JPH09147017A - Production planning system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make an optimum production plan by finding the restriction satisfying degree from the production start data and finding a production device from the production possibility, updating the state is a producing start data NN(neural network) and production NN and thereby changing the production stardata and the production device for each lot according to the assignment state if to other lots. SOLUTION: A dummy lot production means 2 generated the dummy lost of the same number of production devices for each lot against each lot based on the inputted order reception data and production capability data. A production start data NN 3 finds the production start date of each dummy lot, and a production start data NN output evaluation means 4 finde the restriction satisfying genre the formed satisfying degree. A producing device NN 5 decides the prosecution possibility of every production device based degree of each dummy lot based on the output of the NN 3 and output the limit sufficiency 3. A production device NN output evaluation means 6 fined the production service for each lot based on the output of the NN 5 and outputs the producing device information. Based on the output of the means 6, the NN 3 updates the production start data of each dummy lot.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is capable of producing each lot with a plurality of production apparatuses, and in each of the production apparatuses, one production apparatus is used to produce a production plan. The present invention relates to a production planning system that determines devices.

[0002]

2. Description of the Related Art FIG. 11 is disclosed in Japanese Unexamined Patent Publication No. 4-290162 or a collection of papers of the Society of Instrument and Control Engineers Vol. 31,
No. 7 (July 1995) is a flowchart showing a processing flow of the conventional production planning system shown in “Practical use method of plant operation scheduling system”. In the conventional production planning system, as shown in the flowchart of FIG. 11, first, the allocation order for the production devices of all lots is determined (step S110), and the production devices of each lot are determined according to the determined allocation order (step S111). Next, the lot processing order is determined for each device (step S112), and the production start time of each lot is determined in consideration of various constraint conditions according to the determined processing order (step S113). Further, for example, in the case of re-execution after changing the production equipment of the lot or changing the production start time of the lot using the manual correction function, the production equipment is not re-determined but only the production start time is re-determined. ing.

[0003]

Generally, if both the production start time and the production apparatus are determined at the same time and the production apparatus can be changed in the middle of the planning, the solution search range becomes enormous, and the conventional method is difficult. The procedural search method makes it difficult to make a plan within a practical time. Therefore, in the prior art, for example, the production start time of each lot is determined after first determining the production device of each lot. However, in such a method, the production equipment is fixed first, so that the allocation conditions of other lots can be considered in the middle of the planning process, and the constraint conditions such as delivery date and setup change can be most satisfied. There was a problem that the equipment could not be selected, and the planning result obtained would depend on the decision of the first production equipment. In addition, although it is possible to change the production equipment of the lot by the manual correction function after planning, the re-execution function after manual correction also
The lot with the changed production equipment remains fixed to the production equipment after the change, and the other lots remain fixed to the initially determined production equipment.It is not possible to change the production equipment of each lot in order to meet more constraint conditions. There was a problem that it could not be done.

The present invention is for solving the above-mentioned problems, and makes an optimum production plan while automatically changing the production equipment of each lot according to the allocation situation of other lots. The purpose is to obtain a production planning system that can.

[0005]

A first aspect of the invention is to input order data and production capacity data of each production apparatus, and produce each lot based on the input order data and production capacity data. Dummy lot generation means for generating the same number of dummy lots as the number of available devices, and setting, for each dummy lot, a production lot, a producible device of this lot, and an initial value of a production start date, and this dummy lot From the production start date neural network (hereinafter, the neural network is referred to as NN) for updating the production start date of each dummy lot set by the lot generation means, and the production start date of each dummy lot updated by this production start date NN Production start date NN output evaluation means for obtaining the constraint satisfaction degree of each dummy lot, and each set by the dummy lot generation means Production device NN that production apparatus capable of Mirotto determine the likelihood that the each dummy lot is produced
And a production device NN output evaluation means for obtaining a production device for each lot based on the producibility of each dummy lot obtained by the production device NN, wherein the production start date NN is the production device The production start date of each dummy lot is updated based on the production device of each lot obtained by the NN output evaluation means, and the production device NN
The production possibility of each dummy lot is obtained based on the constraint satisfaction degree of each dummy lot obtained by the production start date NN output evaluation means.

A second aspect of the present invention is a convergence for determining whether to repeatedly execute the production start date NN, the production start date NN output evaluation means, the production device NN, and the production device NN output evaluation means. It is provided with a judging means.

A third aspect of the invention is provided with a convergence determination means for performing the determination based on whether or not the output of the production start date NN or the output of the neuron constituting the production apparatus NN has continuously changed a predetermined number of times. is there.

A fourth aspect of the invention is provided with a convergence determination means for performing the determination based on whether or not a predetermined time has elapsed.

A fifth aspect of the invention is provided with a convergence determination means for performing the determination based on whether or not the determination has been performed a predetermined number of times.

A sixth aspect of the present invention is based on the production start date of each of the dummy lots updated by the production start date NN and the production device of each lot obtained by the production device NN output evaluation means. It is provided with a planning result display means for displaying the production start date for producing each lot and the production apparatus.

According to a seventh aspect of the present invention, the production start date NN displayed on the plan result display means or the plan result correction means for correcting the production apparatus, and the production start date NN based on the correction result by the plan result correction means. Alternatively, the NN output changing means for changing the states of the plurality of neurons constituting the production apparatus NN is provided.

