Process optimization scheduling method and system for electro-tricycle assembly line

Abstract

The invention discloses an electro-tricycle assembly line process optimization scheduling method and system, and relates to the related field of scheduling management, and the method comprises the steps: determining a plurality of assembly production nodes to carry out the data sensing of an electro-tricycle, constructing an assembly task dependence graph of the electro-tricycle, and carrying out the scheduling analysis of a plurality of pieces of production process information, the method comprises the following steps: generating an initial scheduling scheme, performing simulation test in combination with an assembly production process data set, generating a simulation scheduling result, performing reinforcement learning on the initial scheduling scheme, optimizing the initial scheduling scheme according to a learning result, and generating a process scheduling scheme to perform adaptive optimization on an electric tricycle assembly production line. The technical problem that the production line is difficult to finely dispatch due to the fact that dynamic changes of the production environment are difficult to adapt is solved, the production environment changes are flexibly adapted by dynamically adjusting the dispatching scheme, and the technical effect of overall production of the electro-tricycle assembly production line is improved.

Description

Technical Field

[0001] The present application relates to fields related to scheduling management, and in particular to a method and system for optimizing scheduling of processes in an electric tricycle assembly production line. Background Art

[0002] Traditional production line scheduling methods mostly rely on manual experience and static rules. These methods are usually unable to be flexibly adjusted according to real-time production conditions, resulting in problems such as resource waste, task delays, and production bottlenecks. The limitations of traditional scheduling methods are particularly obvious when faced with multi-task parallelism and complex task dependencies. With the expansion of production scale and the increase in complexity, traditional scheduling methods are difficult to adapt to real-time changes and optimization needs. Therefore, how to achieve dynamic optimization scheduling of production line processes to improve production efficiency and resource utilization has become an important research topic in production management. Summary of the invention

[0003] The present application provides a method and system for optimizing the scheduling of the process of an electric tricycle assembly production line, thereby solving the technical problem that traditional scheduling methods lack flexibility and are difficult to adapt to dynamic changes in the production environment, resulting in difficulty in fine-grained scheduling of large-scale production lines. The application achieves the technical effect of flexibly adapting to changes in the production environment by dynamically adjusting the scheduling plan, thereby improving the overall production capacity of the electric tricycle assembly production line.

[0004] The present application provides a method for optimizing the scheduling of the process of an electric tricycle assembly production line. The method is applied to a system for optimizing the scheduling of the process of an electric tricycle assembly production line, comprising: extracting an assembly production process data set of an electric tricycle, dividing it based on the assembly production process data set, and determining a plurality of assembly production nodes; performing data sensing on the electric tricycle based on the plurality of assembly production nodes, and constructing an assembly task dependency graph of the electric tricycle; performing scheduling analysis on a plurality of production process information according to the assembly task dependency graph, and generating an initial scheduling plan; performing simulation testing based on the initial scheduling plan in combination with the assembly production process data set, and generating a simulation scheduling result; performing reinforcement learning on the initial scheduling plan according to the simulation scheduling result, optimizing the initial scheduling plan according to the learning result, generating a process scheduling plan, and executing the process scheduling plan to adaptively optimize the electric tricycle assembly production line.

[0005] In a possible implementation, data sensing is performed on the electric tricycle based on the multiple assembly production nodes, an assembly task dependency graph of the electric tricycle is constructed, and the following processing is performed: production analysis is performed on the assembly production line of the electric tricycle based on the assembly production process data set, and an objective function is defined; the multiple assembly production nodes are mapped to the objective function for calculation to obtain multiple process dependency coefficients; weights are assigned to the multiple assembly production nodes based on the multiple process dependency coefficients, and a node priority list is set according to the weight assignment results; the multiple assembly production nodes are connected according to the multiple process dependency coefficients according to the node priority list to construct the assembly task dependency graph.

[0006] In a possible implementation, the multiple assembly production nodes are connected according to the multiple process dependency coefficients according to the node priority list to construct the assembly task dependency graph, and the following processing is performed: the assembly production process data set is normalized based on the multiple assembly production nodes to generate a normalized processing result; the multiple assembly production nodes are traversed according to the normalized processing result to determine multiple node task attributes; the multiple assembly production nodes are mapped to the objective function according to the multiple node task attributes to calculate and obtain multiple node contribution values; a task dependency matrix is ​​constructed based on the multiple node contribution values ​​combined with the multiple assembly production nodes, and the multiple process dependency coefficients are calculated based on the task dependency matrix.

[0007] In a possible implementation, the multiple production process information is scheduled and analyzed according to the assembly task dependency graph to generate an initial scheduling plan, and the following processing is performed: the assembly task dependency graph is traversed to perform graph analysis and construct multiple adjacency lists; depth-first search is used to topologically sort the multiple production process information according to the multiple adjacency lists to generate a dependency order list; the multiple production process information is traversed according to the dependency order list to perform path identification and determine task path parameters; the multiple production process information is allocated to the multiple assembly production nodes according to the task path parameters to determine an initial scheduling plan.

