount (unit: kL) in the lower part. The dark column indicates the mixing plan, and the other columns indicate the filling plan. For example, in the target facility “Tank 1”, 230 kL of the oil type “oil 8” is mixed on July 3rd. Subsequent filling plans are 66 kL on the 4th, 84 kL on the 6th, and 80 kL on the 8th.

  In such a scheduling device 10, a priority matrix that represents tank allocation priority is set for each tank and piping connection relationship, piping conflict, mixing device conflict, cleaning amount, product brand, and product allocation priority. Based on this idea, it is possible to avoid conflicts between pipes and mixers so as to satisfy the delivery date of a given filling request, and to assign work based on a priority matrix. This makes it possible to create a schedule for minimizing the amount of cleaning due to product brand change in consideration of lot aggregation while avoiding competition among a plurality of pipes and a plurality of mixing devices. In addition, by creating a production plan with the scheduling device, it is possible to improve the work efficiency of the planner.

Further, the scheduling device 10, it is possible to fill the job allocation order list L _f, obtaining a near-optimal solution in accordance with the computation time allowed using the simulated annealing method, which is one of meta-heuristics, A production plan with reduced costs can be created.

  Moreover, the scheduling apparatus 10 can perform the allocation that minimizes the penalty value for not satisfying the constraint condition for the filling operation and the mixing operation. Thereby, the cost increase which generate | occur | produces by violating a constraint condition can be suppressed.

  In addition, since the scheduling apparatus has a function of fixing the selected work and creating a reschedule, the schedule can be optimized and cost reduction can be realized.

  As mentioned above, although this invention was concretely demonstrated based on the embodiment, this invention is not limited to the said embodiment. In the above embodiment, the scheduling device is used to create a production plan for a lubricating oil mixing plant. However, it may be used to create a production plan in a manufacturing apparatus that produces a product by mixing other raw materials. Moreover, it is good also as a scheduling apparatus made into the mixing assembly line which has a restriction | limiting in a buffer.

  EXAMPLES Hereinafter, although an Example demonstrates this invention still in detail, this invention is not limited to these Examples.

  In the present embodiment, the scheduling device of the present invention is applied to a lubricating oil manufacturing plant of three mixers (mixing devices), 51 tanks, 51 filling lines, and 166 types of total production items (brands). did.

  The mixer ML has two systems, a batch line used when performing batch processing and a Cornell line used when performing continuous processing. In this embodiment, only the batch line is considered. FIG. 11 is a table showing the performance of the mixer. It is assumed that only one of the three mixers ML1 to ML3 is connected to each tank.

  FIG. 12 is a diagram showing tank facility data. The tank facility data in FIG. 12 shows the tank capacity, the usable mixer, and the pipe between the mixer and the tank. There is not one line between the mixer and the tank in each tank, but there are competing lines as shown in FIG. Tanks connected to the same line cannot be mixed at the same time. The filling line is connected to each tank one by one, and there is no difference in capacity between them.

(Decision variable)
The data given when creating the production plan is filling request data. Filling request data is given as a constraint that a given product must be filled by a given amount by the due date. A detailed production plan for each process is determined so that the production plan satisfies the constraints due to filling requirements and equipment. The following variables are used as decision variables in the production plan of the lubricant manufacturing plant. The decision variables are: “work tank”, “product to be mixed and schedule of mixing process”, “mixing amount”, “product to be filled and schedule of filling process”, “filling amount”, “stored in tank Intermediate inventory amount ”and“ whether or not the product stored in the tank is switched ”. The production plan is determined by determining all these variables.

(Objective function)
In the production plan in the lubricant manufacturing factory, the value of the decision variable is determined so as to satisfy the given filling requirement and minimize the cost. Here, the filling request is treated as a condition that must be satisfied, and the item to be optimized when creating the production plan is the production cost. In this embodiment, two costs, a mixing cost and a cleaning cost, are considered as the costs that can be reduced. The mixing cost is a cost required for setup (preparation work) in the mixing process. The cleaning cost is a cost necessary for cleaning the tank when the brand of the product processed last time (pre-oil brand) and the brand of the manufactured product are different.