An eighth aspect of the present invention is a suspension device for suspending production and a suspension designating means for designating information on a suspension period, information designated by the suspension designating means and the production device NN.
Production start date NN for updating the production start date of each dummy lot based on the production device of each lot obtained by the output evaluation means, information designated by the suspension designation means, and the production start date NN output evaluation And a production device NN for obtaining the productivity of each dummy lot based on the constraint satisfaction degree of each dummy lot obtained by the means.

A ninth aspect of the present invention is a parameter changing means for changing a constraint condition parameter of a production plan, the constraint condition parameter changed by the parameter changing means, and each lot obtained by the production device NN output evaluating means. The production start date NN for updating the production start date of each dummy lot based on the production device, the constraint condition parameter changed by the parameter changing means, and the production of each dummy lot obtained by the raw device NN. The production device NN output evaluation means for obtaining the production device of each lot based on the possibility.

A tenth aspect of the invention is to set the production lot and the producible device set in each of the dummy lots to a plurality of neurons constituting the production start date NN,
The production start date in the case of connecting the neurons having different lots set and the same producible device set to each other and producing the set production lot with the set producible device in each neuron The production start date NN to be obtained is provided.

An eleventh aspect of the invention is a production apparatus NN output evaluation means for determining the dummy lot having the maximum producibility of each lot obtained by the production apparatus NN as a true lot to be actually produced, For each neuron, if the two connected neurons are true lots, the production start date is calculated so that the production periods of the lots do not overlap each other, based on whether or not the determined true lots are true. ,
When one of the two connected neurons is not a true lot, the production start date NN for obtaining the production start date while allowing the production periods of the lots to overlap is provided.

A twelfth aspect of the invention is to set the production lot and the producible device set in each of the dummy lots to a plurality of neurons constituting the production device NN, and the same lot is set and different production is performed. A production device NN that connects the neurons in which the production device is set, and asks each neuron to determine the producibility when the set production lot is produced by the set production device. It is a thing.

In a thirteenth aspect of the present invention, the constraint satisfaction degree of each dummy lot obtained by the production start date NN output evaluating means is set to each of the neurons, and among the plurality of connected neurons, The present invention is provided with a production device NN that finds a neuron that maximizes the set constraint satisfaction level and increases the productivity of the found neuron.

In a fourteenth aspect of the present invention, the constraint satisfaction degree of each dummy lot obtained by the production start date NN output evaluation means is set for each neuron, and among the plurality of connected neurons, the constraint satisfaction degree is set. The present invention is provided with a production device NN that finds a neuron that maximizes the set constraint satisfaction level and reduces the productivity of the other neurons than the found neuron.

[0019]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1 FIG. Embodiment 1 of the production planning system of the present invention will be described in detail below with reference to the drawings. 1 is a block diagram of a production planning system according to a first embodiment of the present invention. In FIG. 1, reference numeral 1 is a data input means for inputting order data and production capacity data of a plurality of production apparatuses for producing each lot, and 2 is based on the order data and the production capacity data input by the data input means 1. , Dummy lot generation means for generating the same number of dummy lots as the number of producible devices of each lot, 3 is a production start date NN for obtaining and outputting the production start date of each dummy lot, 4 is a production start date The production start date NN output evaluation means 5 for obtaining and outputting the constraint satisfaction degree of each dummy lot based on the output of the day NN3 is based on the output of the production start date NN output evaluation means 4 for each producible device of each dummy lot. The production apparatuses NN and 6 which output for the production possibility, obtain the production apparatus of each lot based on the output of the production apparatus NN5, and output the production apparatus information.
The production start date NN3 is an N output evaluation means, and the production start date NN3 updates the production start date of each dummy lot based on the output of the production apparatus NN output evaluation means 6. Reference numeral 7 is a convergence determination means for performing the convergence determination of the production start date NN3 and the production apparatus NN5, and 8 is a planning result display means for displaying the production planning result.

Next, the operation of the production planning system of the present invention will be described with reference to specific examples. For simplicity, it is assumed that the number of lots is 6, the number of production apparatuses is 3, the lots are numbered 1 to 6, and the production apparatuses are numbered A to C.

The data input means 1 inputs order data and production capacity data. The order data includes, for example, the production amount of each lot, the delivery date of each lot, the list of producible devices that shows the devices that can produce each lot for each lot, and the color and size of the product corresponding to the lot. Contains data. An example of the list of producible devices is shown in FIG. In the list of producible devices in FIG. 2, the lot with the lot number 1 can be produced with the three production devices with the device numbers A, B, and C, and the lot with the lot number 2 can be produced with the two production devices with the device numbers A and C. It shows that it can be produced by the device. In addition, the production capacity data indicates the maximum production amount that each production device can produce in one day.

The dummy lot generation means 2 generates a dummy lot based on the order data and the production capacity data input by the data input means 1, and gives a lot number and a device number to each dummy lot. Then, the lot number and the device number of each dummy lot are output to the production start date NN3 and the production device NN5. For example, according to the list of producible devices in FIG. 2, since lot 1 can be produced by three devices A, B, and C,
3 dummy lots are generated for lot 1 and 3
Lot number 1 is given to all three dummy lots, and device numbers A, B, and C are given to each of the three dummy lots. The dummy lot thus generated is shown in FIG. Dummy lots D1A, D1B, D1 in FIG.
C is a dummy lot of the lot 1. Each of the dummy lot D1A has the lot number 1 and the device number A, the dummy lot D1B has the lot number 1 and the device number B, and the dummy lot D1C has the lot number 1 and the device number. C is given respectively. Dummy lots are similarly generated for lots 2 to 6, and lot numbers and device numbers are given to the dummy lots.

