Manufacturing task processing method, device and equipment

Abstract

The invention provides a manufacturing task processing method, device and equipment, and belongs to the technical field of computer information processing. The method comprises the steps of obtaining a manufacturing service optimal scheme initial set of a manufacturing task and a preset low-level heuristic operator queue; determining a low-level heuristic operator performance ranking table and a first solution set according to the manufacturing service optimal scheme initial set and a preset low-level heuristic operator queue; determining a target low-level heuristic operator of the manufacturing service optimal scheme according to the first solution set and a low-level heuristic operator performance ranking table; determining a second solution set according to the first solution set and the target low-level heuristic operator; obtaining a fitness landform vector and a ranking weight; according to the fitness landform vector and the ranking weight, determining a target manufacturing service optimal scheme solution set; and according to the target manufacturing service optimal scheme solution set, determining a manufacturing task corresponding to the manufacturing service optimal scheme and scheduling the manufacturing task. The scheme has self-adaptive capability, and efficient and accurate selection of manufacturing services is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of computer information processing, and in particular to a method, device and equipment for processing a manufacturing task. Background Art

[0002] Intelligent manufacturing leverages technologies such as the Internet of Things (IoT), cyber-physical systems (CPS), big data, and artificial intelligence (AI). It implements intelligent activities such as analysis, reasoning, judgment, conception, and decision-making within the manufacturing process on a cloud platform. Because its implementation relies on cloud platform support, it is also known as intelligent cloud manufacturing. Enterprises virtualize physical manufacturing resources, such as manufacturing equipment, as well as manufacturing capabilities, software, and production plans, into services and upload them to the cloud platform. When users submit manufacturing tasks, the intelligent cloud manufacturing platform combines manufacturing services from different enterprises and regions according to specific rules to meet user needs. The manufacturing service combination process must consider multiple conflicting objectives, such as cost, service quality, delivery time, and service provider capabilities. The manufacturing service selection problem can be categorized into single-service and multi-service selection based on complexity. Due to the increasing complexity of manufacturing tasks, current research is focusing on multi-service selection. Multi-service selection refers to the combination of services encompassing multiple manufacturing subtasks. This involves first breaking down the manufacturing task into subtasks of varying levels and granularity, then automatically selecting services from a pool of candidate services that meet the requirements of each subtask. These services are then aggregated according to specific rules to form a combined service with inherent process logic that collaboratively completes the task.

[0003] Existing approaches to handling multi-task manufacturing tasks primarily include rule-based methods, heuristic algorithms, or machine learning techniques. However, these methods often have limitations when faced with complex manufacturing service environments. Rule-based methods rely too heavily on expert experience and struggle to adapt to dynamically changing environments; machine learning methods, on the other hand, require large amounts of training data and place high demands on data quality. Heuristic algorithms can demonstrate certain efficiency and performance advantages in solving specific problems. However, their performance is less stable when applied across problem domains. This is because heuristic algorithms are tailored to specific problems, and their performance degrades significantly if the problem environment changes. Furthermore, the performance of heuristic algorithms relies on manual parameter adjustments. Therefore, a method for handling manufacturing tasks across problem domains is needed. Summary of the Invention

[0004] The present invention provides a method, device and equipment for processing manufacturing tasks, which solves the problem that the current manufacturing service selection method is difficult to adapt to the dynamically changing environment.

[0005] In order to solve the above technical problems, the technical solutions of the present invention are as follows:

[0006] An embodiment of the present invention provides a method for processing a manufacturing task, comprising:

[0007] Obtaining an initial set of manufacturing service preferred solutions for the manufacturing task and a preset low-level heuristic operator queue for determining the manufacturing service preferred solutions;

[0008] Determining a low-level heuristic operator performance ranking table according to the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue, wherein the preset low-level heuristic operator queue includes a plurality of low-level heuristic operators;

[0009] Determining a first solution set of the manufacturing service preferred solutions according to the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue;

[0010] Determining a target low-level heuristic operator for a manufacturing service optimization solution based on the first solution set and the low-level heuristic operator performance ranking table;

[0011] determining a second solution set of manufacturing service optimization solutions based on the first solution set and a target low-level heuristic operator;

[0012] Obtaining a fitness landscape vector corresponding to the second solution set and a ranking weight of the target low-level heuristic operator;

[0013] Determining a target manufacturing service preferred solution set according to the fitness landscape vector corresponding to the second solution set and the ranking weight of the target low-level heuristic operator;

[0014] According to the target manufacturing service preferred solution set, manufacturing tasks corresponding to the manufacturing service preferred solutions are determined, and the manufacturing tasks are scheduled.

