Multi-stage mixed flow workshop scheduling method for forge piece forming and manufacturing under limited buffer constraint

Abstract

The invention discloses a multi-stage mixed flow workshop scheduling method for forge piece forming and manufacturing under limited buffer constraint, and belongs to the technical field of forge piece hot working process management and control in the discrete manufacturing industry. The method comprises the following steps: constructing a heterogeneous disjunction graph for a hot processing hybrid flow shop based on limited buffer constraints among a plurality of operation procedures in the hot processing hybrid flow shop; performing feature extraction on the heterogeneous disjunction graph to obtain workshop state information corresponding to a plurality of operation processes; the workshop state information is input to a scheduling optimization model, decision action information output by the scheduling optimization model is obtained, the decision action information is used for scheduling the hot-working hybrid flow workshop, and the scheduling optimization model is constructed based on a full-connection neural network. The method can improve the production efficiency of the hot working mixing flow workshop.

Description

Technical Field

[0001] The present application belongs to the technical field of hot working process control of forgings in the discrete manufacturing industry, and in particular to a multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints. Background Art

[0002] The multi-variety variable batch forging forming manufacturing process is a typical discrete manufacturing production mode, which is characterized by the large variability of the variety, batch and quantity of forgings. The multi-stage mixed flow workshop for forging forming manufacturing is the main production carrier of the multi-variety variable batch forging forming manufacturing process. This type of mixed flow workshop has many process nodes, lengthy process flow, complex process constraints, a wide variety of equipment, and frequent random disturbances, which makes dynamic scheduling of the workshop extremely difficult and energy efficiency optimization also faces huge challenges.

[0003] Intelligent optimization algorithm is the core tool for scheduling decision-making in the workshop control system of forging forming process, and is the key means to achieve high-efficiency and intelligent scheduling. Meta-heuristic algorithms represented by genetic algorithms, simulated annealing algorithms and swarm intelligence optimization algorithms have been widely used to solve combinatorial optimization problems in workshop scheduling. However, technical solutions based on such algorithms usually do not take into account the batch, process, buffer and other characteristics of multi-stage mixed-flow workshops for forging forming manufacturing, and have poor practicality. The algorithm structure is also difficult to meet the real-time scheduling requirements.

[0004] Deep reinforcement learning, as an advanced machine learning method that combines deep learning and reinforcement learning, has been successfully applied to the dynamic scheduling problem of cold processing workshops (such as turning, milling, planing, grinding, etc.) in recent years, showing high solution accuracy and efficiency. However, unlike cold processing workshops, forging forming and manufacturing workshops belong to hot processing workshops, and there are significant differences in their production line composition and operation mode. There are also multiple stage buffers, which greatly affect production efficiency. Summary of the invention

[0005] The present application aims to solve at least one of the technical problems existing in the prior art. To this end, the present application proposes a scheduling method for a multi-stage mixed flow workshop for forging forming and manufacturing under limited buffer constraints, which can improve the production efficiency of a hot processing mixed flow workshop.

[0006] In a first aspect, the present application provides a method for scheduling a multi-stage mixed-flow workshop for forging forming manufacturing under a limited buffer constraint, the method comprising: Based on the limited buffer constraints between multiple working procedures in the hot processing hybrid flow shop, a heterogeneous disjunctive graph is constructed for the hot processing hybrid flow shop; Performing feature extraction on the heterogeneous disjunctive graph to obtain workshop status information corresponding to the multiple operation processes; The workshop status information is input into the scheduling optimization model to obtain decision action information output by the scheduling optimization model, and the decision action information is used to schedule the hot processing mixed flow workshop. The scheduling optimization model is constructed based on a fully connected neural network, and the scheduling optimization model is obtained based on multiple training samples, and the training samples include sample workshop status and decision action labels corresponding to the sample workshop status.

[0007] According to one embodiment of the present application, the heterogeneous disjunctive graph includes a node set, a finite buffer, and an edge set; The node set is used to represent the multiple job processes, the edge set is used to represent the sequence relationship and equipment constraints between the multiple job processes, and the finite buffer is used to represent the finite buffer constraint.

[0008] According to an embodiment of the present application, the operation process includes a first operation process of forging and a second operation process of batch heat treatment, and the limited buffer constraint is used to define the triggering condition from the first operation process to the second operation process.

[0009] According to one embodiment of the present application, the workshop status information is represented by a status feature vector :

[0010] in, is the machine utilization rate, is the average utilization of the machine, is the standard deviation of the average utilization of the machine, is the average completion rate of the operation process, is the standard deviation of the average completion rate of the operation process, is the average energy consumption of the machine, is the standard deviation of the average energy consumption of the machine.