Here, the word true lot is defined. Although a plurality of dummy lots are generated for each lot in the dummy lot generation means 2, each lot is actually produced by one of a plurality of producible devices. A dummy lot having a production device number determined to produce each lot is called a true lot. For example, if it is determined that the lot 1 is produced by the device A, the dummy lot D1
A is a true lot, and the remaining dummy lots D1B and D1C
Is not a true lot.

Further, the true lot indicates that the lot is actually produced, and the dummy lot that is not the true lot does not actually produce the lot. Therefore, true lots having the same device number should never overlap in production period. This is because if the production periods overlap, one production apparatus will simultaneously produce a plurality of lots. In contrast, dummy lots are not actually produced, so dummy lots
The true lot and the dummy lot may be duplicated.

FIG. 4 shows an example of the relationship between the true lot and the other dummy lots. The horizontal axis of FIG. 4 represents time.
Dummy lots D1A, D2A, D5A, D in FIG.
6A, D3B, and D4B are true lots, and the other dummy lots are not true lots. In this case, true lots D1A, D2A, D5A, and D6A having the same device number A do not overlap each other, and true lots D3B and D4B having the same device number B do not overlap each other. Dummy lots are allowed to overlap with each other, or dummy lots and true lots are allowed to overlap with each other.

Next, the structure of the production start date NN3 will be described with reference to FIG. Each neuron on the production start date NN3 has a one-to-one correspondence with a dummy lot, is given the same lot number and device number as the corresponding dummy lot, and outputs the production start date of the corresponding dummy lot. For example, the neuron 11A outputs the production start date of the dummy lot D1A, and the neuron 11B outputs the production start date of the dummy lot D1B. In determining the production start date of each dummy lot, it is necessary to consider avoiding duplication of true lots on the same production equipment, but since it is not affected by dummy lots on different production equipment, In each neuron, neurons corresponding to dummy lots assigned to the same production device have connections, and neurons corresponding to dummy lots assigned to different production devices do not have connections. In FIG. 5, for example,
The neurons 11A, 12A, 13A, 15A, 16A corresponding to the dummy lots assigned to the device A are connected to each other, and these neurons are the neurons 11B corresponding to the dummy lots assigned to the different production devices. It has no bond with 13B. If a parallel computer is used, the determination of the production start date for each production device can be processed in parallel.

Next, the operation of the production start date NN3 will be described. The production start date NN3 receives from the production device NN output evaluation means 6 information as to which production device each lot is produced by, that is, which lot is a true lot in each dummy lot, and the same production device NN3 is received. If the dummy lots assigned to each are true lots, the production periods do not overlap.If at least one dummy lot is not a true lot, the production periods are allowed to overlap and all dummy The production start date of each dummy lot is updated so that the lot satisfies the planning constraints. Specifically, these processes are performed by the following steps.

[Step 1] For each true lot, among the other dummy lots (including the true lot) on the same production apparatus, the one that has the highest constraint satisfaction degree when the production is performed next to each true lot. Choose one. At this time, there is a one-to-one correspondence between each true lot and the selected dummy lot. The same dummy lot is never selected for multiple true lots. When the true lot is x and the dummy lot is y, the constraint satisfaction S
y is calculated by the following equation (1). Sy = Pd × Dxy + P1 × (T−Ly) / T (1) The first term on the right side of the equation (1) is for determining the satisfaction degree of the setup change constraint, and Pd is a parameter indicating the strength of the setup change constraint. , Dxy is 0 when setup change occurs between the true lot x and the dummy lot y, and 1 when setup change does not occur. In the second term on the right side of the equation (1), the degree of sufficiency of the delivery date constraint is obtained, Pl is the value of the parameter indicating the strength of the delivery date constraint, T is the number of days in the entire planning period, Ly is the dummy lot y, and the true lot x. Is the number of delivery delay days when the production is started from the day after the production end date, and the value of Ly is 0 when the production is finished earlier than the delivery date. If there is no setup change and there is no delay in delivery, Sy is the maximum value Pd + P
Take l.

[Step 2] The output of the neuron corresponding to the dummy lot paired with the true lot in step 1 is updated, that is, the production start date is updated. The output of the neuron corresponding to the dummy lot z paired with the true lot x is V
z and the internal state are Uz, the output is updated by equation (2),
It is given by equation (3). Uz = Uz + Pa * (Ex-Bz + 1) (2) Vz = g (Uz) (3) Formula (2) is a formula for updating the internal state of the neuron, and Ex
Is the production end date of the true lot x, Bz is the production start date of the dummy lot z, and Pa is a coefficient, and brings the production start date of the dummy lot z closer to the day after the production end date of the true lot x. How close they are to each other is determined by the second term in equation (2). In Expression (3), g (Uz) means that the fractional part of the internal state of the neuron is truncated, and the truncated value is the production start date of the dummy lot x.

[Step 3] The output of the neuron corresponding to the dummy lot w that has not paired with the true lot in step 1 is updated as follows. The output of the neuron corresponding to the dummy lot w is updated so that the production start date is earlier than the dummy lot w and the day after the production end date of the true lot q closest to the dummy lot w is close to the production start date of the dummy lot w. To do. When the output of the neuron corresponding to the dummy lot w is Vw and the internal state is Uw, the state update is given by equations (4) and (5). Uw = Uw + Pa * (Eq-Bw + 1) (4) Vw = g (Uw) (5) In formula (4), Eq is the production end date of the true lot q, B
w is the production start date of the dummy lot w, and the internal state is updated so that the production start date of the dummy lot w approaches the production end date of the true lot q. In Expression (5), g (Uw) means that the fractional part of the internal state of the neuron is truncated, and the truncated value is the production start date of the dummy lot w. All neurons on the production start date NN3 have 1
The above steps 1 to 3 are repeated so that the output is updated each time.