[0008] In a possible implementation, a simulation test is performed based on the initial scheduling plan in combination with the assembly production process data set to generate a simulation scheduling result, and the following processing is performed: historical assembly production record logs are retrieved, and historical assembly production environment parameters are extracted based on the historical assembly production record logs; simulation production environment parameters are constructed based on the historical assembly production environment parameters in combination with the assembly production process data set; the initial scheduling plan is executed according to the dependency order list for simulation based on the simulation production environment parameters to generate multiple scheduling simulation reports; based on the multiple scheduling simulation reports, the initial scheduling plan is traced back to perform testing to generate the simulation scheduling result.

[0009] In a possible implementation, reinforcement learning is performed on the initial scheduling plan according to the simulation scheduling results, and the initial scheduling plan is optimized according to the learning results, and the following processing is performed: the initial scheduling plan is tracked based on the multiple scheduling simulation reports to generate multiple scheduling completion progress data; production evaluation is performed based on the multiple scheduling completion progress data to determine multiple production scheduling scores; the initial scheduling plan is traversed and identified according to the multiple production scheduling scores to generate multiple labels to be optimized; reinforcement learning is performed on the initial scheduling plan based on the multiple labels to be optimized to generate the learning results.

[0010] In a possible implementation, reinforcement learning is performed on the initial scheduling plan based on the multiple labels to be optimized to generate the learning result, and the following processing is performed: data of the initial scheduling plan is divided based on the multiple labels to be optimized, and multiple scheduling parameter sets are determined, and the multiple scheduling parameter sets include an environment scheduling parameter set and a task scheduling parameter set; based on the environment scheduling parameter set, the states of the multiple production process information are defined to construct a scheduling state space; based on the task scheduling parameter set, the actions of the multiple production process information are defined to construct a scheduling action space; a reward function is introduced to combine the scheduling state space with the scheduling action space, and the combination result is synchronized to the reward function for calculation to generate a scheduling feedback value; the scheduling feedback value is added to the learning result.

[0011] The present application also provides an electric tricycle assembly production line process optimization scheduling system, including: a data partitioning module, used to extract the assembly production process data set of the electric tricycle, partition based on the assembly production process data set, and determine multiple assembly production nodes; a data sensing module, used to perform data sensing on the electric tricycle based on the multiple assembly production nodes, and construct an assembly task dependency graph of the electric tricycle; a scheduling analysis module, used to perform scheduling analysis on multiple production process information according to the assembly task dependency graph, and generate an initial scheduling plan; a simulation test module, used to perform simulation testing based on the initial scheduling plan and the assembly production process data set, and generate a simulation scheduling result; a reinforcement learning module, used to perform reinforcement learning on the initial scheduling plan according to the simulation scheduling result, optimize the initial scheduling plan according to the learning result, generate a process scheduling plan, and execute the process scheduling plan to adaptively optimize the electric tricycle assembly production line.

[0012] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0013] The present application provides an electric tricycle assembly production line process optimization scheduling method and system, which relates to the field of scheduling management technology. It solves the technical problem that traditional scheduling methods lack flexibility and are difficult to adapt to dynamic changes in the production environment, resulting in difficulty in fine scheduling of large-scale production lines. It achieves the technical effect of flexibly adapting to changes in the production environment by dynamically adjusting the scheduling plan, thereby improving the overall production capacity of the electric tricycle assembly production line. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical scheme of the embodiment of the present application, the accompanying drawings of the embodiment of the present application will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the system according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or a certain step or several steps of operations can be removed from these processes.

[0015] Figure 1 A flow chart of a method for optimizing the scheduling of processes in an electric tricycle assembly production line provided in an embodiment of the present application.

[0016] Figure 2 A schematic diagram of the structure of a process optimization and scheduling system for an electric tricycle assembly production line provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0018] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0019] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0020] The embodiment of the present application provides a method for optimizing the scheduling of the process of an electric tricycle assembly production line. The method is applied to a system for optimizing the scheduling of the process of an electric tricycle assembly production line. Figure 1 As shown, the method includes:

[0021] Step A100, extracting an assembly production process data set of an electric tricycle, dividing the assembly production process data set based on the assembly production process data set, and determining a plurality of assembly production nodes;

[0022] First, assembly production process-related data are extracted from the production line management system, sensor network and historical production records. The assembly production process data set may include task execution sequence, equipment status, resource usage and timestamp information. The assembly production process data set is divided into several nodes according to equipment resource allocation and execution time. The features of each node are further extracted, including the resources required for the task, time requirements and output quality, etc. The key tasks with the greatest impact on the overall efficiency in the production process are identified and used as key assembly production nodes to ensure that the node division can maximize production efficiency.