[0011] According to an embodiment of the present application, after inputting the workshop status information into the scheduling optimization model and obtaining the decision action information output by the scheduling optimization model, the method further includes: Determine the state transition information of the hot processing hybrid flow shop after scheduling based on the decision action information; Calculating a reward value of the decision action information according to the state transition information; storing a combination of the workshop state information, the state transition information, the decision action information and the reward value as an interaction sample in a memory pool; The interaction samples are extracted from the memory pool to update the parameters of the scheduling optimization model.

[0012] According to an embodiment of the present application, the reward value is calculated based on a reward function, the state transition information includes the energy consumption cost and total production time of the hot processing mixed flow shop, and the reward function is:

[0013] in, is the reward value of the decision action information at time step t; is the energy cost at time step t; is the total production time at time step t; They are , The weight of .

[0014] According to an embodiment of the present application, the scheduling optimization model may include an action network and a target network, and extracting the interaction sample from the memory pool to update the parameters of the scheduling optimization model includes: Inputting the interaction sample into the action network and the target network respectively, obtaining a first output of the action network and a second output of the target network; Calculate the loss value of the first output and the second output by using a loss function; Based on the loss value, the parameters of the target network are optimized by Adam optimizer.

[0015] In a second aspect, the present application provides a multi-stage mixed flow workshop scheduling device for forging forming manufacturing under limited buffer constraints, the device comprising: An acquisition module, configured to construct a heterogeneous disjunctive graph for a hot processing hybrid flow shop based on a finite buffer constraint between multiple operation procedures in the hot processing hybrid flow shop; A first processing module is used to extract features from the heterogeneous disjunctive graph to obtain workshop status information corresponding to the multiple operation processes; The second processing module is used to input the workshop status information into the scheduling optimization model to obtain the decision action information output by the scheduling optimization model, and the decision action information is used to schedule the hot processing mixed flow workshop. The scheduling optimization model is constructed based on a fully connected neural network, and the scheduling optimization model is obtained based on multiple training samples. The training samples include sample workshop status and decision action labels corresponding to the sample workshop status.

[0016] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for scheduling a multi-stage mixed-flow workshop for forging forming manufacturing under limited buffer constraints as described in the first aspect above is implemented.

[0017] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints as described in the first aspect above.

[0018] In a fifth aspect, the present application provides a chip, comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints as described in the first aspect.

[0019] In a sixth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints as described in the first aspect above.

[0020] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application.

[0021] The present application provides a multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints, which has the following beneficial effects compared with the prior art: (1) The heterogeneous disjunctive graph of the hot working hybrid flow shop is constructed by finite buffer constraints, which solves the problems of insufficient optimization of traditional single production links and uncoordinated scheduling decisions, low efficiency and high energy consumption in multi-stage production workshops. It can dynamically coordinate the hot working hybrid flow shop operation processes of the forging workshop and the heat treatment workshop in real time, improve the production efficiency of the hot working hybrid flow shop, reduce production energy consumption, and achieve dynamic scheduling optimization with high efficiency and energy saving.

[0022] (2) The heterogeneous disjunctive graph constructed through finite buffer constraints can realize the collaborative operation between the work processes in the hot working mixed flow workshop during the batch manufacturing process of forging forming, providing an efficient modeling method for the construction of the scheduling environment of the hot working mixed flow forming manufacturing workshop, thereby improving the efficiency and accuracy of workshop decision-making.

[0023] (3) Dynamic scheduling of hot processing mixed flow workshops is achieved through the dual network structure and the adaptive learning of the Adam optimizer. It can independently select the optimal equipment and process path according to the type, batch, quantity of the workpiece and the available status of the equipment under real-time working conditions. Compared with the traditional static scheduling method, this application can effectively respond to equipment failures and process changes, realize dynamic scheduling and real-time adjustment, and improve production efficiency. Based on the processing batch, the heating furnace and aging furnace opening quantity discrimination conditions are adopted to ensure processing efficiency and reduce the energy consumption generated by the heating furnace and aging furnace. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which: Figure 1 It is a flow chart of a multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints provided in an embodiment of the present application; Figure 2 It is a structural schematic diagram of a heterogeneous disjunction graph of a hot processing hybrid flow shop provided in an embodiment of the present application in a state where machine selection is not solved; Figure 3 It is a structural schematic diagram of a heterogeneous disjunction graph of a hot processing hybrid flow shop after solving the machine selection state provided by an embodiment of the present application; Figure 4 It is a flow chart of a two-stage mixed flow workshop environment of forging forming and heat treatment under limited buffer constraints and a decision-making agent training process and a decision-making process provided by an embodiment of the present application; Figure 5 It is a Gantt chart of the sorting and scheduling decision under the condition of parallel machine failure considering the transit time provided by the embodiment of the present application; Figure 6 The embodiment of the present application provides a Gantt chart for scheduling decisions considering three forgings of the same quantity in three same batches; Figure 7 The embodiment of the present application provides a Gantt chart for scheduling decisions considering three different quantities of forgings in three different batches; Figure 8 It is a reward function iteration diagram provided in an embodiment of the present application; Fig. 9 It is a schematic diagram comparing the total energy consumption in the solution process of the DDQN algorithm adopted in the present application and the traditional genetic algorithm and scheduling rules provided in the embodiment of the present application; Fig.10 It is a structural schematic diagram of a multi-stage mixed-flow workshop scheduling device for forging forming manufacturing under limited buffer constraints provided in an embodiment of the present application; Fig.11 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application belong to the scope of protection of this application.