Next, the structure of the production apparatus NN5 will be described with reference to FIG. Each neuron of the production device NN5 has a one-to-one correspondence with a dummy lot, and is given the same lot number and device number as the corresponding dummy lot, and the corresponding dummy lot may become a true lot, that is, a dummy lot. Outputs the possibility that the lot corresponding to the lot number held by will be produced by the production device corresponding to the device number of the dummy lot. The productivity has a value of 0 to 1, and the higher the value, the higher the possibility of being produced. For example, lot 1
, The dummy lots 1-A, 1-B, 1-C are assigned to the dummy lots 1-A, and the dummy lots 1-A have neurons. 21
It is assumed that A corresponds to the neuron 21B in the dummy lot 1-B and the neuron 21C corresponds to the dummy lot 1-C. At this time, the output of the neuron 21A is 0.9,
If the output of the neuron 21B is 0.1 and the output of the neuron 21C is 0.1, the lot 1 is most likely to be produced by the device A.
And Since each lot is produced by one production device,
For each lot, the productivity of one producible device is increased, and the producibility of the remaining producible devices is decreased.
Moreover, since the productivity is independently obtained for each lot, different lots do not affect each other's determination of the productivity. Therefore, neurons having the same lot number have connections, and neurons having different lot numbers do not have connections. It should be noted that if a parallel computer is used, the determination of the production device for each lot can be processed in parallel.

Next, the operation of the production apparatus NN5 will be described. The production apparatus NN5 has a production start date NN output evaluation section 3
Receives the constraint satisfaction of each dummy lot, selects the one with the highest constraint satisfaction among the dummy lots given the same lot number, and increases the productivity of neurons corresponding to that dummy lot. It reduces the productivity of neurons corresponding to other dummy lots having the same lot number. Specifically, these processes are performed by the following steps. [Step 1] The maximum value of the constraint satisfaction degree of the dummy lot having the same lot number and the dummy lot having the maximum value are obtained. [Step 2] Let SMAXi be the maximum value of the constraint satisfaction degree of the dummy lot having the lot number i, and j be the device number of the dummy lot having the maximum value.
The update of the output of the neuron corresponding to the dummy lot having is carried out using the equations (6) to (9).

[0033]

(Equation 1)

Uik is the internal state value of the neuron corresponding to the dummy lot with lot number i and device number k, and is Vik.
Is the output value of the neuron corresponding to the dummy lot of lot number i and device number k, and Pb is a coefficient. In Equation (6), if the dummy lot k is a dummy lot (k = j) that takes the maximum constraint satisfaction degree, the internal state of the neuron corresponding to the dummy lot k is multiplied by the maximum constraint satisfaction degree SMAXi multiplied by the coefficient Pb. If the value is increased and the dummy lot k is a dummy lot (k ≠ j) that does not take the maximum constraint satisfaction degree,
The internal state value of the neuron corresponding to the dummy lot k is decreased by a value obtained by multiplying the maximum constraint satisfaction degree SMAXi by the coefficient Pb. This formula increases the manufacturability of the lot i in the manufacturable devices j and decreases the manufacturability of the lot i in the manufacturable devices other than j. Expression (7) uses the sigmoid function (Expression (8)) to find the output of the neuron from the internal state value. The steps 1 and 2 are performed so that all the neurons of the production apparatus NN5 update the output once.
Is repeated.

Next, the production start date NN output evaluation means 4 will be described. The production start date NN output evaluation means 4 receives the output of the production start date NN3 and obtains the constraint satisfaction degree of each dummy lot. S'is the degree of constraint satisfaction of dummy lot y
y and the dummy lot y have the same device number, the production start date is earlier than the dummy lot y, and the true lot closest to the dummy lot y is the true lot q, S'y is given by the formula (1
0) is used. S′y = Pd × Dqy + Pl × (T−L′y) / T (10) The first term on the right side of the equation (10) seeks the sufficiency of the setup change constraint, and Pd is the strength of the setup change constraint. The value Dqy of the parameter indicating the size is 0 when the setup change occurs between the true lot q and the dummy lot y, and 1 when the setup change does not occur. The second term on the right side of the equation (10) seeks the degree of sufficiency of the delivery time constraint, Pl is the value of the parameter indicating the strength of the delivery time constraint, T is the number of days in the entire planning period, and L'y is the dummy lot y. The number of delivery days that are delayed when production starts from the production start date, and L'when production ends earlier than the delivery date.
The value of y is 0. The difference between the formula (1) and the formula (10) is that the formula (1) finds the constraint satisfaction level when the dummy lot starts production from the day after the production end date of the true lot. The point 10) is that the constraint satisfaction degree when the production is started from the production start date of the dummy lot is obtained. The production start date NN output evaluation means 4 obtains constraint satisfaction levels for all dummy lots, and outputs these values to the production device NN5.