  13 and 14 are tables showing an example of the cleaning cost. For items not listed in the table, the cleaning cost is zero. In the table, the front oil item is described in the left column, and the production item is described in the upper column.

(Restrictions)
A constraint condition refers to a condition that must be satisfied when creating a production plan. The following constraints of the production plan of the lubricating oil manufacturing plant of this embodiment must be satisfied. As the constraint conditions, there are “a constraint condition related to mixing and filling operations”, “a constraint condition related to pre-oil”, “a constraint condition related to switching”, and “a constraint condition related to storage stock”.

  "Restrictions on mixing and filling operations" include: "The piping connected from the mixer is shared by multiple tanks, so the products that can be mixed at the same time are limited." There is an upper limit on the production capacity (number of mixed lots) per day, and it is not possible to mix beyond that, "" Do not process (produce) more than the required amount in mixing and filling operations " Can be mentioned.

  As the “restriction condition regarding the pre-oil”, “after the product is filled, the remaining oil (pre-oil) of the product manufactured in the past remains in the tank until another product is manufactured”. “Constraining conditions for switching” include “when a product different from the pre-oil is manufactured, a cleaning operation must be performed”. “Restrictions on storage stock” include “each tank can only store products up to the capacity of the tank”.

(Details of formulation)
Next, a detailed description will be given of the formulation of the mixed integer linear programming problem (MILP problem) in the production planning model of the lubricating oil factory of the present embodiment.

(constant)
First, the constants used are shown below.
k: Tank number (1)
C _k : Capacity of tank k (2)
W _i , _j : Cleaning cost required when switching from product type i to product type j (3)
f _t , _i : filling amount of item i in period t (4)
k is a tank number and takes a value of k = 1 to 51. C _k is the maximum capacity of the tank k. _ft , _i is the production request amount (filling request amount) of the product i in the t period. The tank number k (1 to 51) is a numerical value specific to the plant, and is a value corresponding to the number of tanks (51) installed in the plant.

(Decision variable)
The following four variables are defined as decision variables.

Expressions (5) and (6) are decision variables (δ ^k _{0, i} = 0, B ^k _{0, i} = 0) representing the mixing operation. When the decision variable δ ^k _{t, i in the} equation (5) is 1, it represents that the product type i is mixed in the t period of the tank k, and the decision variable B ^k _{t, i in} the equation (6) The amount of mixing is shown.

Equations (7) and (8) are decision variables (γ ^k _{0, i} = 0, F ^k _{0, i} = 0) representing the filling operation. When the decision variable γ ^k _{t, i in the} equation (7) is 1, it represents that the product type i is filled in the t period of the tank k, and the decision variable F ^k _{t, i in} the equation (8) is the filling at that time Represents quantity.

Equation (9) is a decision variable (x ^k _{0, i} = 0) representing the capacity of the product type i in the tank k. Equation (10) is a decision variable (α ^k _{0, i} is given) representing the pre-oil type of tank k. For example, if α ² _3,16 is 1, it indicates that the previous oil item in the third period of tank number 2 is product type number 16. Expression (11) is a decision variable that indicates whether or not switching is performed.

式（１２）の第一項では全混合回数を計算し、第二項では総洗浄コストを表す。式（１２）の目的関数を最小とするような生産計画を作成することを目的とする。 (Objective function)
The production planning problem can be described as a problem that minimizes the following function.

The first term of equation (12) calculates the total number of mixings, and the second term represents the total cleaning cost. The purpose is to create a production plan that minimizes the objective function of equation (12).