[0026] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0027] In combination with the accompanying drawings, the multi-stage mixed flow workshop scheduling method for forging forming manufacturing under limited buffer constraints, the multi-stage mixed flow workshop scheduling device for forging forming manufacturing under limited buffer constraints, the electronic device and the readable storage medium provided in the embodiments of the present application are described in detail through specific embodiments and their application scenarios.

[0028] Among them, the multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints can be applied to the terminal, and can be specifically executed by hardware or software in the terminal.

[0029] The terminal includes, but is not limited to, a portable communication device such as a mobile phone or a tablet computer having a touch-sensitive surface (e.g., a touch screen display and / or a touch pad). It should also be understood that in some embodiments, the terminal may not be a portable communication device, but a desktop computer having a touch-sensitive surface (e.g., a touch screen display and / or a touch pad).

[0030] In the following various embodiments, a terminal including a display and a touch-sensitive surface is described. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, a mouse and a joystick.

[0031] The embodiment of the present application provides a multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints. The executor of the multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints can be an electronic device or a functional module or functional entity in the electronic device that can implement the multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints. The electronic devices mentioned in the embodiment of the present application include but are not limited to mobile phones, tablet computers, computers, cameras, wearable devices, etc. The multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints provided in the embodiment of the present application is explained below using electronic devices as the executor as an example.

[0032] like Figure 1 As shown, the multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints includes: steps 110 to 130.

[0033] Step 110: construct a heterogeneous disjunctive graph for the hot working mixed flow shop based on the finite buffer constraints between multiple operation procedures in the hot working mixed flow shop.

[0034] Among them, the hot working mixed assembly line workshop can be a two-level mixed assembly line workshop including a forging forming workshop and a heat treatment workshop, which is used to execute the production process of forgings, including seven steps of heating, blanking, die forging, trimming, solution hardening, quenching and aging. The hot working mixed assembly line workshop produces corresponding forgings according to orders.

[0035] The heterogeneous disjunctive graph may be a heterogeneous disjunctive graph scheduling environment model.

[0036] In this step, the operating procedures in the hot processing hybrid flow shop and the relationship between the operating procedures are determined, the operating procedures are abstracted into corresponding nodes, the finite buffer zone is determined according to the finite buffer constraint, and the edges between the nodes are determined according to the relationship between the operating procedures to obtain the heterogeneous disjunction graph of the hot processing hybrid flow shop.

[0037] This application proposes for the first time to construct the forging forming manufacturing process into a multi-stage mixed flow workshop production mode under the constraints of multi-variety variable batch and limited buffer, and the limited buffer is used to store the work-in-progress of forgings.

[0038] Step 120: extract features from the heterogeneous disjunctive graph to obtain workshop status information corresponding to the multiple operation processes.

[0039] In this step, the node status of each node in the production process is collected in real time through sensors and monitoring systems, and characteristic indicators are determined according to the business needs of forgings. Based on the characteristic indicators, the node status of each node is extracted to obtain the workshop status information of the hot processing mixed flow workshop.

[0040] The node status can be characterized by the execution status of the current process and the filling amount of each finite buffer.

[0041] Step 130: input the workshop status information into the scheduling optimization model to obtain the decision action information output by the scheduling optimization model, wherein the decision action information is used to schedule the hot processing mixed flow workshop. The scheduling optimization model is constructed based on a fully connected neural network, and the scheduling optimization model is obtained based on multiple training samples, wherein the training samples include sample workshop status and decision action labels corresponding to the sample workshop status.

[0042] Among them, the workshop status information can be obtained after normalization and encoding processing to meet the input requirements of the fully connected neural network.

[0043] It can be understood that the scheduling optimization model is a trained fully connected neural network (FCNN).

[0044] Each training sample consists of a sample workshop state label and a corresponding decision action label. The sample workshop state can describe the specific situation of the current production or operation system in the workshop, and the decision action label is the best action that should be taken based on the state.