Next, the production device NN output evaluation means 6 will be described. The production device NN output evaluation means 6 receives the output of the production device NN5, selects among the neurons having the same lot number the maximum output, that is, the one having the highest production possibility, and selects the dummy lot corresponding to the neuron. Is a true lot. True lots are selected for all dummy lots, and information on which dummy lot is a true lot, that is, information on which production apparatus produces each lot is output to the production start date NN3. For example, the production equipment N corresponding to the dummy lots 1-A, 1-B, and 1-C of the lot 1
The output of the N5 neuron is 0.9, 0.1,
If the value is 0.1, the output of the neuron corresponding to dummy lot 1-A is the largest, so dummy lot 1
-A is a true lot, and the set of lot number 1 and device number A is output on the production start date NN3.

Next, the convergence determining means 7 will be described. First, the production start date NN3 and the state update of the production apparatus NN5 will be defined. The state update on the production start date NN3 means that all the neurons on the production start date NN3 are output once and updated. Similarly, the production device NN5
The state update of means that all the neurons of the production apparatus NN5 are output once and updated. In the convergence determination means 7, each time the production start date NN3 and the production device NN5 are updated once each, the production start date NN3 and the production device N are updated.
The output changes before and after the state update of all the neurons of N5 are examined, and if there is no change of the output of all the neurons before and after the state update for a certain number of times in succession, the production start date NN3 and the state of the production apparatus NN5 Finish the update.

Next, the planning result display means 8 will be described. The planning result display means 8 receives information on which dummy lot is a true lot from the production apparatus NN output evaluation means 6, that is, which production apparatus is used to produce each lot, and the true lot from the production start date NN3. Receive the production start date of and display the production plan. In addition, information such as how late the planned plan is behind delivery and how many setup changes occur is also displayed.

Next, the flow of processing for production planning will be described with reference to the flowchart of FIG. First, in step S1, data is input. That is, the data input means 1 inputs the order data and the production capacity data. Step S2
Then, the dummy lot is generated. That is, the dummy lot generation means 2 generates a dummy lot, and outputs the lot number and the device number of each dummy lot to the production start date NN3 and the production device NN5. In step S3, the initial state is set. That is, the dummy lot generation means 2 sets the output of each neuron on the production start date NN3 so that the production of all the dummy lots will be finished by the delivery date, ignoring the duplication of the lots, and the production device of each lot. Is the first device in the list of producible devices, the production device NN
The output of each neuron of 5 is set. In step S4, the production device NN output evaluation means 6 determines the production device of each dummy lot in the initial state.

In step S5, the production start date NN3 is updated once by using the production device information of each lot obtained by the production device NN output evaluation means 6. That is, the outputs of all neurons on the production start date NN3 are updated once.
In step S6, the production start date NN output evaluation means 4 calculates the constraint satisfaction degree of each dummy lot based on the output of the production start date NN3. In step S7, the state of the production apparatus NN5 is updated once using the constraint satisfaction degree obtained by the production start date NN output evaluation means 4. That is, the outputs of all the neurons of the production apparatus NN5 are updated once. In step S8, the production device NN output evaluation means 6 determines the production device of each lot based on the output of the production device NN5 after the state update.

In step S9, the convergence determination means 7 determines whether to end the state update of the production start date NN3 and the two NNs of the production apparatus NN5. When it is determined that the status update is continued, the process returns to step S5, and when it is determined that the status update is completed, the process proceeds to step S10. In step S10, the planning result is displayed.

As described above, according to the present embodiment, the production start date NN output evaluation means 4 obtains the constraint satisfaction degree from the production start date output by the production start date NN3 to obtain the production device NN.
5, the production device NN output evaluation means 6 is the production device NN.
The production device is obtained from the production possibility output by No. 5 and is given to the production start date NN3, and the state update of the production start date NN3 and the production device NN5 is alternately repeated, so that each lot can be assigned according to the allocation status of other lots. It is possible to determine the production start date and production equipment of.

In this embodiment, the list of producible devices for each lot and the production capacity data of each production device are input by the data input means 1. However, the list of producible devices and the production capacity data of each production device are input. It is also possible to prepare in advance and not to input by the data input means 1.

Embodiment 2 FIG. 8 is a configuration diagram of the production planning system according to the second embodiment of the present invention. In FIG. 8, 1 to 8 are the same as those in the first embodiment. 9 is a planning result correction means for the operator to change the production start date or the production device with respect to the planning result displayed on the planning result display means 8, and 10 is a production start date based on the changing operation by the planning result correction means 9. NN output changing means for changing the state of the corresponding neuron of the NN3 or the production device NN5, 1
1 indicates that the operator specifies the suspension device for suspending the production and the information regarding the suspension period, and the production start date NN3 and the production device N
The operator designates a suspension condition designating means for outputting to N5, 12 designates a modification of the constraint condition parameter of the production plan, and a parameter modification for outputting the modified constraint condition parameter value to the production start date NN3 or the production device NN output evaluating means 6. It is a means.

First, the planning result correction means 9 will be described. When the operator wants to partially change the drafted result, the drafting result correction means 9 can change the production start date of the lot or the production device on the screen. For example,
If the operator first determines that it is better to produce the lot 1 in the device B, assuming that the lot 1 is produced in the device A, the lot 1 is displayed on the screen by the planning result correction means 9. It is possible to obtain a more desirable planning result for the operator by changing the production apparatus from No. A to No. B and re-executing the planning after the change.