[0023] Execute step A200, perform data sensing on the electric tricycle based on the multiple assembly production nodes, and construct an assembly task dependency graph of the electric tricycle; in a possible implementation, step A200 further includes step A210, perform production analysis on the assembly production line of the electric tricycle according to the assembly production process data set, and define an objective function;

[0024] First, the features of the assembly production process data set, such as tasks, resource allocation, time requirements, and equipment status, are extracted, and the dependencies between tasks are identified. For example, task A can only be started after task B is completed. Then, the optimization goal is determined, such as minimizing task delays, maximizing resource utilization, or improving production efficiency, and the objective function is defined. The objective function may include minimizing process delays, maximizing equipment utilization, minimizing production costs, etc.

[0025] Execute step A220, map the multiple assembly production nodes to the objective function for calculation, and obtain multiple process dependency coefficients; in a possible implementation, step A220 further includes step A221, normalize the assembly production process data set based on the multiple assembly production nodes to generate a normalized processing result; execute step A222, traverse the multiple assembly production nodes according to the normalized processing result, and determine multiple node task attributes; execute step A223, map the multiple assembly production nodes to the objective function according to the multiple node task attributes, and calculate multiple node contribution values; execute step A224, construct a task dependency matrix based on the multiple node contribution values ​​and the multiple assembly production nodes, and calculate the multiple process dependency coefficients according to the task dependency matrix.

[0026] First, for each node data set in multiple assembly production nodes (such as task time, resource requirements, execution order, etc.), normalization technology is used to normalize it so that the value of the assembly production process data set falls into a fixed range (usually [0,1] or [-1,1]), and the multiple assembly production nodes are traversed according to the normalization processing results, which means that each assembly production node (i.e., production task) has multiple attributes, which may include execution time, resource requirements, and priority. Task dependency, etc., thereby determining multiple node task attributes, mapping the multiple assembly production nodes to the objective function for calculation according to the multiple node task attributes, means obtaining multiple node contribution values ​​by mapping the attributes of each node to the objective function including production efficiency, resource utilization and task completion time, etc. The multiple node contribution values ​​are used to characterize the impact of tasks on production efficiency, and can also dynamically adjust the contribution values ​​of nodes according to feedback information in the production process (such as production delays, equipment failures, etc.) to ensure priority scheduling of the most important tasks in the production process. A task dependency matrix is ​​constructed based on the multiple node contribution values ​​combined with the multiple assembly production nodes. The task dependency matrix is ​​a matrix used to represent the mutual dependence relationship between assembly production tasks. The elements in the task dependency matrix represent the degree of dependence between tasks. Each row of the matrix represents the dependency relationship of tasks, and each column represents the dependency of the task. Further, the multiple process dependency coefficients are calculated according to the task dependency matrix, which can be calculated by the following formula:

[0027] Among them, C i is the contribution value of node i, M ij is the dependency relationship between task i and task j, and the dependency coefficient D ij is the impact of task i on task k, C j is the contribution value of node j, and n is an arbitrary real number.

[0028] The task dependency matrix is ​​synchronized to the above formula for calculation in sequence to obtain multiple process dependency coefficients, optimize resource allocation, improve production efficiency, reduce production delays, and ensure the smooth completion of tasks.

[0029] Execute step A230, assign weights to the multiple assembly production nodes based on the multiple process dependency coefficients, and set a node priority list according to the weight assignment results; execute step A240, connect the multiple assembly production nodes according to the multiple process dependency coefficients according to the node priority list, and construct the assembly task dependency graph.

[0030] According to multiple process dependency coefficients, node weights are assigned to multiple assembly production nodes. For example, if the dependency is strong, the node weight is high, thereby generating a weight assignment result, and multiple assembly production nodes are processed in descending order to construct a priority list. Further, according to the priority list, multiple assembly production nodes are connected according to the process dependency coefficients, and there is a one-to-one correspondence between the multiple assembly production nodes and the multiple process dependencies. The process may include decomposing the entire electric tricycle assembly process into multiple independent subtasks, each subtask corresponding to a production node. By using a directed acyclic graph to represent task dependencies, that is, the nodes of the graph represent assembly tasks, the edges of the graph represent dependencies between tasks, and adding time weights in the graph to represent the time required to complete the task, an assembly task dependency graph is generated to achieve more efficient task scheduling of the electric tricycle assembly production line.

[0031] Execute step A300, perform scheduling analysis on multiple production process information according to the assembly task dependency graph, and generate an initial scheduling plan; in a possible implementation, step A300 further includes step A310, traverse the assembly task dependency graph to perform graph analysis, and construct multiple adjacency lists; execute step A320, use depth-first search to topologically sort the multiple production process information according to the multiple adjacency lists, and generate a dependency order list; execute step A330, traverse the multiple production process information according to the dependency order list to perform path identification, and determine task path parameters; execute step A340, allocate the multiple production process information to the multiple assembly production nodes according to the task path parameters, and determine the initial scheduling plan.