[0045] Through training with multiple training samples, the scheduling optimization model can learn how to make corresponding decisions based on different workshop conditions and continuously adjust its internal parameters so that it can make more appropriate decisions when faced with new workshop conditions.

[0046] In this step, the workshop status information is input into the scheduling optimization model, which extracts and calculates features of the workshop status information, and determines information such as the process scheduling sequence and the limited buffer zone management strategy corresponding to the workshop status information by selecting scheduling rules. This information is integrated into decision action information and output, and multiple work processes in the hot processing mixed flow workshop are controlled to execute operations corresponding to the decision action information, so as to realize the scheduling of the hot processing mixed flow workshop.

[0047] According to the scheduling method of a multi-stage mixed flow workshop for forging forming manufacturing under finite buffer constraints provided in the embodiment of the present application, a heterogeneous disjunction graph of a hot working mixed flow workshop is constructed through finite buffer constraints, which solves the problems of insufficient optimization of traditional single production links and uncoordinated scheduling decisions, low efficiency and high energy consumption in multi-stage production workshops. The method can dynamically coordinate the hot working mixed flow workshop operation procedures of the forging forming workshop and the heat treatment workshop in the hot working mixed flow workshop in real time, improve the production efficiency of the hot working mixed flow workshop, reduce production energy consumption, and achieve efficient and energy-saving dynamic scheduling optimization.

[0048] In some embodiments, the heterogeneous disjunctive graph includes a set of nodes, a finite buffer, and a set of edges; The node set is used to represent the multiple job processes, the edge set is used to represent the sequence relationship and equipment constraints between the multiple job processes, and the finite buffer is used to represent the finite buffer constraint.

[0049] Among them, equipment constraints are used to characterize the processable workload limited by the performance and capacity of production equipment in a hot processing hybrid flow shop.

[0050] The node set V includes the nodes corresponding to each operation process, and the edge set includes the edges between each node. Each edge can have multiple attributes such as weight, direction and validity, which are used to express the complex relationship between the nodes corresponding to each operation process.

[0051] For example, there are three types of forgings in a large-volume order of multiple varieties, and the quantities are divided into the same quantity and different quantities of the three orders. In a hot processing mixed flow shop, since the time of the heat treatment operation process is too long compared to the time of the forging operation process, in order to facilitate visualization, the time of the heat treatment operation process is uniformly shortened to the same order of magnitude as the time of the forging operation process in the heterogeneous disjunction diagram.

[0052] like Figure 2 and Figure 3 As shown in the figure, the scheduling optimization model obtains the decision action information by solving the machine selection state. In the heterogeneous disjunctive graph of three types of workpieces, node S is the beginning of the scheduling task, node E is the end of the scheduling task, node M is the production equipment in the hot processing mixed flow workshop, and node O is the operation process of the workpiece.

[0053] In actual implementation, the number of heating furnaces and aging furnaces to be opened is determined according to the number of forging orders. After determining the workshop composition of the hot processing mixed flow workshop, it is represented as a heterogeneous disjunctive graph using the disjunctive graph model. , where the node set V represents n job processes, and the edge set E represents the sequential relationship and equipment constraints between jobs.

[0054] In this embodiment, by constructing a heterogeneous disjunctive graph including a finite buffer zone, the process requirements of forgings can be met and the efficiency and accuracy of workshop decision-making can be improved.

[0055] In some embodiments, the operation process includes a first operation process of forging and a second operation process of batch heat treatment, and the finite buffer constraint is used to define a trigger condition from the first operation process to the second operation process.

[0056] In actual implementation, the constructed finite buffer constraint area is between the node corresponding to the forging process and the node corresponding to the batch heat treatment process, indicating that when the forgings formed by forging accumulate to a given number, the next stage of batch heat treatment is triggered.

[0057] In this embodiment, the heterogeneous disjunctive graph constructed through finite buffer constraints can realize the collaborative operation between the work processes in the hot working mixed flow workshop during the batch manufacturing process of forging forming, providing an efficient modeling method for the construction of the scheduling environment of the hot working mixed flow forming manufacturing workshop, thereby improving the efficiency and accuracy of the workshop decision-making.

[0058] In some embodiments, the workshop status information is represented by a status feature vector :

[0059] in, is the machine utilization rate, is the average utilization of the machine, is the standard deviation of the average utilization of the machine, is the average completion rate of the operation process, is the standard deviation of the average completion rate of the operation process, is the average energy consumption of the machine, is the standard deviation of the average energy consumption of the machine.

[0060] In actual execution, according to the heterogeneous disjunction graph at each time t, the state feature vector is extracted and preprocessed, and the state feature vector is extracted from the heterogeneous disjunction graph. , including machine utilization, average machine utilization, standard deviation of average machine utilization, average completion rate of operating procedures, standard deviation of average completion rate of operating procedures, average energy consumption of machines, standard deviation of average energy consumption of machines, and then the state feature vectors are standardized and normalized, all of which are values ​​between [0, 1] to adapt to the input requirements of the neural network of the intelligent agent and accelerate the convergence of the neural network.