Next, the NN output changing means 10 will be described. When the production start date of a lot or the production apparatus is changed by the planning result correction means 9, the production start date and the production possibility of the dummy lot corresponding to the changed lot, that is, the production start corresponding to the changed lot The internal states and outputs of the neurons of the day NN3 and the neurons of the production device NN5 must be changed. The NN output changing means 10 changes the internal state and output of each neuron. For example, in the case where the planning result that the production of the lot 1 is started in the device A from April 1st by the planning result correction means 9 such that the production of the lot 1 is started in the device B from March 25th, In the NN output changing means 10, the lot number 1,
Production start date N corresponding to the dummy lot with device number B
The output of the neuron of N3 is changed so that the production start date is March 25, and the output of the neuron of the production device NN5 corresponding to the dummy lot having the lot number 1 and the device number A and the lot number 1 and the device number are changed. The outputs of the neurons of the production apparatus NN5 corresponding to the dummy lot having B are exchanged.

Next, the suspension designating means 11 will be described. The suspension device designation means 11 can designate and change the production device to be suspended and the suspension period when there is a production device to be suspended for part or all of the planned period.
The designation and change can be performed when the initial state is set or when the planning result is corrected, and the suspension device number and the suspension period are output to the production start date NN3 and the production device NN5.

Next, the parameter changing means 12 will be described. The parameter changing unit 12 can change the value of the parameter indicating the strength of the constraint condition, and the changed parameter value is used for planning or re-executing the planning, and the production start date NN3 or It is output to the production device NN output evaluation means 6. The constraint condition parameters include the parameter Pd indicating the strength of the setup change constraint shown in the equations (1) and (10) of the first embodiment and the parameter Pl indicating the strength of the delivery date constraint. In the initial state, when the parameter Pd representing the strength of the setup change constraint is set to a large value, for example, Pd = 0.8, and the parameter Pl representing the strength of the delivery time constraint is set to a small value, for example, Pl = 0.2, the planning is performed. , You can obtain the planning result that gives priority to the setup change constraint, but if you want to make the planning that gives priority to the delivery date constraint, change the value of each parameter on the screen, for example, Pd = 0.2, Pl = 0.8. By re-executing the planning, it is possible to obtain the planning result that prioritizes the delivery date constraint.

Next, the flow of the process of designating the suspension device and the constraint condition parameters at the time of initial setting of the planning will be described with reference to the flowchart of FIG. Steps S1 to S3
Performs the same processing as steps S1 to S3 shown in FIG. 7 of the first embodiment. Next, in step S31, the operator determines whether or not to specify the suspension device. If the suspension device is to be specified, the operation proceeds to step S32, and the suspension specifying means 11 performs the operation to specify the suspension device. If it is not necessary to specify the suspension device, the process of step S32 is not performed.

In step S33, the operator determines whether or not to change the parameter value indicating the strength of the constraint condition, and if the parameter value is to be changed, it is determined in step S33.
In step 34, the parameter changing means 12 changes the parameter value. If it is not necessary to change the parameter value, the process of step S34 is not performed. After this,
Processing similar to steps S4 to S10 of FIG. 7 is performed.

Next, the flow of processing for correcting and re-executing the planning result after the planning will be described with reference to the flowchart of FIG. First, in step S101, the operator determines whether to correct the planning result. When it is determined that the correction is performed, the process proceeds to step S102, the planning result correction unit 9 changes the production device of the lot and the production start date, and based on this change, the NN output changing unit 10
The state of the corresponding neuron of the production start date NN3 or the production device NN5 is changed. If the planning result is not corrected, the process of step S102 is not performed.

Next, in steps S103 to S106,
Similar to steps S31 to S34 of FIG. 9, the operation of changing the suspension device and the operation of changing the constraint condition parameter are performed as necessary. In step S107, the production device NN output evaluation means 6 determines the production device of each lot in the state where the planning result is corrected. After this, steps S5 to S5 in FIG.
The same process as S10 is performed.

As described above, according to the present embodiment, it is possible to specify and change the suspension device for suspending the production and the suspension period by the suspension designating means 11. Further, the parameter changing means 12 can change the parameter indicating the strength of the constraint condition of the production plan. Further, the planning result correcting means 9 is provided with a function for correcting the planning result, and after making a correction such as a change in the production apparatus of the lot or a production start date of the lot, the NN output changing means 10 changes the corrected state. By making a change to the initial state and performing the same processing as at the time of planning, it is possible to perform re-planning based on the modification designation of the planning result.

Embodiment 3 FIG. The convergence determination means 7 of the first embodiment ends the state update after the production start date NN3 and the outputs of the neurons constituting the production device NN5 have continuously changed for a certain number of times, but the production start date NN3 and the production device NN5.
It is also possible to end the state update after repeating the state update for a certain time respectively. For example, production start date NN
3 and the production apparatus NN5 repeat the state update for 30 minutes, and then the state update can be ended.

Embodiment 4 Further, the convergence determination means 7 can also end the state update after the production start date NN3 and the production apparatus NN5 have repeatedly performed the state update a certain number of times. For example, the production start date NN3 and the production device N
After N5 repeats the status update 10,000 times, the status update can be ended.

[0056]

According to the first aspect of the invention, the production start date NN output evaluation means obtains the constraint satisfaction degree from the production start date output by the production start date NN and gives the constraint satisfaction degree to the production device NN.
The N output evaluation means obtains the production device from the production possibility output by the production device NN and gives it to the production start date NN, and the production start date NN and the production device NN update the state.
It is possible to change the production start date and the production apparatus of each lot according to the allocation status of other lots to make an optimal production plan.

According to the second aspect of the present invention, the production start date NN, the production start date NN output evaluation means, the production device NN, and the convergence determination means for determining whether or not to repeatedly execute the production device NN output evaluation means. Since the production start date NN and the production device NN can be alternately and repeatedly updated by providing, it is optimal while changing the production start date and production device of each lot according to the allocation status of other lots. It is possible to make various production plans.