[0015] Optionally, determining a low-level heuristic operator performance ranking table based on the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue includes:

[0016] Processing the initial set of manufacturing service preferred solutions according to the preset low-level heuristic operator queue to determine a plurality of intermediate solution sets;

[0017] Determining fitness landscape vectors corresponding to the multiple intermediate solution sets according to the multiple intermediate solution sets;

[0018] A low-level heuristic operator performance ranking table is determined according to the fitness landscape vectors corresponding to the multiple intermediate solution sets.

[0019] Optionally, determining a first solution set of the manufacturing service preferred solutions according to the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue includes:

[0020] The initial set of manufacturing service preferred solutions is used as an input solution set, and is processed using a low-level heuristic operator queue to obtain a solution set containing multiple solution sets;

[0021] The solution sets are evaluated to obtain a first solution set, where the first solution set is a solution set with the highest evaluation value among the multiple solution sets.

[0022] Optionally, determining a target low-level heuristic operator for the manufacturing service optimization solution based on the first solution set and the low-level heuristic operator performance ranking table includes:

[0023] According to the first solution set, determining a fitness landform feature vector corresponding to the first solution set;

[0024] Determining low-level heuristic operator performance ranking data in the low-level heuristic operator performance ranking table according to the fitness landscape feature vector corresponding to the first solution set;

[0025] A target low-level heuristic operator for the manufacturing service optimization solution is determined according to the low-level heuristic operator performance ranking data.

[0026] Optionally, obtaining the fitness landscape vector corresponding to the second solution set includes:

[0027] Processing the second solution set to determine a central solution and a non-dominated solution set;

[0028] Determine an intermediate solution set based on the central solution and the non-dominated solution set;

[0029] Determine a distance vector based on the intermediate solution set, wherein the distance vector is composed of the Euclidean distance between any two solutions in the intermediate solution set;

[0030] Dividing each solution in the intermediate solution set into a domain according to the distance vector to determine a plurality of domain groups;

[0031] determining a discrete set based on the plurality of domain groups;

[0032] According to the discrete set, a fitness landscape vector corresponding to the second solution set is determined.

[0033] Optionally, obtaining the ranking weight of the target low-level heuristic operator includes:

[0034] Obtaining a quality vector and a selection probability vector of the target low-level heuristic operator;

[0035] processing the first solution set according to the target low-level heuristic operator to determine a reward value;

[0036] updating the mass vector according to the reward value to determine an updated mass vector;

[0037] The selection probability vector is processed according to the updated quality vector to determine a ranking weight of the target low-level heuristic operator.

[0038] Optionally, determining a target manufacturing service preferred solution set according to the fitness landscape vector corresponding to the second solution set and the ranking weight of the target low-level heuristic operator includes:

[0039] determining an iterative low-level heuristic operator according to the ranking weight of the target low-level heuristic operator;

[0040] processing the second solution set according to the iterative low-level heuristic operator to determine a current optimal solution set;

[0041] The current optimal solution set is compared and judged to determine the target manufacturing service preferred solution set.

[0042] An embodiment of the present invention further provides a device for processing a manufacturing task, comprising:

[0043] a first acquisition module, configured to acquire an initial set of manufacturing service preferred solutions for a manufacturing task and a preset low-level heuristic operator queue for determining the manufacturing service preferred solutions;

[0044] A first determination module is configured to determine a low-level heuristic operator performance ranking table based on the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue, wherein the preset low-level heuristic operator queue includes a plurality of low-level heuristic operators; determine a first solution set of the manufacturing service preferred solution based on the initial set of manufacturing service preferred solutions and the preset low-level heuristic operator queue; determine a target low-level heuristic operator of the manufacturing service preferred solution based on the first solution set and the low-level heuristic operator performance ranking table; and determine a second solution set of the manufacturing service preferred solution based on the first solution set and the target low-level heuristic operator;

[0045] a second acquisition module, configured to acquire a fitness landscape vector corresponding to the second solution set and a ranking weight of the target low-level heuristic operator;

[0046] The second determination module is used to determine the target manufacturing service preferred solution set based on the fitness landscape vector corresponding to the second solution set and the ranking weight of the target low-level heuristic operator; determine the manufacturing task corresponding to the manufacturing service preferred solution based on the target manufacturing service preferred solution set, and schedule the manufacturing task.

[0047] An embodiment of the present invention further provides a computing device, comprising: a processor and a memory storing a computer program, wherein the computer program executes the above method when executed by the processor.

[0048] An embodiment of the present invention further provides a computer-readable storage medium storing instructions, which, when executed on a computer, enable the computer to execute the above method.