[0061] In this embodiment, by using the state feature vector containing multiple dimensional states of the hot processing hybrid assembly line as the input of the scheduling optimization model, the scheduling optimization model can be helped to learn the interaction between features, capture deeper relationships, and improve the prediction accuracy of the scheduling optimization model.

[0062] In some embodiments, after inputting the workshop status information into the scheduling optimization model to obtain the decision action information output by the scheduling optimization model, the method further includes: Determine the state transition information of the hot processing hybrid flow shop after scheduling based on the decision action information; Calculating a reward value of the decision action information according to the state transition information; storing a combination of the workshop state information, the state transition information, the decision action information and the reward value as an interaction sample in a memory pool; The interaction samples are extracted from the memory pool to update the parameters of the scheduling optimization model.

[0063] Among them, the state transfer information can be the transformation and related information between different states of the hot processing mixed flow workshop after scheduling based on the decision action information, which can include various links of production, the operating status of equipment, the flow of materials, etc.

[0064] In some embodiments, the reward value is calculated based on a reward function, the state transition information includes the energy consumption cost and total production time of the hot processing hybrid flow shop, and the reward function is:

[0065] in, is the reward value of the decision action information at time step t; is the energy cost at time step t; is the total production time at time step t; They are , The weight of .

[0066] It should be noted that The specific calculation method is: in, e i For the i Energy consumption of a production line (the average power of each device in the production line multiplied by the cumulative working time), p is the unit energy cost.

[0067] It can be understood that based on the decision actions made by the hot processing hybrid assembly line, a reward function that considers production efficiency and production energy consumption is set under the premise of ensuring process constraints. The process constraints include at least one of equipment constraints, process flow constraints, material constraints, time constraints and quality standards.

[0068] The specific calculation method is:

[0069] in, is the production completion time of the i-th production line. The steps in the production process include blanking, heating, billet making, die forging, trimming, solution treatment, quenching and aging.

[0070] In this embodiment, by setting a refined reward function based on energy consumption cost and total production time, the learning of the scheduling optimization model can be guided to achieve more accurate decision-making.

[0071] In some embodiments, the scheduling optimization model may include an action network and a target network, and extracting the interaction sample from the memory pool to update the parameters of the scheduling optimization model includes: Inputting the interaction sample into the action network and the target network respectively, obtaining a first output of the action network and a second output of the target network; Calculate the loss value of the first output and the second output by using a loss function; Based on the loss value, the parameters of the target network are optimized by Adam optimizer.

[0072] Among them, the Adam optimizer can effectively minimize the loss function and improve the performance of the model.

[0073] It can be understood that the solution process of the two-stage mixed flow shop scheduling problem of the hot processing mixed flow shop is modeled as a Markov decision process, including the construction of the scheduling environment, state, action, reward function and decision-making agent.

[0074] like Figure 4 As shown in the figure, a scheduling optimization model based on a double deep Q-network (DDQN) is designed and trained. It is equipped with a target network, an action network, a loss function, a reward function, a memory pool, and a rule base. The hot processing hybrid flow shop includes a forging stage and a heat treatment stage. The state of the heterogeneous disjunction graph (limited buffer multi-stage heterogeneous graph) at each moment is extracted, and a decision is made through the action network. The state s, action a, reward r at this moment, and the state s at the next moment are combined. t The target network and action network are iteratively trained based on the data in the memory pool. The action network outputs C and Q values, and the target network outputs the target Q value (Target Q) according to C. The loss function updates the neural network weights through the Adam method according to the Q value and the target Q value until convergence, meeting the scheduling requirements of the multi-stage mixed-flow workshop for forging forming manufacturing under the limited buffer constraint. k refers to the state at time k, S tk It refers to the state at the next moment after time k.

[0075] In actual execution, the status information of the hot processing mixed flow shop is obtained in real time and the real-time status update formula of the heterogeneous disjunctive graph model is updated as follows:

[0076] in, is the current state feature vector, is the heterogeneous disjunction graph at the current moment, It is the state feature extraction function.

[0077] The number of heating furnaces and aging furnaces to be opened needs to take into account the constraints of minimizing total energy consumption and batch size, and two heating furnaces and three aging furnaces are opened.

[0078] The state characteristics are standardized and normalized. The standardization formula is:

[0079] in, is the original eigenvalue, Features The mean of Features The standard deviation of . The normalization formula is:

[0080] in, and are the minimum and maximum eigenvalues, respectively.

