Workpiece multi-process collaborative machining method

Abstract

The application discloses a kind of workpiece multi-process collaborative processing methods, belong to intelligent manufacturing processing technical field, to solve the technical problems of path redundancy, low resource utilization in multi-workpiece, multi-station collaborative processing, specifically includes: based on global processing map and workpiece demand, with the minimum shared station number as target, combined with processing precision constraint to complete workpiece multi-objective clustering, generate the to-be-processed cluster containing shared station;Again, to-be-processed cluster is used as vertex to construct station associated topological graph, calculate process associated value and select candidate station;Iterative selection optimal shared station, with path length minimization as target to solve collaborative processing path, dynamically update topological graph and candidate station;Finally, collaborative path is summarized and the remaining workpiece independent path is supplemented, and the final scheme is output;The application improves processing efficiency, resource utilization and path planning rationality, and is suitable for flexible manufacturing workshop multi-process collaborative processing scene.

Description

Technical Field

[0001] This invention belongs to the field of intelligent manufacturing processing technology, specifically a method for multi-process collaborative processing of workpieces. Background Technology

[0002] In the fields of flexible manufacturing and multi-process collaborative processing, parallel processing of multiple workpieces and multiple workstations has become the mainstream production mode. Current processing planning mostly adopts independent scheduling for single workpieces or static clustering scheduling, which has obvious drawbacks: on the one hand, workpiece processing workstations are scattered, the utilization rate of shared workstations is low, which easily leads to idle equipment resources and path redundancy, resulting in low overall processing efficiency; on the other hand, existing scheduling strategies lack a global perspective, do not combine processing accuracy constraints and dynamic working condition adjustments, and are difficult to adapt to the differentiated processing needs of workpieces, easily leading to problems such as process conflicts and unreasonable path planning.

[0003] Furthermore, traditional processing planning relies heavily on fixed workstation sequences and static path calculations, failing to respond in real-time to dynamic information such as equipment status and workstation distance changes, making it difficult to achieve efficient collaboration among multiple processes and workpieces. Existing technologies lack a global planning scheme centered on shared workstations that balances accuracy constraints and path optimization, failing to effectively balance processing efficiency, resource utilization, and processing accuracy, thus hindering the intelligent and efficient development of flexible manufacturing systems. Therefore, a collaborative processing method for multiple workpiece processes is urgently needed to address the technical bottlenecks of resource waste, rigid planning, and poor collaboration in existing technologies. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention proposes a multi-process collaborative machining method for workpieces, aiming to solve the technical problems of path redundancy and low resource utilization in multi-workpiece, multi-station collaborative machining. Specifically, it includes: based on a global machining map and workpiece requirements, minimizing the number of shared workstations as the objective, and combining machining accuracy constraints to complete multi-objective clustering of workpieces, generating clusters of workpieces to be processed containing shared workstations; then, constructing a workstation association topology graph with the workpiece clusters as vertices, calculating the process association value, and screening candidate workstations; iteratively selecting the optimal shared workstation, solving the collaborative machining path with the objective of minimizing path length, and dynamically updating the topology graph and candidate workstations; finally, summarizing the collaborative paths and supplementing the remaining independent workpiece paths, and outputting the final solution. This invention improves machining efficiency, resource utilization, and path planning rationality, and is suitable for multi-process collaborative machining scenarios in flexible manufacturing workshops.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A multi-process collaborative machining method for workpieces is applied to a machining planning system. The machining planning system includes an autonomous planning agent, multiple workpieces, and machining equipment. The autonomous planning agent is configured with a global machining map. The method includes:

[0007] Based on the global processing map and the demand information of the workpieces to be processed, multi-objective clustering is performed on the workpieces to be processed to obtain multiple clusters to be processed; each cluster to be processed includes at least one shared workstation.

[0008] Using all the clusters to be processed as vertices, establish process association relationships between different clusters to be processed that share a workstation through the corresponding shared workstation, construct a workstation association topology graph, calculate the initial upper bound of the value of each process association relationship in the workstation association topology graph, and output a candidate workstation set after filtering.

[0009] Iterative processing of candidate workstation set: Each time, a shared workstation is selected from the candidate workstation set. Based on all the processing clusters associated with the shared workstation, a collaborative processing path is solved. The workpieces in the collaborative processing path are synchronously deleted from the corresponding processing clusters. The workstation association topology and the candidate workstation set are updated until the candidate workstation set is empty.

[0010] The collaborative processing paths generated in each iteration are summarized, and the remaining unprocessed workpieces in the cluster to be processed after each iteration are supplemented as independent paths. The final collaborative processing path is then integrated and output.

[0011] Specifically, based on the global processing map and the demand information of the workpieces to be processed, multi-objective clustering is performed on the workpieces to be processed to obtain multiple clusters to be processed, including:

[0012] Based on the global machining map, the initial machining station sequence required for each workpiece to be processed is obtained; the initial machining station sequence includes the identification of the machining equipment station that the workpiece passes through in sequence.

[0013] Based on the requirements of the workpieces to be processed, the target processing accuracy level of each workpiece is obtained; the target processing accuracy level is used to constrain the range of workstations that can be shared during the clustering process.

[0014] A multi-objective clustering algorithm is adopted, with minimizing the total number of shared workstations in all clusters to be processed as the first optimization objective and the target processing accuracy level corresponding to each workpiece as the constraint condition. Clustering operation is performed on all workpieces to be processed to obtain multiple clusters to be processed.

[0015] For each cluster to be processed, extract the processing stations common to all workpieces within the cluster, and use each extracted processing station as a shared station for the cluster. Finally, output the cluster identifier, the list of workpiece identifiers contained in the cluster, and the corresponding list of shared stations for each cluster to be processed.

[0016] Specifically, the step of establishing process association relationships between different clusters of work to be processed through corresponding shared workstations, using all clusters to be processed as vertices, and constructing a workstation association topology graph, includes:

[0017] Each cluster identifier of the cluster to be processed is treated as a vertex;

[0018] Iterate through any two different vertices that are to be processed, and for each pair of clusters to be processed, check whether there is a common shared workstation identifier in the shared workstation list of each pair of clusters to be processed.

[0019] If there are identical shared workstation identifiers, the shared workstation identifiers will be used as the process association between the clusters to be processed, and an edge will be used to connect the two vertices corresponding to the clusters to be processed.

[0020] All generated vertices and edges are combined to construct an undirected workstation association topology graph; each edge in the workstation association topology graph records the identifier of the shared workstation it is associated with, as well as the cluster identifiers of the two clusters to be processed connected by the edge.

[0021] Specifically, the calculation of the initial upper bound of the value of each process association in the workstation association topology graph, and the output of a candidate workstation set after filtering, includes:

[0022] Obtain the cluster identifier of the first cluster to be processed and the cluster identifier of the second cluster to be processed associated with each edge in the workstation association topology graph;

[0023] Based on the number of remaining unprocessed workpieces in each current processing cluster, the current number of remaining unprocessed workpieces in the first processing cluster is obtained as the first quantity, and the current number of remaining unprocessed workpieces in the second processing cluster is obtained as the second quantity.

[0024] Take the minimum value between the first quantity and the second quantity, and multiply the minimum value by a preset unit processing efficiency coefficient to obtain the initial upper bound of the edge value; the unit processing efficiency coefficient is the efficiency value generated by the processing equipment to process a single workpiece once on a shared workstation, which is pre-stored by the global processing map;

[0025] Create an empty set of candidate workstations, filter out edges whose initial value upper bound is greater than zero, and add the shared workstation identifier corresponding to each filtered edge to the set of candidate workstations.