According to the third aspect of the invention, the convergence determination means makes the determination based on whether or not the output of the production start date NN or the output of the neurons constituting the production apparatus NN has continuously changed a predetermined number of times. , Planning can be automatically stopped without updating the status more than necessary.

According to the fourth aspect of the present invention, the convergence determination means makes the determination based on whether or not a predetermined time has elapsed.
It is possible to prevent N from repeating the state update indefinitely without convergence.

According to the fifth aspect of the invention, the convergence determination means makes the determination based on whether or not the determination has been executed a predetermined number of times, so that it is possible to prevent the state update from being infinitely repeated without the NN converging. it can.

According to the sixth aspect of the invention, since the planning result display means for displaying the production start date for producing each lot and the production apparatus is provided, the operator can see the obtained planning result on the display screen. it can.

According to the seventh aspect of the present invention, the production start date displayed on the plan result display means or the plan result correction means for correcting the production device, and the production start date based on the correction result by the plan result correction means. Since the NN or the NN output changing means for changing the states of the plurality of neurons constituting the production apparatus NN is provided, the operator can freely change the obtained planning result as he desires, and based on this change designation. Production plan can be re-established.

According to the eighth aspect of the present invention, there is provided a suspension device for suspending the production and a suspension designating means for designating information on the suspension period, and the designation information by this suspension designating means is used for the production start date N.
Since N or the production apparatus NN refers to it, it is possible to specify or change the suspension device and suspension period before or after the planning.

According to the ninth aspect of the invention, the parameter changing means for changing the constraint condition parameter of the production plan is provided, and the change information by the parameter changing means is referred to by the production start date NN or the production device NN output evaluation means. The constraint condition parameters can be specified or changed before or after the planning of the plan.

According to the tenth aspect of the invention, the production start date NN is always set by providing the neuron, which has a one-to-one correspondence with each dummy lot and outputs the production start date of each dummy lot, on the production start date NN. The production start date that satisfies the most restrictive condition when each lot is produced by each of the plural producible devices can be obtained for each producible device. Production start date can be determined. Further, on the production start date NN, dummy lots on different production devices do not affect each other's determination of the production start date, so only neurons corresponding to the dummy lots on the same production device are By combining them, the production start date of each dummy lot can be determined independently for each production device.

According to the eleventh aspect, the true lot is determined by the output NN output evaluation means of the production apparatus, and when the two connected neurons are true lots according to the production start date NN, the production period of the lot is The production start date is calculated so as not to overlap, and when one of the two connected neurons is not a true lot, the production start date is calculated while allowing the production periods of the lots to overlap. Even on a producible device that does not actually perform production, it is possible to obtain the production start date that satisfies the most restrictive conditions. For this reason, at some point during the planning process, a dummy lot that is not a true lot may have a higher degree of constraint satisfaction than a true lot.However, by enabling such a situation The production equipment can be changed.

According to the twelfth aspect of the invention, by providing the production apparatus NN with a neuron that corresponds to each dummy lot in a one-to-one correspondence and outputs the productivity of each dummy lot, the production apparatus NN always operates in each lot. Can know which production device can produce the most satisfying constraint condition, and can determine the production device for each lot. Further, in the production apparatus NN, dummy lots of different lots do not affect each other's determination of producibility of each other. Therefore, by connecting only neurons corresponding to dummy lots having the same lot number. , The production device for each dummy lot can be determined independently for each lot.

According to the thirteenth invention, the production apparatus NN is
The constraint satisfaction degree of each dummy lot obtained by the production start date NN output evaluation means is set for each neuron, and the neuron having the maximum constraint satisfaction degree among the plurality of connected neurons is obtained. Since the productivity of the neurons is increased, it is possible to automatically change the production apparatus of each lot in the middle stage of the planning to the production apparatus that further satisfies the constraint condition.

According to the fourteenth invention, the production apparatus NN is
The constraint satisfaction degree of each dummy lot obtained by the production start date NN output evaluation means is set for each neuron, and the neuron having the maximum constraint satisfaction degree among the plurality of connected neurons is obtained and obtained. Since the production possibilities other than the neurons are reduced, the production apparatus of each lot in the middle stage of the planning can be automatically changed to the production apparatus that further satisfies the constraint condition.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a configuration of a production planning system according to a first embodiment.

FIG. 2 is an explanatory diagram illustrating a list of producible devices for each lot.

FIG. 3 is an explanatory diagram illustrating a dummy lot.

FIG. 4 is an explanatory diagram illustrating an overlapping relationship between a true lot and a dummy lot that is not.

FIG. 5 is a configuration diagram showing a configuration of a production start date NN according to the first embodiment.

FIG. 6 is a configuration diagram showing a configuration of the production apparatus NN of the first embodiment.

FIG. 7 is a flowchart showing a flow of production planning processing according to the first embodiment.

FIG. 8 is a configuration diagram showing a configuration of a production planning system according to a second embodiment.

FIG. 9 is a flowchart showing a flow of processing for designating a suspension device and constraint condition parameters according to the second embodiment.

FIG. 10 is a flow chart showing a flow of processing of correction and re-execution of the production planning result of the second embodiment.

FIG. 11 is a flowchart showing a flow of production planning processing in the prior art.