[0032] Traversing the assembly task dependency graph for graph parsing refers to traversing the assembly task dependency graph, obtaining each node in the assembly task dependency graph and its corresponding dependency relationship, parsing the dependency relationship, and building connection information between nodes, and building multiple adjacency lists, wherein the adjacency list is used to store the direct dependency relationship of each node. For each node in the assembly task dependency graph, the adjacency list stores a list corresponding to the node, which contains all nodes in the assembly task dependency graph that have a dependency relationship with the node.

[0033] Further, the multiple production process information is topologically sorted according to the multiple adjacency lists using depth-first search. First, the graph is traversed using depth-first search, each node is visited in sequence, and the order of its dependency relationships is recorded. For each node, all dependent nodes (i.e., child nodes) are visited first, and then the current node is visited to ensure the correctness of the dependency relationships. Then, multiple adjacency lists are traversed using depth-first search to mark the visited nodes. Whenever all dependent tasks (child nodes) of a node are visited, they are added to the topological order. Then, the topological sorting result is searched using depth-first search to obtain a dependency order list of tasks, wherein each task (node) is executed after the dependent task (predecessor task) to ensure the correctness of the scheduling.

[0034] In order to find the complete execution path from the start task to the end task and ensure that the task dependencies and execution order do not conflict, multiple production process information is traversed according to the dependency order list for path identification, which means traversing all task nodes according to the dependency order list, and determining the path and order of each task execution, traversing the order list, tracking the predecessor tasks of the current task, determining the task path and calculating the task path parameters (such as execution time, resource requirements, etc.), and the task path parameters may include task execution time, resource requirements, task dependencies, etc.

[0035] Finally, the multiple production process information is allocated to the multiple assembly production nodes according to the task path parameters, which means that each task is allocated to the corresponding assembly production node according to the task path parameters and resource requirements, and the execution order of each task is determined, that is, a preliminary scheduling plan is generated according to the task path parameters and task dependencies. The initial scheduling plan includes the execution order, resource allocation, priority and execution time of each task, ensuring that the task dependencies are correctly handled, thereby improving the efficiency of production line scheduling and resource utilization.

[0036] Execute step A400, perform simulation testing based on the initial scheduling plan in combination with the assembly production process data set, and generate a simulation scheduling result; in a possible implementation, step A400 further includes step A410, calling the historical assembly production record log, and extracting the historical assembly production environment parameters based on the historical assembly production record log; execute step A420, analyze the historical assembly production environment parameters in combination with the assembly production process data set, and construct the simulation production environment parameters; execute step A430, execute the initial scheduling plan according to the dependency order list based on the simulation production environment parameters for simulation, and generate multiple scheduling simulation reports; execute step A440, trace back to the initial scheduling plan according to the multiple scheduling simulation reports for testing, and generate the simulation scheduling result.

[0037] First, the historical production data of the electric tricycle assembly production line is collected through an automated data acquisition system or manual records. The record content includes the start and end time of each assembly task, the resources used (such as equipment, workers, tools, etc.), the quality of task completion, etc., and the historical assembly production record log can cover various production stages and equipment status, such as equipment failure, resource occupation, etc., and further extract historical assembly production environment parameters based on the historical assembly production record log. The historical assembly production environment parameters can include equipment performance status, production environment conditions (such as temperature, humidity), resource allocation (such as manpower, equipment), task execution time, etc., and then simulate the possible production environment parameters based on the historical assembly production environment parameters. Based on the historical data, the possible fluctuation range of environmental parameters and the correlation between them are assumed, and different production scenarios are simulated according to the historical data, such as the impact of equipment failure, resource shortage, environmental changes, etc. on task execution. The simulated production environment parameters can be generated by using statistical methods or machine learning models (such as Monte Carlo simulation, Bayesian network).

[0038] Furthermore, the simulation is performed based on the simulated production environment parameters and the initial scheduling scheme is executed according to the dependency order list, which means that the dependency order list is used as input to simulate the initial scheduling scheme, and the entire production process can be simulated by simulation software or a custom simulation algorithm (such as discrete event simulation DES). The simulation process needs to consider the dependency relationship, resource usage, production environment conditions, etc. of each production node. At the same time, during the simulation process, tasks are executed one by one in a predetermined order, and the system feeds back information such as resource usage, task completion, and task delay in real time. After the simulation process is completed, multiple scheduling simulation reports are generated, and the reports contain key information such as the execution status, resource utilization, task completion time, and task delay of each task.

[0039] Finally, based on the data in the simulation report, we trace back to the initial scheduling plan. We use the task completion time, resource allocation and other information contained in the report to determine whether the scheduling plan is reasonable, and check whether the execution order, resource utilization, time window, etc. between tasks are in line with expectations. If problems are found in the scheduling plan during the simulation (such as delays in certain tasks, uneven resource use, etc.), we can use backtracking analysis to identify bottlenecks or unreasonable parts. Based on the backtracking analysis, we can adjust the execution order, resource allocation or dependency of the tasks, generate the simulation scheduling results, optimize task scheduling, and improve the overall operating efficiency of the production line.