[0026] Specifically, the process of solving the collaborative processing path includes:

[0027] Select the shared workstation with the largest upper bound of initial value from the current set of candidate workstations, and set the shared workstation as the currently selected shared workstation;

[0028] Based on the constructed workstation association topology, obtain the cluster identifiers of all the clusters to be processed associated with the currently selected shared workstation, and use the set of cluster identifiers as the current associated cluster set;

[0029] Obtain the remaining workpieces in each cluster to be processed in the current associated cluster set that have not yet been included by any generated collaborative processing path, and construct all the obtained remaining workpieces into a queue of workpieces to be processed in the current batch.

[0030] With the goal of minimizing the total processing path length of the current batch of workpieces to be processed, the collaborative processing path of the current batch of workpieces to be processed on the currently selected shared workstation is solved, and finally the identifier of the currently selected shared workstation and its corresponding collaborative processing path are output.

[0031] Specifically, solving the collaborative processing path of the current batch of workpieces to be processed on the currently selected shared workstation includes:

[0032] The currently selected shared workstation is designated as the pivot workstation, and the search state is initialized. The search state includes: the current workstation position, the set of completed workpieces, and the current cumulative path length. The initial value of the current workstation position is set to the pivot workstation, the initial value of the set of completed workpieces is set to an empty set, and the initial value of the current cumulative path length is set to zero.

[0033] Take the current batch of workpieces to be processed as the set of workpieces to be processed. If the set of workpieces to be processed is empty, then take the current cumulative path length as the global optimal path length.

[0034] If the set of workpieces to be processed is not empty, then select a candidate workpiece that has not been marked as completed from the set of workpieces to be processed, and obtain a complete sequence of processing stations consisting of all necessary stations that start from the current station position and pass through the candidate workpiece in sequence, according to the global processing map; the necessary stations include the pivot station and the other necessary stations of the candidate workpiece recorded in the global processing map.

[0035] Calculate the path increment length required to execute a complete processing station sequence; the path increment length is the sum of the distances between adjacent stations in the complete processing station sequence.

[0036] Specifically, solving the collaborative processing path of the current batch of workpieces to be processed on the currently selected shared workstation further includes:

[0037] If the current cumulative path length plus the path increment length is less than the currently recorded global optimal path length, then the current workstation position is updated to the last workstation of the complete processing workstation sequence, the candidate workpiece is added to the set of completed processing workpieces, the current cumulative path length is increased by the path increment length, the candidate workpiece is marked as completed, and the next incomplete workpiece is processed; otherwise, another candidate workpiece is selected directly.

[0038] When all workpieces in the set of workpieces to be processed are marked as completed, the current cumulative path length is taken as the global optimal path length, and the corresponding complete processing station sequence is recorded as the collaborative processing path on the currently selected shared station.

[0039] Specifically, obtaining the complete machining station sequence consisting of all necessary stations that start from the current workstation position and sequentially pass through the candidate workpiece includes:

[0040] Use the current workstation position as the starting workstation identifier;

[0041] The initial processing station sequence of the candidate workpiece is obtained from the global processing map, and all station identifiers before the pivot station are extracted from the initial processing station sequence as the preceding station set, and all station identifiers after the pivot station are extracted as the following station set.

[0042] According to the original order in the initial processing station sequence, the station identifiers in the preceding station set and the following station set are sorted to generate the preceding station execution sequence and the following station execution sequence.

[0043] Based on the shortest path distance between any two workstations recorded in the global processing map, the shortest path from the starting workstation identifier to the first workstation in the preceding workstation execution sequence, the shortest path between two adjacent workstations in the preceding workstation execution sequence, the shortest path from the last workstation in the preceding workstation execution sequence to the pivot workstation, the shortest path from the pivot workstation to the first workstation in the following workstation execution sequence, and the shortest path between two adjacent workstations in the following workstation execution sequence are calculated sequentially.

[0044] All shortest paths are concatenated end to end in the execution order to generate a complete sequence of processing stations.

[0045] Specifically, the collaborative processing path generated by the iteration takes the remaining unprocessed workpieces within the cluster to be processed after the iteration as independent paths to supplement the final collaborative processing path, including:

[0046] Obtain all collaborative processing paths output in each iteration, and arrange them in the order of their generation to obtain the planned path sequence;

[0047] Obtain the remaining unprocessed workpieces in each cluster to be processed, and summarize all the remaining unprocessed workpieces into a set of remaining workpieces;

[0048] For each workpiece in the remaining workpiece set, the initial processing station sequence of the workpiece is obtained from the global processing map, and the initial processing station sequence is used as the independent processing path of the workpiece.

[0049] All generated independent processing paths are appended to the end of the planned path sequence to generate a complete processing path sequence, which is then output as the final collaborative processing path.

[0050] Specifically, the global processing map configured within the autonomous planning intelligent body is stored using a hierarchical graph structure, which includes a workstation layer, an equipment layer, and a path layer.

[0051] The workstation layer includes the workstation identifier, workstation position coordinates, and processing capacity vector for each processing workstation;

[0052] The equipment layer includes the equipment identifier of each processing device, a list of workstation identifiers under its jurisdiction, and equipment movement speed parameters;

[0053] The path layer contains a set of reachable paths between any two workstations, and each reachable path records the path length and path travel time parameters;

[0054] The autonomous planning agent also receives the actual path travel time reported by the processing equipment through the update interface, and corrects the path travel time parameters in the path layer online.

[0055] Compared with the prior art, the beneficial effects of the present invention are:

[0056] 1. This invention proposes a multi-process collaborative processing method for workpieces. By performing multi-objective clustering on workpieces and extracting shared workstations, the method effectively reduces the repeated occupation of workstation resources and improves the utilization rate of processing equipment and workstations. At the same time, it constructs a workstation association topology map and filters high-value candidate workstations, realizing global collaborative planning for multiple workpieces and multiple processes, reducing the overall processing path length, and improving processing efficiency and production flexibility.

[0057] 2. This invention proposes a multi-process collaborative machining method for workpieces, which adopts a dynamic iterative optimization strategy. It can update the topology and machining path according to the real-time working conditions, adapt to different machining accuracy requirements, and avoid process conflicts and resource idleness. The hierarchical global machining map supports online correction of path parameters, which enhances the real-time performance and reliability of the planning scheme, and provides an efficient and stable multi-process collaborative machining solution for flexible manufacturing workshops. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of a multi-process collaborative machining method for a workpiece according to the present invention;

[0059] Figure 2 This is a flowchart illustrating the principle of a multi-process collaborative machining method for workpieces according to the present invention. Detailed Implementation

[0060] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0061] Please see Figures 1-2 The present invention provides an embodiment of a multi-process collaborative machining method for workpieces, applied to a machining planning system. The machining planning system includes an autonomous planning agent, multiple workpieces, and machining equipment. The autonomous planning agent is configured with a global machining map. The method includes steps S1 to S4 and is applicable to the field of pen tip machining, comprising the following steps:

[0062] S1: Based on the global processing map and the demand information of the workpieces to be processed, perform multi-objective clustering on the workpieces to be processed to obtain multiple clusters to be processed; each cluster to be processed includes at least one shared workstation.

[0063] The global processing map configured within the autonomous planning intelligent body is stored using a hierarchical graph structure, which includes a workstation layer, an equipment layer, and a path layer.

[0064] The workstation layer includes workstation identifiers, workstation location coordinates, and processing capability vectors for each processing workstation. The workstation identifier is a unique identification code for each workstation, generated using a unified coding rule across the workshop to avoid duplicate or confusing identifiers. All subsequent association, matching, and scheduling operations involving workstations will use the workstation identifier as the sole identification basis. The workstation location coordinates are marked using a unified Cartesian coordinate system or a three-dimensional spatial coordinate system established by the workshop, accurately recording the physical location of each workstation within the workshop. The processing capability vector details the types of processing procedures that the workstation can complete, the range of workpiece materials it can process, size limitations, upper limit of processing accuracy, maximum quantity processed in a single operation, equipment operating power, and other core capability parameters, directly determining the types of workpieces that the workstation can adapt to.