(Restrictions on mixing and filling operations)
The constraint conditions handled in this embodiment will be described. In the following equation, T ₀ represents a planning period, iεP ₀ , P ₀ represents a productivity variety set, and kεK ₀ , K ₀ represents an entire tank set. The following are the restrictions on mixing and filling operations. Here, TC is a tank set having a competing relationship between the mixer and the tank. K ₁ , K ₂ , and K ₃ represent tank sets connected to the mixers ML1, ML2, and ML3, respectively.
K ₁ = { _{1, 9, 10,} 12-14, 17-19, 36, 37, 45, 46, 49}
K ₂ = {2-8, 11, 15, 16, 20-35}
K ₃ = {38 to 44, 47, 49 to 51}

式（１３）〜式（１６）は混合作業の制約を表す式である。式（１３）は混合機−タンク間の配管の設備条件によるによる混合回数の制限を表している。式（１４）〜式（１６）は混合機の能力の上限による混合回数の制約を表す式である。式（１４）は混合機ＭＬ１に繋がれているタンクの混合作業の制約を表す式である。式（１５）は混合機ＭＬ２に繋がれているタンクの混合作業の制約を表す式である。式（１６）は混合機ＭＬ３に繋がれているタンクの混合作業の制約を表す式である。 Also, K _C represents a set of tanks that share the same pipe. The constraint conditions are shown below.

Expressions (13) to (16) are expressions representing restrictions on the mixing work. Expression (13) represents the limitation of the number of mixing depending on the equipment condition of the pipe between the mixer and the tank. Expressions (14) to (16) are expressions representing restrictions on the number of times of mixing depending on the upper limit of the mixer capacity. Formula (14) is a formula showing the restrictions of the mixing operation of the tank connected to the mixer ML1. Expression (15) is an expression representing the restriction of the mixing operation of the tank connected to the mixer ML2. Expression (16) is an expression representing the restriction of the mixing operation of the tank connected to the mixer ML3.

式（１７），式（１８）は、上記式（５）の「混合作業を表す決定変数δ^ｋ _ｔ，ｉ」の定義式である。上記式（６）の「混合作業の混合量を表す決定変数Ｂ^ｋ _ｔ，ｉ」が１〜Ｃ_ｋの間の値を取るときのみ、上記式（５）の「混合作業を表す決定変数δ^ｋ _ｔ，ｉ」の値は１を取ることを表している。

Expressions (17) and (18) are definition expressions of “determining variable δ ^k _{t, i} representing the mixing operation” in the above expression (5). Only when the “decision variable B ^k _{t, i} representing the mixing amount of the mixing operation” in the above equation (6) takes a value between ₁ and C _k , the “decision variable δ representing the mixing operation” in the above equation (5). The value of “ ^k _{t, i} ” represents taking 1.

式（１９），式（２０）は、上記式（７）の「充填作業を表す決定変数γ^ｋ _ｔ，ｉ」の定義式である。上記式（８）の「充填作業の充填量を表す決定変数Ｆ^ｋ _ｔ，ｉ」が１〜Ｃ_ｋの間の値を取るときのみ、上記式（７）の「充填作業を表す決定変数γ^ｋ _ｔ，ｉ」の値は１を取ることを表している。

Expressions (19) and (20) are definition expressions of “decision variable γ ^k _{t, i} representing the filling operation” in the above expression (7). "Decision variable F ^{k t} which represents the loading of the filling _{operation, i"} of the above formula (8) only when takes a value between 1 through C _k, the decision variable representing the "filling operation of the above formula (7) gamma The value of “ ^k _{t, i} ” represents taking 1.

式（２１）は充填要求と充填量総和が等しいことを表している。式（２２）は、総混合量と総充填量が等しいことを示している。式（２３）は各タンクで１期間に行える仕事は、混合作業、または充填作業のどちらか１作業のみということを示している。

Expression (21) represents that the filling request and the filling amount sum are equal. Equation (22) shows that the total mixing amount and the total filling amount are equal. Equation (23) shows that the work that can be performed in each tank in one period is only one work of either the mixing work or the filling work.

式（２４）は、各タンクには必ずタンク内に１種類の前油が残っていることを表している。式（２５）はタンクｋのｔ期において製品ｉの混合作業をする場合、タンクｋのｔ期の前油が製品ｉに変わることを示している。式（２６），式（２７）のＫは十分大きい整数を表している。また、ｔ＝１のときの前油種を表す決定変数α^ｋ _{ｔ−１，ｉ}、すなわち決定変数α^ｋ _０，ｉは、生産計画を行う前の０期の前油データを表しており、所与のデータである。これらの式（２６），式（２７）は混合作業がない場合は、一つ前の期の前油がそのままタンク内に残ることを表している。 (Restrictions on pre-oil)
The following formulas (24) to (27) show the constraint conditions for the pre-oil.