[0040] Next, step A500 is executed to perform reinforcement learning on the initial scheduling plan according to the simulation scheduling result, optimize the initial scheduling plan according to the learning result, generate a process scheduling plan, and execute the process scheduling plan to adaptively optimize the electric three-wheel assembly production line.

[0041] In a possible implementation, step A500 further includes step A510, tracking the initial scheduling plan based on the multiple scheduling simulation reports to generate multiple scheduling completion progress data; executing step A520, performing production evaluation according to the multiple scheduling completion progress data to determine multiple production scheduling scores; executing step A530, traversing and identifying the initial scheduling plan according to the multiple production scheduling scores to generate multiple tags to be optimized;

[0042] Tracking the initial scheduling scheme based on the multiple scheduling simulation reports refers to tracking the actual completion status of each production process. For example, the actual start time of the task, the actual completion time, and whether the task is completed on time are recorded. At the same time, in the multiple scheduling simulation reports generated during the simulation process, the execution time, resource usage, task completion status, delay and other information of each task are recorded. Based on the multiple scheduling reports, the execution status of each task in the initial scheduling scheme is tracked to ensure that all tasks start and complete on time, thereby generating multiple scheduling completion progress data, which may include task identification, actual task start time, actual task completion time, task completion time difference, and resource consumption during task execution.

[0043] Further, production evaluation is performed according to the multiple scheduling completion progress data, which means measuring the efficiency of the production process and the execution quality of the task through the tracked scheduling progress data. The evaluation indicators may include the timeliness of task completion, the efficiency of resource use, task delay, etc., and a scheduling score is generated for each production node. The scoring basis may be the degree of task completion, resource utilization, task delay, etc., traverse all task nodes in the initial scheduling plan, calculate the production scheduling score of each node, thereby obtaining multiple production scheduling scores, and identify each task node according to the multiple production scheduling scores to determine which task nodes need to be optimized. If there is a task node with a low score (such as large task delay or low resource utilization), a corresponding label to be optimized is generated for the node. The label to be optimized is used to characterize the task node that needs to be further optimized, which is usually a bottleneck or inefficient part in the production scheduling plan, thereby determining multiple labels to be optimized.

[0044] Execute step A540 to perform reinforcement learning on the initial scheduling scheme based on the multiple labels to be optimized to generate the learning result.

[0045] In one possible implementation, step A540 further includes step A541, dividing the data of the initial scheduling plan based on the multiple labels to be optimized, and determining multiple scheduling parameter sets, wherein the multiple scheduling parameter sets include an environment scheduling parameter set and a task scheduling parameter set; executing step A542, defining the states of the multiple production process information based on the environment scheduling parameter set, and constructing a scheduling state space; executing step A543, defining the actions of the multiple production process information based on the task scheduling parameter set, and constructing a scheduling action space; executing step A544, introducing a reward function, combining the scheduling state space with the scheduling action space, synchronizing the combination result to the reward function for calculation, and generating a scheduling feedback value; executing step A545, adding the scheduling feedback value to the learning result.

[0046] In order to determine the scheduling parameters for each task node and production process, the initial scheduling plan is first divided according to the multiple tags to be optimized as determined above. The focus of data division is to distinguish between the environmental scheduling parameter set and the task scheduling parameter set, so that the scheduling optimization can be more targeted and efficient. According to the information in the multiple tags to be optimized, the initial scheduling plan is split into multiple subsets, each of which corresponds to a different task node or production process. Each subset will correspond to its specific environment and task scheduling parameters, and multiple scheduling parameter sets will be determined. The environmental scheduling parameter set includes production environment variables (such as equipment status, resource availability, production line configuration, etc.), which affect the scheduling strategy of the entire production process. The task scheduling parameter set includes the specific attributes of each task node (such as task execution time, resource requirements, dependencies, etc.).

[0047] Further, the status of each production process is defined according to the environmental scheduling parameter set. For example, whether the equipment is available, whether the resources meet the task requirements, whether the production environment is suitable, etc. For each task node (such as welding, assembly, testing, etc.), there may be different execution results under different environmental conditions, so the status corresponding to each task can have multiple possible expressions. For example, if equipment A is available and task B has not been completed, the status may be "equipment A is available, task B needs to wait".

[0048] Further, the operability of each task node is defined according to the task scheduling parameter set. The execution order, priority, resource allocation, etc. of the tasks can be regarded as part of the action space, where the actions of each production node can include executing tasks, rescheduling, and adjusting resources. For example, for task A, possible actions include: "allocating resources" or "adjusting the task order to prioritize execution".