[0065] The equipment layer includes equipment identifiers for each processing device, a list of managed workstation identifiers, and equipment movement speed parameters. The equipment identifier is a unique code for each processing device, independent of the workstation identifiers. One processing device can correspond to one or more processing workstations, enabling multi-workstation collaborative operation. The list of managed workstation identifiers clearly lists all workstation codes that the device can drive and control, clarifying the affiliation between devices and workstations, avoiding confusion regarding workstation ownership during equipment scheduling, and facilitating analysis of equipment load and optimization of equipment utilization. The equipment movement speed parameters record the movement rate of the processing device within the workshop, including no-load movement speed, movement speed with workpiece, and turning response speed. This is a crucial reference for calculating workpiece handling time, evaluating path planning efficiency, and optimizing equipment scheduling sequence, ensuring that the planned processing path matches the actual operating capacity of the equipment and avoiding production delays caused by mismatches between path planning and equipment operating speed.

[0066] The path layer contains a set of reachable paths between any two workstations. Each reachable path records the path length and path travel time parameters. The path length is the actual physical distance between the two workstations, including the length data of different path forms such as straight-line distance, detour distance, and obstacle avoidance distance. The path travel time parameter comprehensively considers actual production factors such as equipment movement speed, workshop aisle congestion, queuing probability in front of workstations, and workpiece loading and unloading time to estimate the time required for the equipment to move from one workstation to another along the path. It is not simply calculated from distance and speed, but is more in line with the actual production scenario in the workshop.

[0067] Meanwhile, the autonomous planning agent is also equipped with an update interface, which can receive the actual path travel time reported by the processing equipment in real time. When there is a deviation between the actual travel time and the parameters stored in the path layer, the agent automatically corrects the path travel time parameters in the path layer online, ensuring that subsequent path planning, time estimation, benefit evaluation and other operations are always based on the latest and most accurate data, and avoiding the planning scheme from deviating from the actual production situation due to data lag.

[0068] Furthermore, the requirements information for the workpieces to be processed includes key information such as workpiece identifier, workpiece type, workpiece material, workpiece size, list of required processing steps, target processing accuracy level, minimum processing time, latest delivery time, and priority level. Among these, the target processing accuracy level directly determines the range of workstations that the workpiece can share and is a core constraint in the clustering process. The workpiece identifier serves as a unique code for the workpiece and is used for subsequent workpiece tracking, intra-cluster workpiece management, path matching, and other operations. The workpiece type, material, and size are used to select suitable processing workstations to ensure that the workstations within the clustered workpieces can meet the workpiece processing requirements.

[0069] Based on the global processing map and the demand information of the workpieces to be processed, multi-objective clustering is performed on the workpieces to be processed to obtain multiple clusters to be processed, including:

[0070] S1.1: Based on the global processing map, obtain the initial processing station sequence required for each workpiece to be processed; the initial processing station sequence includes the processing equipment station identifiers that the workpiece passes through in sequence;

[0071] Furthermore, the specific process of obtaining the initial machining station sequence required for each workpiece to be processed includes: the autonomous planning agent retrieves the pre-stored standard machining process document for the workpiece based on the workpiece identifier. This process document clearly marks all the processes required for workpiece processing and the corresponding machining station types for each process. Subsequently, the agent traverses the station layer of the global machining map, filters out the machining stations that meet the requirements of each process station type, and arranges the corresponding station identifiers in sequence according to the process sequence specified in the process document, ultimately generating the initial machining station sequence unique to the workpiece to be processed. For example, if the machining process of any shaft-type workpiece is turning, drilling, grinding, and polishing, the agent filters out the identifiers of turning, drilling, grinding, and polishing stations from the global machining map and arranges them in the process sequence to form the initial machining station sequence for the workpiece. Each workpiece to be processed corresponds to an independent initial machining station sequence. The sequence length varies depending on the number of workpiece processing processes. Each station identifier in the sequence comes from the global machining map, ensuring that the station actually exists and has the corresponding processing capability.

[0072] S1.2: Based on the requirements of the workpieces to be processed, obtain the target machining accuracy level for each workpiece;

[0073] Furthermore, the target machining accuracy level is a core indicator for measuring the quality of workpiece machining. It directly specifies the accuracy standards that each machining process of the workpiece must achieve. Different workpieces have significantly different target machining accuracy levels depending on their usage scenarios and performance requirements. For example, the target machining accuracy level of precision instrument parts is much higher than that of ordinary mechanical parts. The target machining accuracy level is usually divided using industry-standard accuracy grading. Different levels correspond to different allowable ranges of machining errors, surface roughness requirements, dimensional tolerance requirements, etc. The core function of this level is to constrain the range of shareable workstations during the clustering process. That is, only workstations with a machining accuracy upper limit not lower than the target machining accuracy level of the workpiece in the machining capability vector can be included in the candidate range of shared workstations in the cluster to be processed of that workpiece. This avoids the workpiece machining quality being unqualified due to insufficient machining accuracy of the workstation from the source, and also narrows the screening range of shared workstations, improving the efficiency and rationality of clustering.

[0074] Furthermore, the specific process for obtaining the target machining accuracy level is as follows: the autonomous planning agent retrieves the corresponding requirement information document based on the workpiece identifier of the workpiece to be processed, and extracts the pre-set target machining accuracy level parameters from the requirement information document. The target machining accuracy level parameters are determined by the production technology department based on the workpiece design drawings, usage requirements, industry standards, etc., and have uniqueness and authority.

[0075] S1.3: A multi-objective clustering algorithm is adopted, with minimizing the total number of shared workstations in all clusters to be processed as the first optimization objective, and the target processing accuracy level corresponding to each workpiece as the constraint condition. Clustering operation is performed on all workpieces to be processed to obtain multiple clusters to be processed.

[0076] It should be explained that a shared workstation refers to a processing workstation that is used by multiple workpieces within a cluster to be processed. The fewer the number of shared workstations, the more workpieces a single shared workstation can serve, and the higher the utilization rate of workstation resources.

[0077] Furthermore, the specific process of clustering operations includes:

[0078] (1) The autonomous planning agent forms an initial sample set of all workpieces to be processed, and defines each workpiece to be processed in the initial sample set as an independent initial cluster, thus completing the initial cluster division;

[0079] (2) The similarity calculation rules are preset. The similarity calculation rules are composed of five dimensions: the overlap of the initial processing station sequence, the matching degree of the workpiece type, the matching degree of the workpiece material, the matching degree of the target processing accuracy level, and the matching degree of the station space distance. The weight parameters of each dimension are set as follows: the weight of the overlap of the initial processing station sequence is 0.35, the weight of the matching degree of the workpiece type is 0.20, the weight of the matching degree of the workpiece material is 0.15, the weight of the matching degree of the target processing accuracy level is 0.20, and the weight of the matching degree of the station space distance is 0.10. The total weight is 1. The autonomous planning agent calculates the similarity between any two initial clusters in the initial sample set according to the similarity calculation rules, and obtains the similarity value corresponding to each pair of initial clusters. The similarity calculation formula is the prior art in this field and is not an inventive solution of this application. It will not be elaborated here.

[0080] (3) The autonomous planning agent sorts all initial cluster pairs in descending order of similarity value to generate a cluster pair sorting sequence. Then, starting from the first position of the cluster pair sorting sequence, it selects the pair of initial clusters with the highest similarity value as the cluster pairs to be merged.

[0081] (4) Perform a merging operation on the clusters to be merged to generate a new cluster. After the merging is completed, the autonomous planning agent counts the number of processing stations shared by all workpieces to be processed in the new cluster, obtains the number of shared stations in the new cluster, and verifies the target processing accuracy level corresponding to all workpieces to be processed in the new cluster. Confirm that the upper limit of the processing accuracy of each shared station in the new cluster is not lower than the target processing accuracy level of all workpieces to be processed in the new cluster, and completes the accuracy constraint verification.