[Explanation of symbols]

1 data input means, 2 dummy lot generation means, 3
Production start date NN, 4 Production start date NN output evaluation means, 5
Production device NN, 6 Production device NN output evaluation means, 7
Convergence determination means, 8 planning result display means, 9 planning result correction means, 10 NN output changing means, 11 suspension device specifying means, 12 parameter changing means

Claims (14)

[Claims] 1. A production planning system for determining a plurality of production devices for producing each lot for a plurality of lots and a production start date of each lot, by inputting order data and production capacity data of each production device. , Based on the input order data and production capacity data, the same number of dummy lots as the number of producible devices are generated for each lot, and for each dummy lot, a lot to be produced and the production of this lot Feasible device and the initial value of the production start date and the dummy lot generating means, and the production start date neural network (hereinafter, the neural network is referred to as NN) for updating the production start date of each dummy lot set by the dummy lot generating means. )
A production start date NN output evaluation means for obtaining a constraint satisfaction degree of each dummy lot from the production start date of each dummy lot updated by this production start date NN; and each dummy set by the dummy lot generation means. A production apparatus NN that obtains the possibility that each dummy lot will be produced by the lot producible apparatus, and obtains the production apparatus of each lot based on the producibility of each dummy lot obtained by this production apparatus NN A production device NN output evaluation means, wherein the production start date NN updates the production start date of each dummy lot based on the production device of each lot obtained by the production device NN output evaluation means, The production device NN is configured so that each dummy lot is based on the constraint satisfaction degree of each dummy lot obtained by the production start date NN output evaluation means. Production planning system and obtaining a lot of production possibility. 2. The production start date NN, the production start date N
The N output evaluation means, the production apparatus NN, and a convergence determination means for determining whether or not to repeatedly execute the production apparatus NN output evaluation means are provided.
Production planning system described. 3. The production start date N is defined by the convergence determining means.
The production planning system according to claim 2, wherein the determination is performed based on whether or not the output of N or the neuron that constitutes the production apparatus NN has continuously changed a predetermined number of times. 4. The production planning system according to claim 2, wherein the convergence determination means makes the determination based on whether or not a predetermined time has elapsed. 5. The production planning system according to claim 2, wherein the convergence determination means makes the determination based on whether or not the determination has been performed a predetermined number of times. 6. Each of the lots is produced based on the production start date of each of the dummy lots updated by the production start date NN and the production device of each lot obtained by the production device NN output evaluation means. 2. The production planning system according to claim 1, further comprising a planning result display means for displaying a production start date and a production device to be produced. 7. The production start date displayed on the planning result display means or the planning result correction means for correcting the production apparatus, and the production start date NN or the production apparatus based on the correction result by the planning result correction means. 7. The production planning system according to claim 6, further comprising NN output changing means for changing the states of a plurality of neurons constituting the NN. 8. A suspension device for suspending production and a suspension designating means for designating information regarding suspension period, wherein the production start date NN is determined by the information designated by the suspension designating means and the production device NN output evaluation means. The production start date of each dummy lot is updated based on the obtained production device of each lot, and the production device NN obtains the information designated by the suspension designating means and the production start date NN output evaluation means. 2. The production planning system according to claim 1, wherein the production possibility of each dummy lot is obtained based on the constraint satisfaction degree of each dummy lot. 9. A parameter changing unit for changing a constraint condition parameter of a production plan, wherein the production start date NN is the constraint condition parameter changed by the parameter changing unit and the production apparatus NN.
The production start date of each dummy lot is updated based on the production apparatus of each lot obtained by the output evaluation unit, and the production apparatus NN output evaluation unit sets the constraint condition parameter changed by the parameter changing unit. 2. The production planning system according to claim 1, wherein a production device for each lot is obtained based on the producibility of each dummy lot obtained by the raw device NN. 10. The production start date NN sets the production lot and the producible device set in each of the dummy lots to a plurality of neurons constituting the production start date NN, and different lots are set. And connecting the neurons to which the same producible device is set, and causing each neuron to obtain a production start date when the set production lot is produced by the set producible device. The production planning system according to claim 1, which is characterized in that. 11. The production device NN output evaluation means determines, as the true lot to be actually produced, the dummy lot having the maximum productivity in each lot obtained by the production device NN, and starts the production. The day NN produces each neuron based on whether it is the determined true lot or not, when the two connected neurons are true lots, the production periods of the lots do not overlap. 11. The production plan according to claim 10, wherein the start date is obtained, and when one of the two connected neurons is not a true lot, the production start date is obtained while allowing overlapping of production periods of the lots. system. 12. The production apparatus NN sets the production lot set in each of the dummy lots and the producible apparatus in a plurality of neurons constituting the production apparatus NN, and the same lot is set. Characterized in that the neurons in which different producible devices are set are connected to each other, and each neuron is requested for producibility when the set production lot is produced by the set producible device. The production planning system according to claim 1. 13. The production device NN sets, in each of the neurons, the constraint satisfaction degree of each of the dummy lots obtained by the production start date NN output evaluation means, among the plurality of connected neurons. 13. The production planning system according to claim 12, wherein a neuron that maximizes the set constraint satisfaction degree is obtained in step (3), and the productivity of the obtained neuron is increased. 14. The production device NN sets, in each of the neurons, the constraint satisfaction degree of each of the dummy lots obtained by the production start date NN output evaluation means, and among the plurality of connected neurons. 13. The production planning system according to claim 12, wherein a neuron that maximizes the set constraint satisfaction degree is obtained in, and the productivity of the other neurons than the obtained neuron is reduced.