[0049] Finally, a reward function is introduced, which is used to measure the quality of scheduling actions. Good scheduling results will receive high rewards, while unsatisfactory scheduling results will receive low rewards or penalties. The current scheduling state and action space are combined to obtain all possible state-action pairs. Each state-action pair corresponds to a calculated reward value. The reward value is calculated for each state-action pair according to the reward function to obtain the scheduling feedback value. The feedback value reflects the quality of the decision. The scheduling feedback value includes positive feedback value and negative feedback value. The calculated scheduling feedback value is used to update the strategy in the reinforcement learning process, generate learning results, reduce resource waste, ensure that production tasks are completed on time, and improve overall production efficiency.

[0050] Then, based on the results of reinforcement learning, the execution order, resource requirements, dependencies and other parameters of each task node are first extracted from the initial scheduling plan, and these parameters are optimized according to the learning results. For example, for common bottlenecks or delays in production lines, reinforcement learning may have adjusted the task order, rearranged the priority of resources, and even adjusted the layout of the production line. Combined with the learning results and production needs, an optimized process scheduling plan is generated, and appropriate resources and time are allocated to each production process to ensure that each task can be completed on time and efficiently. The optimized task priority, resource requirements, task execution time and other information are integrated, and a process scheduling plan is constructed according to the dependency and priority. The optimized process scheduling plan is applied to the actual production line, so that each task in the scheduling plan is executed according to the predetermined priority and resource allocation to ensure the smooth operation of the production line. Each task node is scheduled according to the completion of the predecessor task, and the task is executed using appropriate resources. At the same time, if a production node (such as a welding process) has equipment failure during the execution process, causing a delay in the production line, the system can automatically adjust the task order based on the learning results, give priority to tasks that do not depend on the node, and continue to execute related tasks after the failure is repaired. Through this dynamic adjustment, the production line can maintain an efficient operating state when facing various practical problems. Real-time feedback and dynamic optimization enable the production line to continuously adapt to changes in the external environment and achieve long-term efficient and stable production.

[0051] The embodiments of the present application solve the technical problem that traditional scheduling methods lack flexibility and are difficult to adapt to dynamic changes in the production environment, resulting in difficulty in fine-grained scheduling of large-scale production lines. The embodiment of the present application achieves the technical effect of flexibly adapting to changes in the production environment by dynamically adjusting the scheduling plan and improving the overall production capacity of the electric tricycle assembly production line.

[0052] In the above, refer to Figure 1 A method for optimizing the process scheduling of an electric tricycle assembly production line according to an embodiment of the present application is described in detail. Figure 2 A process optimization scheduling system for an electric tricycle assembly production line according to an embodiment of the present application is described.

[0053] According to an embodiment of the present application, an electric tricycle assembly production line process optimization scheduling system is used to solve the technical problem that the traditional scheduling method lacks flexibility and is difficult to adapt to the dynamic changes of the production environment, resulting in difficulty in fine scheduling of large-scale production lines, and achieves the technical effect of flexibly adapting to changes in the production environment by dynamically adjusting the scheduling scheme to improve the overall production capacity of the electric tricycle assembly production line. An electric tricycle assembly production line process optimization scheduling system includes: a data partitioning module 10, a data sensing module 20, a scheduling analysis module 30, a simulation test module 40, and a reinforcement learning module 50.

[0054] A data partitioning module 10 is used to extract an assembly production process data set of an electric tricycle, and to partition the assembly production process data set to determine a plurality of assembly production nodes;

[0055] A data sensing module 20, configured to perform data sensing on the electric tricycle based on the plurality of assembly production nodes, and construct an assembly task dependency graph of the electric tricycle;

[0056] A scheduling analysis module 30, configured to perform scheduling analysis on multiple production process information according to the assembly task dependency graph and generate an initial scheduling plan;

[0057] A simulation test module 40, configured to perform simulation testing based on the initial scheduling scheme in combination with the assembly production process data set to generate a simulation scheduling result;

[0058] The reinforcement learning module 50 is used to perform reinforcement learning on the initial scheduling plan according to the simulation scheduling results, optimize the initial scheduling plan according to the learning results, generate a process scheduling plan, and execute the process scheduling plan to adaptively optimize the electric three-wheel assembly production line.

[0059] The specific configuration of the data sensing module 20 will be described in detail below. As described above, based on the multiple assembly production nodes, data sensing is performed on the electric tricycle to construct the assembly task dependency graph of the electric tricycle. The data sensing module 20 may further include: performing production analysis on the assembly production line of the electric tricycle according to the assembly production process data set, defining the objective function; mapping the multiple assembly production nodes to the objective function for calculation, and obtaining multiple process dependency coefficients; weighting the multiple assembly production nodes based on the multiple process dependency coefficients, and setting a node priority list according to the weight allocation result; connecting the multiple assembly production nodes according to the multiple process dependency coefficients according to the node priority list, and constructing the assembly task dependency graph.