[0082] (5) If the accuracy constraint verification result is qualified, the autonomous planning agent replaces the two initial clusters in the cluster pair to be merged with the new cluster, updates the composition of the initial sample set, and recalculates the similarity value between the new cluster and all other unmerged initial clusters in the initial sample set, and updates the similarity of all cluster pairs.

[0083] (6) If the accuracy constraint verification result is unqualified, the autonomous planning agent abandons the current merging operation, skips the current cluster pair to be merged, and selects the next cluster pair with the highest similarity value from the cluster pair sorting sequence as the new cluster pair to be merged.

[0084] (7) Perform iterative execution until the merging operation can no longer further reduce the total number of shared workstations in all clusters to be processed, or the preset clustering termination condition is met. The clustering termination condition includes two items: the first item is that the number of clusters to be processed reaches the preset first similarity threshold, which is set to 8; the second item is that the similarity value between any two clusters to be processed is lower than the preset second similarity threshold, which is set to 0.2. When either termination condition is met, the clustering operation ends, and the autonomous planning agent outputs the final multiple clusters to be processed.

[0085] S1.4: For each cluster to be processed, extract the processing stations common to all workpieces in the cluster, and use each extracted processing station as a shared station for the cluster. Finally, output the cluster identifier of each cluster to be processed, the list of workpiece identifiers contained in the cluster, and the list of shared stations corresponding to the cluster.

[0086] Furthermore, the process of extracting shared workstations includes:

[0087] (1) The autonomous planning agent traverses each cluster to be processed and retrieves the initial processing station sequence corresponding to all workpieces to be processed in the current cluster;

[0088] (2) Summarize the initial processing station sequences corresponding to all workpieces to be processed in the current processing cluster to generate an initial processing station sequence set;

[0089] (3) Compare and filter the initial processing station sequence set, extract the station identifiers that appear in each initial processing station sequence within the initial processing station sequence set, and generate a common station identifier set.

[0090] (4) Retrieve the upper limit parameters of the machining accuracy corresponding to each common workstation identifier in the global machining map. The upper limit parameters of the machining accuracy are pre-stored by the workstation layer of the global machining map. The unit is mm. The accuracy level is divided into five levels: the upper limit of the first level is 0.01 mm, the upper limit of the second level is 0.05 mm, the upper limit of the third level is 0.10 mm, the upper limit of the fourth level is 0.50 mm, and the upper limit of the fifth level is 1.00 mm.

[0091] (5) Retrieve the target machining accuracy level corresponding to each workpiece in the current processing cluster, which also corresponds to the first to fifth level accuracy standards;

[0092] (6) The autonomous planning agent compares and verifies the upper limit of the processing accuracy corresponding to each workstation identifier in the shared workstation identifier set with the target processing accuracy level corresponding to all workpieces to be processed in the current processing cluster, and removes workstation identifiers whose upper limit of processing accuracy is lower than the target processing accuracy level of any workpiece to be processed, and generates a shared workstation identifier set.

[0093] (7) Define the processing station corresponding to each station identifier in the shared station identifier set as the shared station of the current processing cluster, and complete the extraction of the shared station of a single processing cluster;

[0094] (8) Repeat steps (1) to (7) until all the shared workstations of the clusters to be processed are extracted, and output the list of shared workstations corresponding to each cluster to be processed.

[0095] S2: Using all the clusters to be processed as vertices, establish process association relationships between different clusters to be processed that have shared workstations through the corresponding shared workstations, construct a workstation association topology graph, calculate the initial value upper bound of each process association relationship in the workstation association topology graph, and output a candidate workstation set after filtering.

[0096] It should be explained that the workstation association topology graph is an abstract graph with the clusters to be processed as vertices and the shared workstation association relationships as edges. It can intuitively and clearly show the process association network formed between different clusters to be processed through shared workstations.

[0097] The process of establishing process association relationships between different clusters of workers to be processed, using all clusters to be processed as vertices and sharing workstations, and constructing a workstation association topology graph, includes:

[0098] S2.1: Treat the cluster identifier of each cluster to be processed as a vertex;

[0099] S2.2: Traverse any two different vertices that are to be processed, and for each pair of to be processed, check whether there is the same shared workstation identifier in the shared workstation list of each pair of to be processed.

[0100] Furthermore, it checks whether the clusters to be processed have the same shared workstation identifier in their respective shared workstation lists, including:

[0101] (1) The autonomous planning agent traverses all generated vertices, arbitrarily selects two different vertices, and determines the current pair of vertices to be detected;

[0102] (2) Define the two clusters to be processed corresponding to the current vertex pair to be detected as the current cluster pair to be processed;

[0103] (3) The autonomous planning agent retrieves the shared workstation list corresponding to the first cluster to be processed in the current cluster pair, and records it as the first shared workstation list; retrieves the shared workstation list corresponding to the second cluster to be processed in the current cluster pair, and records it as the second shared workstation list;

[0104] (4) The autonomous planning agent compares the identifiers of the first shared workstation list and the second shared workstation list. The comparison rule is set to complete matching of the identifier strings. The matching process iterates through each workstation identifier in the first shared workstation list and compares the strings with each workstation identifier in the second shared workstation list one by one.

[0105] (5) Record the comparison results. If there is a completely matching workstation identifier, it is determined that the current cluster pair to be processed has the same shared workstation identifier; if there is no completely matching workstation identifier, it is determined that the current cluster pair to be processed does not have the same shared workstation identifier.

[0106] (6) Repeat (1) to (5) until all different vertex corresponding to the cluster pairs to be processed are detected, and output the detection results of all cluster pairs to be processed.

[0107] S2.3: If there is a shared workstation identifier, the shared workstation identifier shall be used as the process association between the clusters to be processed, and an edge shall be used to connect the two vertices corresponding to the clusters to be processed.

[0108] Furthermore, the specific steps in S2.3 include:

[0109] (1) The autonomous planning agent receives the detection results of the cluster pairs to be processed from the detection results of the cluster pairs to be processed and selects the cluster pairs to be processed that have the same shared workstation identifier. The selected cluster pairs to be processed are defined as valid associated cluster pairs.

[0110] (2) For each group of valid associated clusters, extract the common shared workstation identifiers between the two groups of clusters to be processed, and define the extracted shared workstation identifiers as process association identifiers. The process association identifiers are used to uniquely identify the process association relationship between the two groups of clusters to be processed.

[0111] (3) The autonomous planning agent retrieves the vertex set of the workstation association topology. Each vertex in the vertex set corresponds to a cluster to be processed. The vertex corresponding to the first cluster to be processed in the effective association cluster pair is defined as the starting vertex, and the vertex corresponding to the second cluster to be processed in the effective association cluster pair is defined as the ending vertex.

[0112] (4) The autonomous planning agent performs edge construction operation. The edge construction operation adopts a preset topology connection algorithm. The parameters of the topology connection algorithm are set as follows: the connection type parameter is undirected connection, the connection weight parameter is initially set to 1, and the connection state parameter is set to valid. It is used to establish a connection relationship between the starting vertex and the ending vertex.

[0113] (5) Assign the process association identifier to the currently constructed connection relationship so that the current connection relationship and the process association identifier form a one-to-one binding relationship. After the binding is completed, the current connection relationship becomes the topological edge representing the process association relationship between the two groups of clusters to be processed.

[0114] (6) The autonomous planning agent adds the completed topological edges to the edge set of the workstation association topological graph. The edge set is used to store all topological edges that represent the process association relationship between the clusters to be processed, and completes the construction of topological edges for a single effective association cluster pair.

[0115] (7) Repeat (1)-(6) until the topological edge construction of all valid associated cluster pairs is completed, and output the updated workstation associated topological graph. The updated workstation associated topological graph contains all vertices and all topological edges.

[0116] S2.4: Combine all generated vertices and edges to construct an undirected workstation association topology graph; each edge in the workstation association topology graph records the identifier of the shared workstation it is associated with, as well as the cluster identifiers of the two clusters to be processed connected by the edge. The process of constructing the graph based on nodes and edges is existing technology in this field and is not an inventive solution of this application, so it will not be described in detail here.