[0081] The decision action information is generated by the DDQN algorithm by selecting one of the following nine scheduling rules: (1) Select the workpiece with the shortest processing time; (2) Select the workpiece with the longest processing time in the operation process; (3) Select the workpiece with the smallest ratio between the processing time of the current operation and the processing time of the subsequent operation; (4) Select the workpiece with the shortest subsequent processing time; (5) Select the workpiece with the lowest remaining processing energy consumption; (6) Select the workpiece with the lowest energy consumption in the operation process; (7) Select the workpiece with the highest energy consumption in the operation process; (8) Select the workpiece with the smallest ratio of the processing energy consumption of the operation process to the processing energy consumption of the subsequent operation process; (9) Select the workpiece with the lowest energy consumption for subsequent processing.

[0082] When the sorting action of the forging forming stage is completed, the constraint on whether the number of workpieces in the finite buffer zone meets the conditions for entering the next stage of processing is:

[0083]

[0084] in, It is the minimum number of workpieces required to enter the next stage of processing after the forging process is completed. Q Indicates the current number of artifacts in the finite buffer. It's time t+ The number of workpieces in the finite buffer at 1, It's time t The number of workpieces in the finite buffer at time M, the number of parallel machines currently processing at time M, It is j The processing speed of the parallel machines (the number of workpieces processed per unit time), It is j The state variable of the parallel machine, 1 means it is processing and 0 means it is idle.

[0085] Considering how to achieve rescheduling when a parallel machine fails, the rescheduling time adjustment formula is:

[0086] in, is the total completion time after rescheduling, is the planned completion time before the failure occurs, It is the time lost due to failure.

[0087] The optimization goal of minimizing the impact of failure is:

[0088] in, F Indicates the number of workpieces that stopped processing due to a fault. is the processing speed of the reallocated spare machine. The optimization goal is to minimize the time affected by the failure, that is, to quickly resume production.

[0089] Reselection Constraints for Parallel Machines

[0090] in, is the standby parallel machine selection variable, 1 means selecting the k A spare machine, 0 means not selected.

[0091] like Figure 5 As shown, the horizontal axis is time in seconds, and the vertical axis is the machine. It considers the sorting scheduling of three types of workpieces when the transit time is 0 or 3 seconds and the M31 parallel machine fails, with only one workpiece of each type.

[0092] like Figure 6 As shown, the horizontal axis is time in seconds, and the vertical axis is the machine. It is mainly for the sorting and scheduling of three batches of 5, 10 and 20 under the same batch size of multiple varieties.

[0093] like Figure 7 As shown, the horizontal axis is time in seconds, and the vertical axis is the machine. It mainly shows the sorting and scheduling of three batches of 5, 10 and 20 under the same batch size of multiple varieties.

[0094] In actual implementation, the scheduling optimization model of the DDQN architecture includes an action network and a target network. The action network and the target network are composed of fully connected neural networks. The standardized workshop status information is processed to obtain the state feature vector , respectively input into the action network and target network, and output the state feature vector through the action network The Q value of the target network outputs the state feature vector The action network is used for decision output, and the target network is used for stable training.

[0095] The action network is used to estimate the Take decision action information The Q value is expressed as:

[0096] in is the Q-value function, and are the current state feature vector and the next state feature vector respectively, and are the current decision action information and the next decision action information respectively. For the current reward, A discount factor that weighs the importance of current rewards and future rewards, are the network parameters of the action network, are the network parameters of the target network.

[0097] The training steps of the action network and the target network use experience replay and fixed Q target technology. Experience replay alleviates the correlation between samples, and fixed Q target is used to stabilize training. The formula for experience replay is:

[0098] in, is the loss function, A buffer for experience replay.

[0099] The parameter adjustment and scheduling strategy of the scheduling optimization model are expressed as:

[0100] in, are model parameters, is the learning rate, is the gradient of the loss function.

[0101] like Figure 8 As shown in the figure, the horizontal axis is the number of iterations (Iteration), and the vertical axis is the reward value (Reward). During the iterative process of the reward function, it gradually converges after about 150 iterations.

[0102] like Fig. 9 As shown in the figure, the horizontal axis is used to represent DDQN, genetic algorithm (GA) and scheduling rule (SR), and the vertical axis is total energy consumption (Total Energy Consumption). The number of three types of workpieces is 8, 10, and 13 respectively. By comparing with the genetic algorithm and the ordinary scheduling rules, it is found that the sorting scheduling driven by this algorithm can significantly reduce energy consumption.