Expression (24) indicates that one type of front oil always remains in each tank. Equation (25) indicates that when the product i is mixed in the t period of the tank k, the previous oil in the t period of the tank k is changed to the product i. K in the equations (26) and (27) represents a sufficiently large integer. Moreover, the decision variable α ^k _{t−1, i} representing the previous oil type when t = 1, that is, the decision variable α ^k _{0, i} represents the previous oil data in the zero period before the production plan is performed, Given data. These formulas (26) and (27) represent that the previous oil in the previous period remains in the tank as it is when there is no mixing operation.

式（２８）は、タンクに貯めておく貯蔵在庫の最小値を表している。式（２９）は０期の各タンクの貯蔵在庫がすべて０であることを表している。式（３０）は貯蔵在庫の最大値を表している。式（３１）はタンクｋのｔ期に製品ｉをＦ^ｋ _ｔ，ｉだけ充填を行わなければならない場合、タンクｋのｔ−１期における製品ｉの貯蔵在庫量を表すｘ^ｋ _{ｔ−１，ｉ}は、充填量Ｆ^ｋ _ｔ，ｉ以上でなければならないことを表す。式（３２）は、ｔ−１期末の在庫ｘ^ｋ _{ｔ−１，ｉ}に対して、ｔ期にＦ^ｋ _ｔ，ｉだけ充填され、Ｂ^ｋ _ｔ，ｉだけ混合された場合のｔ期末の貯蔵在庫量を表すものである。式（３３）はタンクｋのｔ期における前油が製品ｉのときタンク内には製品ｉ以外の貯蔵在庫量がないことを表している。 (Restrictions on storage stock)
The following are the constraints on storage stock.

Expression (28) represents the minimum value of the storage stock stored in the tank. Expression (29) represents that the storage stock of each tank in the zero period is all zero. Expression (30) represents the maximum value of the storage stock. Equation (31) indicates that if product i has to be filled by F ^k _{t, i} in t period of tank k, x ^k _{t-1, i} indicates that the filling amount must be greater than or equal to F ^k _{t, i} . Equation (32) shows that the stock at the end of t-1 x ^k _{t-1, i is} filled at the end of t by F ^k _{t, i and} mixed at the end of t when B ^k _{t, i is} mixed. It represents the stock quantity. Expression (33) represents that when the previous oil in the t period of the tank k is the product i, there is no storage stock amount other than the product i in the tank.

式（３４），式（３５）は、前油種を表す決定変数α^ｋ _{ｔ−１，ｉ}，α^ｋ _ｔ，ｊがともに１である場合、切り替えの有無を表す決定関数β^ｋ _{ｔ，ｉ，ｊ}の値を１とすることを表している。すなわち、これら二つの式（３４），式（３５）は、タンクｋのｔ期末において、製品ｉから製品ｊに製造製品の切り替えの有無を表している。 (Restrictions on switching)
The restriction conditions for switching are shown below.

Expressions (34) and (35) are determined by the decision function β ^k _{t, i} indicating whether or not switching is performed when the decision variables α ^k _{t-1, i} and α ^k _{t, j} representing the previous oil type are both 1. _{, J} is set to 1. That is, these two formulas (34) and (35) indicate whether or not the manufactured product is switched from the product i to the product j at the end of the period of the tank k.

(Assignment priority matrix)
Conventionally, an operator of a lubricant manufacturing factory has decided a certain tank to be used for manufacturing each product in consideration of a later production plan. The allocation priority matrix is a matrix that reflects such job allocation criteria in numerical values. 15 and 16 are diagrams showing an allocation priority matrix according to the present embodiment. In FIG.15 and FIG.16, the tank number is shown in the left column, and the product oil kind is shown in the upper stage. The numerical values in the figure represent “3: always assigned”, “2: occasionally assigned”, and “1: not assigned”.