[0117] It should be noted that an undirected topological graph means that the edges connecting two vertices have no directionality. It only indicates that there is a bidirectional process association between two clusters to be processed, and there is no unidirectional dependency. This is consistent with the actual characteristics of equal association and bidirectional collaboration between clusters to be processed in the workshop.

[0118] The initial upper bound of the value of each process association in the workstation association topology graph is calculated, and a candidate workstation set is output after filtering, including:

[0119] S2.5: Obtain the cluster identifier of the first cluster to be processed and the cluster identifier of the second cluster to be processed associated with each edge in the workstation association topology graph;

[0120] It should be noted that each edge corresponds to a unique cluster to be processed. The agent traverses all edges in the topology graph and reads the two cluster identifiers recorded in the attribute information of each edge one by one. These identifiers are defined as the first cluster identifier and the second cluster identifier, respectively. They are only used as identifiers to distinguish between two related clusters and have no order or priority difference.

[0121] S2.6: Based on the number of remaining unprocessed workpieces in each current processing cluster, obtain the current number of remaining unprocessed workpieces in the first processing cluster as the first quantity, and obtain the current number of remaining unprocessed workpieces in the second processing cluster as the second quantity.

[0122] Furthermore, the process of obtaining the current number of remaining unprocessed workpieces is as follows: The autonomous planning agent maintains the workpiece status table of each cluster to be processed in real time, and records the processing status of each workpiece in the cluster, such as unprocessed, included in the path, and processed. The agent traverses the workpiece status table of each cluster to be processed and counts the number of workpieces with the status of unprocessed, which is the current number of remaining unprocessed workpieces in that cluster. When S2 is executed, all workpieces are not included in the collaborative processing path, so in the initial state, the number of remaining unprocessed workpieces in each cluster to be processed is equal to the total number of workpieces contained in that cluster. As the iteration process progresses, workpieces are gradually included in the path and processed, and the remaining number will dynamically decrease. The agent updates the status table in real time to ensure that the obtained remaining number is always the latest data.

[0123] S2.7: Take the minimum value between the first quantity and the second quantity, and multiply the minimum value by a preset unit processing efficiency coefficient to obtain the initial value upper bound of the edge; the unit processing efficiency coefficient is the efficiency value generated by the processing equipment for processing a single workpiece once at the shared workstation, which is pre-stored by the global processing map;

[0124] Furthermore, the specific steps in S2.7 include:

[0125] (1) The autonomous planning agent receives the data of the number of remaining workpieces in the cluster to be processed, retrieves the number of remaining unprocessed workpieces corresponding to the first cluster to be processed associated with the current edge to be calculated, and defines the number as the first number, which is an integer greater than or equal to zero.

[0126] (2) Retrieve the number of remaining unprocessed workpieces corresponding to the second cluster to be processed associated with the edge to be calculated, and define this number as the second number, which is an integer greater than or equal to zero;

[0127] (3) Compare the first quantity with the second quantity, select the smaller value, and obtain the minimum value;

[0128] (4) Retrieve the unit processing efficiency coefficient corresponding to the shared workstation associated with the current edge in the global processing map. In this embodiment, the unit processing efficiency coefficient is set to 5, which is pre-stored in the global processing map and cannot be changed arbitrarily.

[0129] (5) Multiply the minimum value by the unit processing efficiency coefficient to obtain the initial value upper bound of the current edge. The initial value upper bound is an integer greater than or equal to zero, which is used to characterize the potential benefit upper limit of the process association relationship corresponding to the current edge.

[0130] (6) Repeat (1)-(5) until the calculation of the upper bound of the initial value corresponding to all edges in the workstation association topology is completed, and output the upper bound data of the initial value corresponding to all edges.

[0131] S2.8: Create an empty set of candidate workstations, filter out edges whose initial value upper bound is greater than zero, and add the shared workstation identifier corresponding to each filtered edge to the set of candidate workstations.

[0132] Furthermore, the screening process includes: the autonomous planning agent traverses all edges in the workstation association topology graph, reads the initial value upper bound value corresponding to each edge one by one, and determines whether the value is greater than zero. Since the unit processing efficiency coefficient is positive and the minimum number of remaining unprocessed workpieces is a non-negative integer, the initial value upper bound is a non-negative number. Only when the minimum value is greater than zero is the initial value upper bound greater than zero. That is, only shared workstations that are associated with two processing clusters that both have remaining unprocessed workpieces will be included in the candidate workstation set. After screening, the agent adds the shared workstation identifiers corresponding to the edges that meet the conditions to the pre-created empty candidate workstation set one by one, ensuring that there are no duplicate workstation identifiers or invalid workstation identifiers in the set, and finally outputs the candidate workstation set.

[0133] S3: Iterative processing of candidate workstation set: Each time, a shared workstation is selected from the candidate workstation set. Based on all the processing clusters associated with the shared workstation, a collaborative processing path is solved. The workpieces in the collaborative processing path are synchronously deleted from the corresponding processing clusters. The workstation association topology and the candidate workstation set are updated until the candidate workstation set is empty.

[0134] The process of solving the collaborative processing path includes:

[0135] S3.1: Select the shared workstation with the largest upper bound of initial value from the current set of candidate workstations, and use this shared workstation as the currently selected shared workstation;

[0136] Furthermore, the initial value upper bound selection process is as follows: The autonomous planning agent reads the initial value upper bound values ​​of all shared workstations in the current candidate workstation set, compares them one by one, and selects the shared workstation with the largest value; if there are multiple shared workstations with the same value upper bound and all of them being the maximum value, then further compares the auxiliary indicators such as the number of clusters to be processed associated with them, the total number of remaining workpieces in the cluster, and the concentration of workstation spatial distribution, and prioritizes the shared workstation with more associated clusters, a larger total number of remaining workpieces, and a more concentrated distribution to ensure the rationality of the selection. After selection, the shared workstation is the core processing workstation of the current round, that is, the currently selected shared workstation.

[0137] S3.2: Based on the constructed workstation association topology, obtain the cluster identifiers of all the clusters to be processed associated with the currently selected shared workstation, and use the set of cluster identifiers as the current associated cluster set;

[0138] Furthermore, the autonomous planning agent traverses all edges in the workstation association topology graph, filtering out edges whose shared workstation identifiers recorded in the edge attribute information are completely consistent with the currently selected shared workstation identifier. The clusters to be processed corresponding to the two vertices connected by these edges are the clusters to be processed associated with the currently selected shared workstation. The agent summarizes the cluster identifiers of these clusters to be processed, removes duplicate identifiers, and forms the current associated cluster set, ensuring that all clusters to be processed in the set have a direct process association with the currently selected shared workstation, with no irrelevant clusters mixed in, thus narrowing the scope of workpiece screening and improving the targeting of path planning.

[0139] S3.3: Obtain the remaining workpieces in each cluster to be processed in the current associated cluster set that have not yet been included by any generated collaborative processing path, and construct all the obtained remaining workpieces into a queue of workpieces to be processed in the current batch.

[0140] Furthermore, the process of acquiring the remaining workpieces and constructing a queue is as follows: The autonomous planning agent traverses each cluster to be processed in the current associated cluster set, retrieves the workpiece status table of that cluster, filters out workpieces in the unprocessed state, that is, those not yet included in any generated collaborative processing path, and extracts key information such as workpiece identifier, initial processing station sequence, target processing accuracy level, workpiece size, and material of these workpieces; all the filtered workpiece information is arranged in a preset order, such as workpiece priority, workpiece size, and initial processing station sequence matching degree, to form the queue of workpieces to be processed in the current batch. Each workpiece in the queue of workpieces to be processed has complete and unique information, and there are no duplicate workpieces.