[0103] In this embodiment, dynamic scheduling of the hot processing mixed flow shop is achieved through the adaptive learning of the dual network structure and the Adam optimizer. It can independently select the optimal equipment and process path according to the type, batch, quantity of the workpiece and the available status of the equipment under real-time working conditions. Compared with the traditional static scheduling method, the present application can effectively respond to equipment failures and process changes, realize dynamic scheduling and real-time adjustment, and improve production efficiency. Based on the processing batch, the opening quantity discrimination conditions of the heating furnace and the aging furnace are adopted to ensure the processing efficiency and reduce the energy consumption generated by the heating furnace and the aging furnace.

[0104] The embodiment of the present application provides a method for scheduling a multi-stage mixed flow workshop for forging forming manufacturing under limited buffer constraints, and the execution body may be a scheduling device for a multi-stage mixed flow workshop for forging forming manufacturing under limited buffer constraints. In the embodiment of the present application, the scheduling method for a multi-stage mixed flow workshop for forging forming manufacturing under limited buffer constraints is performed by a scheduling device for a multi-stage mixed flow workshop for forging forming manufacturing under limited buffer constraints as an example to illustrate the scheduling device for a multi-stage mixed flow workshop for forging forming manufacturing under limited buffer constraints provided in the embodiment of the present application.

[0105] The embodiment of the present application also provides a multi-stage mixed-flow workshop scheduling device for forging forming and manufacturing under limited buffer constraints.

[0106] like Fig.10 As shown, the multi-stage mixed flow workshop scheduling device for forging forming manufacturing under limited buffer constraints includes: An acquisition module 1010 is used to construct a heterogeneous disjunctive graph for a hot processing hybrid flow shop based on a finite buffer constraint between multiple operation procedures in the hot processing hybrid flow shop; The first processing module 1020 is used to extract features from the heterogeneous disjunction graph to obtain the workshop status information corresponding to the multiple operation processes; The second processing module 1030 is used to input the workshop status information into the scheduling optimization model to obtain the decision action information output by the scheduling optimization model, and the decision action information is used to schedule the hot processing mixed flow workshop. The scheduling optimization model is constructed based on a fully connected neural network, and the scheduling optimization model is obtained based on multiple training samples. The training samples include sample workshop status and decision action labels corresponding to the sample workshop status.

[0107] According to the multi-stage mixed flow workshop scheduling device for forging forming manufacturing under finite buffer constraints provided in the embodiment of the present application, a heterogeneous disjunction graph of a hot working mixed flow workshop is constructed through finite buffer constraints, which solves the problems of insufficient optimization of traditional single production links and uncoordinated scheduling decisions, low efficiency and high energy consumption in multi-stage production workshops. It can dynamically coordinate the hot working mixed flow workshop operation procedures of the forging forming workshop and the heat treatment workshop in the hot working mixed flow workshop in real time, improve the production efficiency of the hot working mixed flow workshop, reduce production energy consumption, and achieve efficient and energy-saving dynamic scheduling optimization.

[0108] The multi-stage mixed-flow workshop scheduling device for forging forming manufacturing under limited buffer constraints in the embodiment of the present application can be an electronic device, or a component in an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal, or a device other than a terminal. Exemplarily, the electronic device can be a tablet computer, a laptop computer, a PDA, an Ultra-Mobile Personal Computer (UMPC), a netbook, or a Personal Digital Assistant (PDA), etc., which is not specifically limited in the embodiment of the present application.

[0109] The scheduling device for multi-stage mixed-flow workshop for forging forming manufacturing under limited buffer constraints in the embodiment of the present application can be a device with an operating system. The operating system can be a Linux operating system, a Windows operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0110] The multi-stage mixed flow workshop scheduling device for forging forming and manufacturing under limited buffer constraints provided in the embodiment of the present application can realize the various processes implemented by the multi-stage mixed flow workshop scheduling method for forging forming and manufacturing under limited buffer constraints in the above-mentioned embodiment. To avoid repetition, they will not be repeated here.

[0111] In some embodiments, Fig.11As shown, an embodiment of the present application also provides an electronic device 1100, including a processor 1101, a memory 1102, and a computer program stored in the memory 1102 and executable on the processor 1101. When the program is executed by the processor 1101, each process of the embodiment of the multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0112] It should be noted that the electronic devices in the embodiments of the present application include the mobile electronic devices and non-mobile electronic devices mentioned above.

[0113] An embodiment of the present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, each process of the embodiment of the multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0114] The processor is the processor in the electronic device in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0115] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under the above-mentioned limited buffer constraint.

[0116] The processor is the processor in the electronic device in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk.

[0117] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, which are coupled to the processor, and the processor is used to run programs or instructions to implement the various processes of the embodiment of the multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0118] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0119] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0120] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, disk, CD), including a number of instructions for a terminal (which can be a mobile phone, computer, server, or network device, etc.) to execute the multi-stage mixed flow workshop scheduling method for forging forming manufacturing under limited buffer constraints of each embodiment of the present application.