[0049] The technical solution of the present invention includes at least the following effects:

[0050] The above-mentioned scheme of the present invention obtains an initial set of manufacturing service preferred solutions for manufacturing tasks and a preset low-level heuristic operator queue for determining the manufacturing service preferred solutions; determines a low-level heuristic operator performance ranking table based on the initial set of manufacturing service preferred solutions and the preset low-level heuristic operator queue, wherein the preset low-level heuristic operator queue includes multiple low-level heuristic operators; determines a first solution set of the manufacturing service preferred solution based on the initial set of manufacturing service preferred solutions and the preset low-level heuristic operator queue; determines a target low-level heuristic operator of the manufacturing service preferred solution based on the first solution set and the low-level heuristic operator performance ranking table; determines a second solution set of the manufacturing service preferred solution based on the first solution set and the target low-level heuristic operator; obtains the fitness landscape vector corresponding to the second solution set and the ranking weight of the target low-level heuristic operator; determines the target manufacturing service preferred solution set based on the fitness landscape vector corresponding to the second solution set and the ranking weight of the target low-level heuristic operator; determines the manufacturing task corresponding to the manufacturing service preferred solution based on the target manufacturing service preferred solution set, and schedules the manufacturing task, thereby realizing efficient and accurate selection of manufacturing services. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is a flow chart of a method for processing a manufacturing task provided by an embodiment of the present invention;

[0052] Figure 2 Schematic diagram of the random walk process of the method for processing a manufacturing task provided by an embodiment of the present invention;

[0053] Figure 3 1 is a schematic diagram of a process for dividing domains of a method for processing a manufacturing task provided by an embodiment of the present invention;

[0054] Figure 4 This is a schematic diagram of coding for manufacturing service optimization problems in a method for processing manufacturing tasks provided by an embodiment of the present invention;

[0055] Figure 5 1 is a schematic diagram of a subtask sequential crossover process of a manufacturing task processing method provided by an embodiment of the present invention;

[0056] Figure 6 2. It is a schematic diagram of a service selection and cross-processing method of a manufacturing task provided by an embodiment of the present invention;

[0057] Figure 7 1 is a schematic diagram of a variation process of a method for processing a manufacturing task provided by an embodiment of the present invention;

[0058] Figure 8 This is a flowchart of the processing and manufacturing tasks of automotive parts provided by an embodiment of the present invention;

[0059] Figure 9 is a structural diagram of a manufacturing task processing device provided by an embodiment of the present invention;

[0060] Figure 10 It is a structural diagram of a computing device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0061] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0062] like Figure 1 As shown, an embodiment of the present invention provides a method for processing a manufacturing task, comprising:

[0063] Step 11, obtaining an initial set of manufacturing service preferred solutions for the manufacturing task and a preset low-level heuristic operator queue for determining the manufacturing service preferred solutions;

[0064] Step 12: determining a low-level heuristic operator performance ranking table based on the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue, wherein the preset low-level heuristic operator queue includes a plurality of low-level heuristic operators;

[0065] Step 13, determining a first solution set of the manufacturing service preferred solutions according to the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue;

[0066] Step 14: determining a target low-level heuristic operator for the manufacturing service optimization solution based on the first solution set and the low-level heuristic operator performance ranking table;

[0067] Step 15, determining a second solution set of the manufacturing service optimization solution based on the first solution set and the target low-level heuristic operator;

[0068] Step 16, obtaining the fitness landscape vector corresponding to the second solution set and the ranking weight of the target low-level heuristic operator;

[0069] Step 17: determining a target manufacturing service preferred solution set based on the fitness landscape vector corresponding to the second solution set and the ranking weight of the target low-level heuristic operator;

[0070] Step 18: Determine the manufacturing tasks corresponding to the target manufacturing service preferred solution according to the target manufacturing service preferred solution set, and schedule the manufacturing tasks.

[0071] In this embodiment, the initial set of preferred manufacturing service solutions is a set of possible manufacturing service solutions for a specific manufacturing task, collected through methods such as historical data, expert experience, and preliminary screening. These solutions cover different manufacturing processes, equipment options, and supplier combinations, each representing a possible approach to solving the manufacturing task. For example, for a manufacturing task involving automotive parts, the initial set includes solutions using different machine tools and different raw material sources.

[0072] The preset low-level heuristic operator queue is a tool or rule used to manipulate and optimize the manufacturing service preferred solution. The preset low-level heuristic operator queue consists of multiple low-level heuristic operators arranged in a specific order. Their function is to improve and adjust the manufacturing service preferred solution to find a more optimal solution.