[0141] S3.4: With minimizing the total processing path length of the current batch of workpieces to be processed as the optimization objective, solve for the collaborative processing path of the current batch of workpieces to be processed on the currently selected shared workstation, and finally output the identifier of the currently selected shared workstation and its corresponding collaborative processing path.

[0142] Solving the collaborative processing path of the current batch of workpieces to be processed at the currently selected shared workstation includes:

[0143] S3.4.1: Set the currently selected shared workstation as the pivot workstation and initialize the search state, which includes: the current workstation position, the set of completed workpieces, and the current cumulative path length; the initial value of the current workstation position is set to the pivot workstation, the initial value of the set of completed workpieces is set to an empty set, and the initial value of the current cumulative path length is set to zero.

[0144] It should be noted that the pivot station is the core station of the current collaborative machining path. The machining paths of all workpieces revolve around the pivot station. Workpieces must first complete the processes before the pivot station, then complete the core processes at the pivot station, and finally complete the processes after the pivot station. This ensures that the core processes are concentrated in high-value shared stations, maximizing station utilization. The search state is used to record key dynamic information during the path solving process, enabling gradual exploration and optimal selection of paths: the current station position records the latest station position of the workpiece during machining, initially the pivot station; the set of completed workpieces records the identifiers of the completed workpieces in the current path, initially empty, and gradually added; the current cumulative path length records the total path distance from the initial position to the current position, initially zero, and gradually accumulated as workpieces are machined.

[0145] S3.4.2: Take the queue of workpieces to be processed in the current batch as the set of workpieces to be processed. If the set of workpieces to be processed is empty, then take the current cumulative path length as the global optimal path length.

[0146] It should be explained that the set of workpieces to be processed is all the workpieces that need to be processed in the current batch. If the set of workpieces to be processed is empty, it means that there are no workpieces to be processed in the current batch. There is no need to plan a path. The initial cumulative path length of zero is taken as the optimal length. If the path is empty, the update step is directly entered.

[0147] S3.4.3: If the set of workpieces to be processed is not empty, then select a candidate workpiece that has not been marked as completed from the set of workpieces to be processed, and obtain a complete sequence of processing stations consisting of all the necessary stations that start from the current station position and pass through the candidate workpiece in sequence, according to the global processing map; the necessary stations include the pivot station and the other necessary stations of the candidate workpiece recorded in the global processing map;

[0148] The step of obtaining the complete machining station sequence consisting of all necessary stations that must pass through the candidate workpiece starting from the current station position includes:

[0149] S3.4.3.1: The current workstation position is used as the starting workstation identifier, where the starting workstation is the current position of the workpiece. Path planning starts from this workstation to ensure path continuity and avoid repeatedly planning completed paths.

[0150] S3.4.3.2: Obtain the initial processing station sequence of the candidate workpiece from the global processing map; the initial processing station sequence is a list of station identifiers that the candidate workpiece needs to pass through sequentially to complete processing, and includes pivot stations;

[0151] S3.4.3.3: Extract all station identifiers before the pivot station from the initial machining station sequence as the preceding station set, and extract all station identifiers after the pivot station as the following station set. The preceding station set contains all stations corresponding to all processes that need to be completed before the workpiece reaches the pivot station, and the following station set contains all stations corresponding to all processes that need to be completed after the workpiece leaves the pivot station. This clarifies the machining process of the workpiece before and after the pivot station and avoids process confusion.

[0152] S3.4.3.4: According to the original order in the initial processing station sequence, sort the station identifiers in the preceding station set and the following station set respectively to generate the preceding station execution sequence and the following station execution sequence;

[0153] S3.4.3.5: Based on the shortest path distance between any two workstations recorded in the global processing map, calculate the shortest path from the starting workstation identifier to the first workstation in the preceding workstation execution sequence, the shortest path between two adjacent workstations in the preceding workstation execution sequence, the shortest path from the last workstation in the preceding workstation execution sequence to the pivot workstation, the shortest path from the pivot workstation to the first workstation in the following workstation execution sequence, and the shortest path between two adjacent workstations in the following workstation execution sequence in sequence.

[0154] Furthermore, the process of calculating the shortest path from the starting workstation identifier to the first workstation in the preceding workstation execution sequence includes:

[0155] (1) The autonomous planning agent first confirms the starting station identifier required for this calculation. This starting station identifier is output by S3.4.3.1 and represents the current station position of the workpiece.

[0156] (2) Determine the target station for this calculation. The target station is the first station in the previous station execution sequence. The previous station execution sequence is generated by the previous steps in the process order and remains fixed.

[0157] (3) The autonomous planning agent retrieves the basic data stored in the global processing map path layer. This basic data includes the connectivity of all workstations in the workshop and the direct path distance between adjacent workstations. The unit of the path distance is uniformly set to meters, and all values ​​are real numbers greater than zero. The data is collected and stored in advance by the system.

[0158] (4) Input the starting workstation identifier and the target workstation into the Dijkstra algorithm. The Dijkstra algorithm starts from the starting workstation and traverses all adjacent and connected workstations layer by layer. At the same time, it accumulates the direct path distance between each adjacent workstation and updates the shortest distance record corresponding to each traversed workstation. It retains only the smaller distance value and updates the corresponding path information until the Dijkstra algorithm reaches the target workstation and triggers the termination condition. The Dijkstra algorithm is the prior art in this field and is not the inventive solution of this application. It will not be described in detail here.

[0159] (5) Extract the shortest path distance value from the starting station identifier to the first station in the execution sequence of the preceding station from the final output of Dijkstra's algorithm.

[0160] Furthermore, the calculation process of the shortest path between two adjacent workstations in the preceding workstation execution sequence includes: the autonomous planning agent first obtains the preceding workstation execution sequence; selects two adjacent workstations sequentially from the preceding workstation execution sequence, defines the workstation that comes first as the current starting workstation, and defines the workstation that follows immediately as the current target workstation, forming a pair of workstations to be calculated; inputs the current starting workstation and the current target workstation into the Dijkstra algorithm, starting from the current starting workstation, traversing all adjacent and connected workstations layer by layer, while accumulating the direct path distance between each adjacent workstation segment, continuously updating the shortest distance record corresponding to each traversed workstation, until the current target workstation is reached and the termination condition is triggered; extracts the shortest path distance value corresponding to the current starting workstation to the current target workstation from the final output of the Dijkstra algorithm.

[0161] Furthermore, the calculation process of the shortest path from the last workstation in the preceding workstation execution sequence to the pivot workstation includes: the autonomous planning agent confirms the starting workstation for this path calculation, which is the last workstation in the preceding workstation execution sequence; confirms the ending workstation for this path calculation, which is the pivot workstation, set as the currently selected shared workstation, and is the core workstation for workpiece processing; inputs the starting workstation and the ending workstation into the Dijkstra algorithm, which starts from the starting workstation and traverses all adjacent and connected workstations layer by layer, while accumulating the direct path distance between each adjacent workstation segment, and continuously updating the shortest distance record corresponding to each traversed workstation until the algorithm reaches the ending workstation and triggers the termination condition; the autonomous planning agent extracts the shortest path distance value from the last workstation in the preceding workstation execution sequence to the pivot workstation from the final output of the shortest path calculation algorithm.

[0162] Furthermore, the calculation process of the shortest path from the pivot station to the first station in the subsequent station execution sequence includes: the autonomous planning agent first confirms the starting station of this path calculation, which is the pivot station; confirms the ending station of this path calculation, which is the first station in the subsequent station execution sequence, and the subsequent station execution sequence is generated according to the original sequence of the initial processing station sequence of the workpiece; the determined starting station and ending station are input into the Dijkstra algorithm, which starts from the starting station and traverses all adjacent and connected stations layer by layer, while accumulating the direct path distance between each adjacent station segment, and continuously updating the shortest distance record corresponding to each traversed station until the algorithm reaches the ending station and triggers the termination condition; from the final output of the shortest path calculation algorithm, the shortest path distance value corresponding to the first station in the subsequent station execution sequence is extracted.