[0121] In the description of this application, "first feature" or "second feature" may include one or more of the features.

[0122] In the description of the present application, “plurality” means two or more.

[0123] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

[0124] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0125] Although the embodiments of the present application have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the claims and their equivalents.

Claims

1. A multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints, characterized in that: include: Based on the limited buffer constraints between multiple working procedures in the hot processing hybrid flow shop, a heterogeneous disjunctive graph is constructed for the hot processing hybrid flow shop; Performing feature extraction on the heterogeneous disjunctive graph to obtain workshop status information corresponding to the multiple operation processes; The workshop status information is input into the scheduling optimization model to obtain decision action information output by the scheduling optimization model, and the decision action information is used to schedule the hot processing mixed flow workshop. The scheduling optimization model is constructed based on a fully connected neural network, and the scheduling optimization model is obtained based on multiple training samples, and the training samples include sample workshop status and decision action labels corresponding to the sample workshop status.

2. The method for scheduling a multi-stage mixed-flow workshop for forging forming manufacturing under limited buffer constraints according to claim 1 is characterized in that: The heterogeneous disjunctive graph includes a node set, a finite buffer, and an edge set; The node set is used to represent the multiple job processes, the edge set is used to represent the sequence relationship and equipment constraints between the multiple job processes, and the finite buffer is used to represent the finite buffer constraint.

3. The method for scheduling a multi-stage mixed-flow workshop for forging forming and manufacturing under limited buffer constraints according to claim 1 is characterized in that: The operation procedures include a first operation procedure of forging and a second operation procedure of batch heat treatment, and the finite buffer constraint is used to define a triggering condition from the first operation procedure to the second operation procedure.

4. The method for scheduling a multi-stage mixed-flow workshop for forging forming and manufacturing under limited buffer constraints according to claim 1 is characterized in that: The workshop status information is represented by a status feature vector : ； in, is the machine utilization rate, is the average utilization of the machine, is the standard deviation of the average utilization of the machine, is the average completion rate of the operation process, is the standard deviation of the average completion rate of the operation process, is the average energy consumption of the machine, is the standard deviation of the average energy consumption of the machine.

5. The method for scheduling a multi-stage mixed-flow workshop for forging forming and manufacturing under limited buffer constraints according to claim 1 is characterized in that: After inputting the workshop status information into the scheduling optimization model to obtain the decision action information output by the scheduling optimization model, the method further includes: Determine the state transition information of the hot processing hybrid flow shop after scheduling based on the decision action information; Calculating a reward value of the decision action information according to the state transition information; storing a combination of the workshop state information, the state transition information, the decision action information and the reward value as an interaction sample in a memory pool; The interaction samples are extracted from the memory pool to update the parameters of the scheduling optimization model.

6. The method for scheduling a multi-stage mixed-flow workshop for forging forming and manufacturing under limited buffer constraints according to claim 5 is characterized in that: The reward value is calculated based on a reward function, the state transition information includes the energy consumption cost and total production time of the hot processing mixed flow workshop, and the reward function is: ； in, is the reward value of the decision action information at time step t; is the energy cost at time step t; is the total production time at time step t; They are , The weight of .

7. The method for scheduling a multi-stage mixed-flow workshop for forging forming and manufacturing under limited buffer constraints according to claim 5 is characterized in that: The scheduling optimization model may include an action network and a target network, and extracting the interaction sample from the memory pool to update the parameters of the scheduling optimization model includes: Inputting the interaction sample into the action network and the target network respectively, obtaining a first output of the action network and a second output of the target network; Calculate the loss value of the first output and the second output by using a loss function; Based on the loss value, the parameters of the target network are optimized by Adam optimizer.

8. A multi-stage mixed-flow workshop scheduling device for forging forming and manufacturing under limited buffer constraints, characterized in that: include: An acquisition module, configured to construct a heterogeneous disjunctive graph for a hot processing hybrid flow shop based on a finite buffer constraint between multiple operation procedures in the hot processing hybrid flow shop; A first processing module is used to extract features from the heterogeneous disjunctive graph to obtain workshop status information corresponding to the multiple operation processes; The second processing module is used to input the workshop status information into the scheduling optimization model to obtain the decision action information output by the scheduling optimization model, and the decision action information is used to schedule the hot processing mixed flow workshop. The scheduling optimization model is constructed based on a fully connected neural network, and the scheduling optimization model is obtained based on multiple training samples. The training samples include sample workshop status and decision action labels corresponding to the sample workshop status.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the multi-stage mixed-flow workshop scheduling method for forging forming manufacturing under limited buffer constraints as described in any one of claims 1-7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for scheduling a multi-stage mixed-flow workshop for forging forming manufacturing under limited buffer constraints as described in any one of claims 1 to 7 is implemented.

Images