[0060] The specific configuration of the data sensing module 20 will be described in detail below. As described above, the multiple assembly production nodes are connected according to the multiple process dependency coefficients according to the node priority list to construct the assembly task dependency graph. The data sensing module 20 may further include: normalizing the assembly production process data set based on the multiple assembly production nodes to generate a normalized processing result; traversing the multiple assembly production nodes according to the normalized processing result to determine multiple node task attributes; mapping the multiple assembly production nodes to the objective function according to the multiple node task attributes to calculate and obtain multiple node contribution values; constructing a task dependency matrix based on the multiple node contribution values ​​combined with the multiple assembly production nodes, and calculating the multiple process dependency coefficients according to the task dependency matrix.

[0061] The specific configuration of the scheduling analysis module 30 will be described in detail below. As described above, the scheduling analysis of multiple production process information is performed according to the assembly task dependency graph to generate an initial scheduling plan. The scheduling analysis module 30 may further include: traversing the assembly task dependency graph for graph parsing and constructing multiple adjacency lists; topologically sorting the multiple production process information according to the multiple adjacency lists using depth-first search to generate a dependency order list; traversing the multiple production process information according to the dependency order list to perform path identification and determine task path parameters; allocating the multiple production process information to the multiple assembly production nodes according to the task path parameters to determine the initial scheduling plan.

[0062] The specific configuration of the simulation test module 40 will be described in detail below. As described above, based on the initial scheduling scheme combined with the assembly production process data set, the simulation test is performed to generate a simulation scheduling result. The simulation test module 40 may further include: retrieving the historical assembly production record log, extracting the historical assembly production environment parameters based on the historical assembly production record log; analyzing the historical assembly production environment parameters in combination with the assembly production process data set to construct the simulation production environment parameters; executing the initial scheduling scheme according to the dependency order list based on the simulation production environment parameters to perform simulation simulation and generate multiple scheduling simulation reports; tracing back to the initial scheduling scheme according to the multiple scheduling simulation reports to perform testing and generate the simulation scheduling result.

[0063] The specific configuration of the reinforcement learning module 50 will be described in detail below. As described above, the initial scheduling scheme is reinforced learned based on the simulation scheduling results, and the initial scheduling scheme is optimized based on the learning results. The reinforcement learning module 50 may further include: tracking the initial scheduling scheme based on the multiple scheduling simulation reports to generate multiple scheduling completion progress data; performing production evaluation based on the multiple scheduling completion progress data to determine multiple production scheduling scores; traversing and identifying the initial scheduling scheme according to the multiple production scheduling scores to generate multiple labels to be optimized; and reinforcing the initial scheduling scheme based on the multiple labels to be optimized to generate the learning results.

[0064] The specific configuration of the reinforcement learning module 50 will be described in detail below. As described above, before the set of items to be used is classified according to the storage order, the reinforcement learning module 50 may further include: dividing the data of the initial scheduling plan based on the multiple tags to be optimized, determining multiple scheduling parameter sets, and the multiple scheduling parameter sets include an environment scheduling parameter set and a task scheduling parameter set; defining the state of the multiple production process information based on the environment scheduling parameter set, and constructing a scheduling state space; defining the action of the multiple production process information based on the task scheduling parameter set, and constructing a scheduling action space; introducing a reward function, combining the scheduling state space with the scheduling action space, synchronizing the combination result to the reward function for calculation, and generating a scheduling feedback value; adding the scheduling feedback value to the learning result.

[0065] An electric tricycle assembly production line process optimization scheduling system provided in an embodiment of the present application can execute an electric tricycle assembly production line process optimization scheduling method provided in any embodiment of the present application, and has functional modules and beneficial effects corresponding to the execution method.

[0066] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present application.

[0067] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A method for optimizing the process scheduling of an electric tricycle assembly production line, characterized in that: The method comprises: Extracting an assembly production process data set of an electric tricycle, dividing the assembly production process data set based on the assembly production process data set, and determining a plurality of assembly production nodes; Performing data sensing on the electric tricycle based on the multiple assembly production nodes to construct an assembly task dependency graph of the electric tricycle; Perform scheduling analysis on multiple production process information according to the assembly task dependency graph to generate an initial scheduling plan; Perform simulation testing based on the initial scheduling plan and the assembly production process data set to generate a simulation scheduling result; The initial scheduling plan is reinforced learned according to the simulation scheduling result, the initial scheduling plan is optimized according to the learning result, a process scheduling plan is generated, and the process scheduling plan is executed to adaptively optimize the electric three-wheel assembly production line.

2. The method for optimizing the process scheduling of an electric tricycle assembly production line according to claim 1, characterized in that: Based on the multiple assembly production nodes, data sensing is performed on the electric tricycle to construct an assembly task dependency graph of the electric tricycle, and the method includes: Performing production analysis on the assembly line of the electric tricycle according to the assembly production process data set, and defining an objective function; Mapping the multiple assembly production nodes to the objective function for calculation to obtain multiple process dependency coefficients; Based on the multiple process dependency coefficients, weights are assigned to the multiple assembly production nodes, and a node priority list is set according to the weight assignment result; According to the node priority list, the multiple assembly production nodes are connected according to the multiple process dependency coefficients to construct the assembly task dependency graph.