[0163] Further, the calculation process of the shortest path between two adjacent workstations in the subsequent workstation execution sequence includes: the autonomous planning agent first obtains the subsequent workstation execution sequence, which is generated according to the original sequence of the initial processing workstations of the workpiece, and is used to characterize the order of workstations that need to be processed after the pivot workstation; from the subsequent workstation execution sequence, two adjacent workstations are selected sequentially, the workstation that comes first is defined as the current calculation starting point workstation, and the workstation that follows immediately is defined as the current calculation ending point workstation, forming a pair of adjacent workstations to be calculated; the current calculation starting point workstation and the current calculation ending point workstation are input into the Dijkstra algorithm, and the Dijkstra algorithm... The algorithm starts from the current starting workstation and traverses all adjacent and connected workstations layer by layer. It accumulates the direct path distance between each adjacent workstation segment, continuously updating the shortest distance record for each traversed workstation. This process gradually narrows the search range and approaches the current endpoint workstation. It continuously compares the accumulated distance of the new path with the recorded shortest distance, retaining only the smaller distance values ​​and updating the corresponding path information to avoid invalid paths interfering with the calculation results. This continues until the algorithm reaches the current endpoint workstation and triggers the termination condition, ending the path search for the current pair of adjacent workstations. The shortest path distance value from the current starting workstation to the current endpoint workstation is extracted from the final output of the shortest path calculation algorithm.

[0164] S3.4.3.6: Concatenate all shortest paths in the execution order to generate a complete sequence of processing stations.

[0165] S3.4.4: Calculate the path increment length required to execute a complete machining station sequence; the path increment length is the sum of the distances between adjacent stations in the complete machining station sequence;

[0166] Furthermore, the specific steps of S3.4.4 include:

[0167] (1) The autonomous planning agent obtains the complete processing station sequence. The stations in the complete processing station sequence are arranged according to the order of workpiece processing. Each station corresponds to a unique station identifier. The sequence length is greater than or equal to 2.

[0168] (2) Initialize the path increment length calculation variable. The initial value of this variable is set to zero, and the data type of the variable is set to real number. It is used to accumulate the total distance between all adjacent stations in the complete processing station sequence.

[0169] (3) Starting from the beginning of the complete processing station sequence, select two adjacent stations in sequence, define the station that is in front as the current front station, and define the station that follows it as the current back station, to form a pair of adjacent stations with a distance to be calculated.

[0170] (4) The autonomous planning agent retrieves the shortest path distance value from the current front workstation to the current back workstation, adds the shortest path distance value corresponding to the current adjacent workstation pair to the existing value of the current path increment length calculation variable, and updates the value of the path increment length calculation variable.

[0171] (5) Repeat (3)-(4) to iterate through all adjacent workstation pairs in the complete processing workstation sequence until the cumulative calculation of the distance between all adjacent workstations is completed. At this time, the value of the path increment length calculation variable is the path increment length corresponding to the complete processing workstation sequence.

[0172] S3.4.5: If the current cumulative path length plus the path increment length is less than the currently recorded global optimal path length, then update the current workstation position to the last workstation of the complete processing workstation sequence, add the candidate workpiece to the set of completed processing workpieces, increase the current cumulative path length by the path increment length, mark the candidate workpiece as completed, and continue processing the next incomplete workpiece; otherwise, directly select another candidate workpiece.

[0173] S3.4.6: When all workpieces in the set of workpieces to be processed are marked as completed, the current cumulative path length is taken as the global optimal path length, and the corresponding complete processing station sequence is recorded as the collaborative processing path on the currently selected shared station.

[0174] Synchronously delete the workpieces in the collaborative processing path from the corresponding clusters to be processed, update the station association topology and the candidate station set until the candidate station set is empty, including:

[0175] (1) Obtain the collaborative processing path corresponding to the currently selected shared workstation, and extract all workpiece identifiers contained in the collaborative processing path as a list of processed workpiece identifiers;

[0176] (2) Obtain the constructed current associated cluster set, and for each cluster to be processed in the current associated cluster set, delete the workpiece from the list of processed workpiece identifiers from the remaining unprocessed workpieces of the cluster to be processed;

[0177] (3) For each vertex of the cluster to be processed in the constructed workstation association topology graph, if the number of unprocessed workpieces remaining after the cluster to be processed is zero, then delete the vertex corresponding to the cluster to be processed and all edges connected to the vertex from the workstation association topology graph.

[0178] (4) For each remaining edge in the updated workstation association topology graph, recalculate the current value upper bound of the edge according to the method for calculating the initial value upper bound;

[0179] (5) Clear the output candidate workstation set, add the shared workstation identifiers corresponding to the edges whose current value upper bound is greater than zero to the cleared candidate workstation set, and obtain the updated candidate workstation set;

[0180] (6) Determine whether the updated candidate workstation set is empty. If it is empty, end the iteration.

[0181] S4: Summarize the collaborative processing paths generated in each iteration, supplement the remaining unprocessed workpieces in the cluster to be processed after the iteration as independent paths, and integrate and output the final collaborative processing path.

[0182] The collaborative processing path generated by the iteration takes the remaining unprocessed workpieces within the cluster to be processed after the iteration as an independent path to supplement the final collaborative processing path, including:

[0183] S4.1: Obtain all collaborative processing paths output in each iteration, arrange them in the order of their generation, and obtain the planned path sequence;

[0184] S4.2: Obtain the remaining unprocessed workpieces in each cluster to be processed, and summarize all the remaining unprocessed workpieces into a set of remaining workpieces;

[0185] S4.3: For each workpiece in the remaining workpiece set, obtain the initial processing station sequence of the workpiece from the global processing map, and use the initial processing station sequence as the independent processing path of the workpiece;

[0186] S4.4: Append all generated independent processing paths to the end of the planned path sequence to generate a complete processing path sequence, and output the complete processing path sequence as the final collaborative processing path.

[0187] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments under the guidance of the present invention without departing from the spirit and scope of the present invention. All of these variations are within the protection scope of the present invention.

Claims

1. A multi-process collaborative machining method for workpieces, applied to a machining planning system, characterized in that, The machining planning system includes an autonomous planning agent, multiple workpieces, and machining equipment. The autonomous planning agent is configured with a global machining map. The method includes: Based on the global processing map and the demand information of the workpieces to be processed, multi-objective clustering is performed on the workpieces to be processed to obtain multiple clusters to be processed; each cluster to be processed includes at least one shared workstation. Using all the clusters to be processed as vertices, establish process association relationships between different clusters to be processed that share a workstation through the corresponding shared workstation, construct a workstation association topology graph, calculate the initial upper bound of the value of each process association relationship in the workstation association topology graph, and output a candidate workstation set after filtering. Iterative processing of candidate workstation set: Each time, a shared workstation is selected from the candidate workstation set. Based on all the processing clusters associated with the shared workstation, a collaborative processing path is solved. The workpieces in the collaborative processing path are synchronously deleted from the corresponding processing clusters. The workstation association topology and the candidate workstation set are updated until the candidate workstation set is empty. The collaborative processing paths generated in each iteration are summarized, and the remaining unprocessed workpieces in the cluster to be processed after each iteration are supplemented as independent paths. The final collaborative processing path is then integrated and output.

2. The multi-process collaborative machining method for a workpiece as described in claim 1, characterized in that, Based on the global processing map and the demand information of the workpieces to be processed, multi-objective clustering is performed on the workpieces to be processed to obtain multiple clusters to be processed, including: Based on the global machining map, the initial machining station sequence required for each workpiece to be processed is obtained; the initial machining station sequence includes the identification of the machining equipment station that the workpiece passes through in sequence. Based on the requirements of the workpieces to be processed, the target processing accuracy level of each workpiece is obtained; the target processing accuracy level is used to constrain the range of workstations that can be shared during the clustering process. A multi-objective clustering algorithm is adopted, with minimizing the total number of shared workstations in all clusters to be processed as the first optimization objective and the target processing accuracy level corresponding to each workpiece as the constraint condition. Clustering operation is performed on all workpieces to be processed to obtain multiple clusters to be processed. For each cluster to be processed, extract the processing stations common to all workpieces within the cluster, and use each extracted processing station as a shared station for the cluster. Finally, output the cluster identifier, the list of workpiece identifiers contained in the cluster, and the corresponding list of shared stations for each cluster to be processed.