(Experimental result)
In this example, the scheduling apparatus described in the above embodiment is applied to a production plan model for a lubricant manufacturing factory. Here, a numerical experiment was performed using a computer of CPU: Celeron (trade name) 2 GHz, memory: 512 MB. As described above, FIG. 10 shows a Gantt chart of a production plan based on experimental results.

  In the scheduling device of the present invention, monthly filling data could be created in about 30 minutes. Conventionally, it takes about 15 hours when a production planner creates monthly filling data. Further, in the scheduling device of the present invention, when preparing the blending data based on the filling data and creating the allocation plan for the equipment, it can be performed in about 30 minutes. Conventionally, it takes about 6 hours for a production planner to create an allocation plan for equipment. As a result, it is possible to shorten the monthly work time of the person in charge of production planning by about 20 hours. In other words, a reduction in work time of about 240 hours per year can be achieved.

  Further, in the scheduling device of the present invention, in the weekly production plan, when the blending data is created based on the filling data and the allocation plan to the equipment is created, it can be performed in about 30 minutes. Conventionally, it took about two hours for a production planner to create an allocation plan for equipment. As a result, it is possible to shorten the work time for one week of the production planner by about 1 hour 30 minutes. In other words, a reduction in work time of about 72 hours can be achieved annually.

DESCRIPTION OF SYMBOLS 1 ... Lubricating oil mixing plant, 4 ... Filling device, 5 ... Drum can, 6 ... Roller, 10 ... Scheduling device, 11 ... Input part, 12 ... Output part, 20 ... Electronic control unit, 21 ... Schedule calculating part, 31 ... Restriction Condition DB, 32 ... Job DB, 33 ... Product tank DB, 34 ... Product DB, T _{A to} T _C ... Raw material tank, T _{1 to} T ₄ ... Product tank, L _{1 to} L ₆ ... Pipe, M _k ... Mixing device , M ₁ ... continuous processing apparatus, M ₂ ... batch blender.

Claims (8)

A scheduling device for creating a work schedule in a mixing plant having a plurality of tanks, a mixing device, a pipe connecting the tank and the mixing device,
A priority matrix setting means for setting a priority matrix indicating the allocation priority of the tank for each product;
Schedule calculation means for creating a work schedule by assigning work based on the priority matrix using a backward method that avoids competition of the piping and competition of the mixing device so as to satisfy the delivery date of the product A scheduling apparatus comprising:   2. The schedule calculation means is capable of obtaining a nearly optimal solution according to an allowable calculation time using a simulated annealing method, and creating an assignment order list of the work. Scheduling device. The priority matrix setting means expresses the priority matrix with three kinds of integers,
The priority matrix is represented by a first integer that represents a situation that must be assigned, a second integer that represents a situation that can be assigned, and a third integer that represents a situation that cannot be assigned. The scheduling apparatus according to claim 1 or 2, characterized in that: The scheduling calculation means allocates the work so that a penalty value for not satisfying the constraint condition in the plant is minimized.
The scheduling apparatus according to claim 1, wherein the penalty value is set according to an increase in cost that occurs due to not satisfying the constraint condition.   The scheduling apparatus according to claim 1, wherein the scheduling calculation unit fixes the selected work and creates a reschedule.   The scheduling calculation means distinguishes the mixing amount and mixing timing in the tank into a general-purpose tank and a dedicated tank, considering the minimum mixing amount and the maximum mixing amount of the tank, so that the cleaning amount is minimized. The scheduling apparatus according to claim 1, wherein a work schedule for mixing, inspection, and cleaning is created. Piping data storage means for storing piping data, which is data relating to the piping connection relationship, piping competition between multiple brands, and flow velocity in the piping;
The scheduling apparatus according to claim 1, wherein the schedule calculation unit creates a work schedule in consideration of piping competition between a plurality of brands and the flow velocity. It further comprises brand data storage means for storing brand data that is data relating to a plurality of brands,
The scheduling apparatus according to claim 1, wherein the schedule calculation unit additionally updates the work schedule in accordance with addition or change of the brand data.

Images