3. The method for optimizing the process scheduling of an electric tricycle assembly production line according to claim 2, characterized in that: The multiple assembly production nodes are mapped to the objective function for calculation to obtain multiple process dependency coefficients, the method comprising: Normalizing the assembly production process data set based on the multiple assembly production nodes to generate a normalized processing result; Traversing the plurality of assembly production nodes according to the normalization processing result to determine a plurality of node task attributes; According to the multiple node task attributes, the multiple assembly production nodes are mapped to the objective function for calculation to obtain multiple node contribution values; A task dependency matrix is ​​constructed according to the multiple node contribution values ​​and the multiple assembly production nodes, and the multiple process dependency coefficients are calculated according to the task dependency matrix.

4. The method for optimizing the process scheduling of an electric tricycle assembly production line according to claim 1, characterized in that: The method of performing scheduling analysis on the plurality of production process information according to the assembly task dependency graph and generating an initial scheduling plan includes: Traversing the assembly task dependency graph to perform graph parsing and construct multiple adjacency lists; Using depth-first search to topologically sort the multiple production process information according to the multiple adjacency lists to generate a dependency order list; Traversing the plurality of production process information according to the dependency order list to perform path identification and determine task path parameters; The plurality of production process information are allocated to the plurality of assembly production nodes according to the task path parameters, and an initial scheduling scheme is determined.

5. The method for optimizing the process scheduling of an electric tricycle assembly production line according to claim 4, characterized in that: Based on the initial scheduling plan and the assembly production process data set, a simulation test is performed to generate a simulation scheduling result, the method comprising: Retrieving a historical assembly production record log, and extracting historical assembly production environment parameters based on the historical assembly production record log; Analyze the historical assembly production environment parameters in combination with the assembly production process data set to construct simulation production environment parameters; Execute the initial scheduling scheme according to the dependency order list based on the simulated production environment parameters to perform simulation and generate multiple scheduling simulation reports; According to the multiple scheduling simulation reports, the initial scheduling plan is traced back to perform testing to generate the simulation scheduling result.

6. The method for optimizing the process scheduling of an electric tricycle assembly production line according to claim 5, characterized in that: Reinforcement learning is performed on the initial scheduling scheme according to the simulation scheduling result, and the initial scheduling scheme is optimized according to the learning result. The method includes: Tracking the initial scheduling plan based on the multiple scheduling simulation reports to generate multiple scheduling completion progress data; Performing production evaluation based on the plurality of scheduling completion progress data to determine a plurality of production scheduling scores; Traversing and marking the initial scheduling scheme according to the multiple production scheduling scores to generate multiple labels to be optimized; Reinforcement learning is performed on the initial scheduling scheme based on the multiple labels to be optimized to generate the learning result.

7. The method for optimizing the process scheduling of an electric tricycle assembly production line according to claim 6, characterized in that: Performing reinforcement learning on the initial scheduling scheme based on the multiple labels to be optimized to generate the learning result, the method comprising: Divide the initial scheduling scheme into data based on the multiple tags to be optimized, and determine multiple scheduling parameter sets, wherein the multiple scheduling parameter sets include an environment scheduling parameter set and a task scheduling parameter set; Based on the environmental scheduling parameter set, the states of the plurality of production process information are defined to construct a scheduling state space; Based on the task scheduling parameter set, actions are defined for the plurality of production process information to construct a scheduling action space; Introducing a reward function, combining the scheduling state space with the scheduling action space, synchronizing the combination result to the reward function for calculation, and generating a scheduling feedback value; The scheduling feedback value is added to the learning result.

8. An electric tricycle assembly production line process optimization scheduling system, characterized in that: The system is used to implement the method for optimizing the scheduling of the process of an electric tricycle assembly production line according to any one of claims 1 to 7, and the system comprises: A data partitioning module is used to extract an assembly production process data set of an electric tricycle, and to partition the assembly production process data set to determine a plurality of assembly production nodes; A data sensing module, used for performing data sensing on the electric tricycle based on the multiple assembly production nodes, and constructing an assembly task dependency graph of the electric tricycle; A scheduling analysis module, used to perform scheduling analysis on multiple production process information according to the assembly task dependency graph and generate an initial scheduling plan; A simulation test module, used to perform simulation test based on the initial scheduling plan combined with the assembly production process data set to generate a simulation scheduling result; The reinforcement learning module is used to perform reinforcement learning on the initial scheduling plan according to the simulation scheduling results, optimize the initial scheduling plan according to the learning results, generate a process scheduling plan, and execute the process scheduling plan to adaptively optimize the electric three-wheel assembly production line.

Images