3. The multi-process collaborative machining method for a workpiece as described in claim 1, characterized in that, The process of establishing process association relationships between different clusters of workers to be processed, using all clusters to be processed as vertices and sharing workstations, and constructing a workstation association topology graph, includes: Each cluster identifier of the cluster to be processed is treated as a vertex; Iterate through any two different vertices that are to be processed, and for each pair of clusters to be processed, check whether there is a common shared workstation identifier in the shared workstation list of each pair of clusters to be processed. If there are identical shared workstation identifiers, the shared workstation identifiers will be used as the process association between the clusters to be processed, and an edge will be used to connect the two vertices corresponding to the clusters to be processed. All generated vertices and edges are combined to construct an undirected workstation association topology graph; each edge in the workstation association topology graph records the identifier of the shared workstation it is associated with, as well as the cluster identifiers of the two clusters to be processed connected by the edge.

4. The multi-process collaborative machining method for a workpiece as described in claim 1, characterized in that, The initial upper bound of the value of each process association in the workstation association topology graph is calculated, and a candidate workstation set is output after filtering, including: Obtain the cluster identifier of the first cluster to be processed and the cluster identifier of the second cluster to be processed associated with each edge in the workstation association topology graph; Based on the number of remaining unprocessed workpieces in each current processing cluster, the current number of remaining unprocessed workpieces in the first processing cluster is obtained as the first quantity, and the current number of remaining unprocessed workpieces in the second processing cluster is obtained as the second quantity. Take the minimum value between the first quantity and the second quantity, and multiply the minimum value by a preset unit processing efficiency coefficient to obtain the initial upper bound of the edge value; the unit processing efficiency coefficient is the efficiency value generated by the processing equipment to process a single workpiece once on a shared workstation, which is pre-stored by the global processing map; Create an empty set of candidate workstations, filter out edges whose initial value upper bound is greater than zero, and add the shared workstation identifier corresponding to each filtered edge to the set of candidate workstations.

5. The multi-process collaborative machining method for a workpiece as described in claim 1, characterized in that, The process of solving the collaborative processing path includes: Select the shared workstation with the largest upper bound of initial value from the current set of candidate workstations, and set the shared workstation as the currently selected shared workstation; Based on the constructed workstation association topology, obtain the cluster identifiers of all the clusters to be processed associated with the currently selected shared workstation, and use the set of cluster identifiers as the current associated cluster set; Obtain the remaining workpieces in each cluster to be processed in the current associated cluster set that have not yet been included by any generated collaborative processing path, and construct all the obtained remaining workpieces into a queue of workpieces to be processed in the current batch. With the goal of minimizing the total processing path length of the current batch of workpieces to be processed, the collaborative processing path of the current batch of workpieces to be processed on the currently selected shared workstation is solved, and finally the identifier of the currently selected shared workstation and its corresponding collaborative processing path are output.

6. The multi-process collaborative machining method for a workpiece as described in claim 5, characterized in that, Solving the collaborative processing path of the current batch of workpieces to be processed at the currently selected shared workstation includes: The currently selected shared workstation is designated as the pivot workstation, and the search state is initialized. The search state includes: the current workstation position, the set of completed workpieces, and the current cumulative path length. The initial value of the current workstation position is set to the pivot workstation, the initial value of the set of completed workpieces is set to an empty set, and the initial value of the current cumulative path length is set to zero. Take the current batch of workpieces to be processed as the set of workpieces to be processed. If the set of workpieces to be processed is empty, then take the current cumulative path length as the global optimal path length. If the set of workpieces to be processed is not empty, then select a candidate workpiece that has not been marked as completed from the set of workpieces to be processed, and obtain a complete sequence of processing stations consisting of all necessary stations that start from the current station position and pass through the candidate workpiece in sequence, according to the global processing map; the necessary stations include the pivot station and the other necessary stations of the candidate workpiece recorded in the global processing map. Calculate the path increment length required to execute a complete processing station sequence; the path increment length is the sum of the distances between adjacent stations in the complete processing station sequence.

7. The multi-process collaborative machining method for a workpiece as described in claim 6, characterized in that, Solving the collaborative processing path of the current batch of workpieces to be processed on the currently selected shared workstation also includes: If the current cumulative path length plus the path increment length is less than the currently recorded global optimal path length, then the current workstation position is updated to the last workstation of the complete processing workstation sequence, the candidate workpiece is added to the set of completed processing workpieces, the current cumulative path length is increased by the path increment length, the candidate workpiece is marked as completed, and the next incomplete workpiece is processed; otherwise, another candidate workpiece is selected directly. When all workpieces in the set of workpieces to be processed are marked as completed, the current cumulative path length is taken as the global optimal path length, and the corresponding complete processing station sequence is recorded as the collaborative processing path on the currently selected shared station.

8. The multi-process collaborative machining method for a workpiece as described in claim 7, characterized in that, The process of obtaining the complete machining station sequence consisting of all necessary stations that must pass through the candidate workpiece starting from the current station position includes: Use the current workstation position as the starting workstation identifier; The initial processing station sequence of the candidate workpiece is obtained from the global processing map, and all station identifiers before the pivot station are extracted as the preceding station set, and all station identifiers after the pivot station are extracted as the following station set. According to the original order in the initial processing station sequence, the station identifiers in the preceding station set and the following station set are sorted to generate the preceding station execution sequence and the following station execution sequence. Based on the shortest path distance between any two workstations recorded in the global processing map, the shortest path from the starting workstation identifier to the first workstation in the preceding workstation execution sequence, the shortest path between two adjacent workstations in the preceding workstation execution sequence, the shortest path from the last workstation in the preceding workstation execution sequence to the pivot workstation, the shortest path from the pivot workstation to the first workstation in the following workstation execution sequence, and the shortest path between two adjacent workstations in the following workstation execution sequence are calculated sequentially. All shortest paths are concatenated end to end in the execution order to generate a complete sequence of processing stations.

9. The multi-process collaborative machining method for a workpiece as described in claim 1, characterized in that, The collaborative processing path generated by the iteration takes the remaining unprocessed workpieces within the cluster to be processed after the iteration as an independent path to supplement the final collaborative processing path, including: Obtain all collaborative processing paths output in each iteration, and arrange them in the order of their generation to obtain the planned path sequence; Obtain the remaining unprocessed workpieces in each cluster to be processed, and summarize all the remaining unprocessed workpieces into a set of remaining workpieces; For each workpiece in the remaining workpiece set, the initial processing station sequence of the workpiece is obtained from the global processing map, and the initial processing station sequence is used as the independent processing path of the workpiece. All generated independent processing paths are appended to the end of the planned path sequence to generate a complete processing path sequence, which is then output as the final collaborative processing path.

10. The multi-process collaborative machining method for a workpiece as described in claim 1, characterized in that, The global processing map configured within the autonomous planning intelligent body is stored using a hierarchical graph structure, which includes a workstation layer, an equipment layer, and a path layer. The workstation layer includes the workstation identifier, workstation position coordinates, and processing capacity vector for each processing workstation; The equipment layer includes the equipment identifier of each processing device, a list of workstation identifiers under its jurisdiction, and equipment movement speed parameters; The path layer contains a set of reachable paths between any two workstations, and each reachable path records the path length and path travel time parameters; The autonomous planning agent also receives the actual path travel time reported by the processing equipment through the update interface, and corrects the path travel time parameters in the path layer online.

Images