[0073] By applying each low-level heuristic operator in a preset low-level heuristic operator queue to an initial set of preferred manufacturing service solutions, the optimization effect of each low-level heuristic operator on the solution is evaluated, thereby determining the performance ranking of the low-level heuristic operators. Specifically, the low-level heuristic operators in the preset low-level heuristic operator queue are sequentially applied to the solutions in the initial set to generate a series of new solutions. These new solutions are then evaluated based on certain evaluation metrics, such as cost, quality, and time, and the average performance improvement or degree of optimization of the solutions generated by each low-level heuristic operator is calculated. Finally, the low-level heuristic operators are ranked from high to low based on performance, forming a low-level heuristic operator performance ranking table. For example, if the solutions generated by a low-level heuristic operator reduce cost by an average of 10%, while other low-level heuristic operators reduce cost by a smaller margin, then this low-level heuristic operator will be ranked higher in the ranking table.

[0074] The solutions in the initial set are repeatedly processed and optimized using low-level heuristic operators from a preset low-level heuristic operator queue to generate a new set of optimal manufacturing service solutions, which constitute the first solution set. Specifically, a number of solutions are selected from the initial set and combined using low-level heuristic operators from the preset low-level heuristic operator queue, such as a crossover followed by a mutation operation. The new solutions generated after each operation are added to the first solution set. After multiple iterations, the first solution set will contain a large number of new solutions processed by different low-level heuristic operators, which, to a certain extent, optimize and improve upon the initial solutions.

[0075] Combined with the optimization of the solutions in the first solution set and the performance ranking table of low-level heuristic operators, select the low-level heuristic operator with the most significant optimization effect on the current solution set as the target low-level heuristic operator. Specifically: analyze which low-level heuristic operators are used to obtain each solution in the first solution set, and how these solutions perform in terms of evaluation indicators. Referring to the performance ranking table of low-level heuristic operators, select those low-level heuristic operators that perform outstandingly in the process of generating the first solution set and can significantly improve the key indicators of the solutions as target low-level heuristic operators. For example, if the solution generated by a low-level heuristic operator in the first solution set performs well in terms of quality and cost, and the low-level heuristic operator is also ranked high in the performance ranking table of low-level heuristic operators, then it can be selected as the target low-level heuristic operator.

[0076] Using the target low-level heuristic operator as the primary operating method, the solutions in the first solution set are further optimized and processed to obtain the second solution set. Specifically, the target low-level heuristic operator is applied multiple times to each solution in the first solution set to generate new solutions. These new solutions, under the influence of the target low-level heuristic operator, are significantly optimized in certain aspects. Collecting all these newly generated solutions forms the second solution set.

[0077] The fitness landscape vector describes the fitness of each solution in the second solution set across different evaluation metrics. Each element of the vector corresponds to an evaluation metric, such as cost, quality, or time, and the value of the element represents the solution's fitness for that metric. For example, for a manufacturing task, the fitness landscape vector can be represented as [cost fitness, quality fitness, time fitness]. A solution's fitness landscape vector is [0.8, 0.9, 0.7], indicating that the solution's fitness for cost, quality, and time are 0.8, 0.9, and 0.7, respectively. The ranking weight of the target low-level heuristic operator is determined based on its performance in previous steps and its position in the low-level heuristic operator performance ranking table. The ranking weight reflects the importance of the target low-level heuristic operator in the current optimization process. For example, if the target low-level heuristic operator ranks high in the low-level heuristic operator performance ranking table and plays a significant role in generating the second solution set, its ranking weight will be higher.

[0078] Taking into account the fitness landscape vectors of the solutions in the second solution set and the ranking weights of the target low-level heuristic operators, the solutions in the second solution set are screened and sorted to determine the target manufacturing service preferred solution set. Specifically: First, the solutions in the second solution set are evaluated according to the fitness landscape vectors, and the comprehensive fitness value of each solution is calculated. The comprehensive fitness value can be obtained by weighted average based on the weights of each evaluation indicator. Then, the comprehensive fitness value is adjusted in combination with the ranking weight of the target low-level heuristic operator. For example, if the ranking weight of the target low-level heuristic operator is high, then the solution generated by the low-level heuristic operator or with a higher degree of optimization will have a greater advantage in the final ranking. Finally, the solutions in the second solution set are sorted according to the adjusted comprehensive fitness value, and the top-ranked solutions are selected to form the target manufacturing service preferred solution set.

[0079] The optimal solution is selected from the set of optimal solutions for the target manufacturing service, and the specific manufacturing task is determined based on this solution. For example, the solution specifies the use of a specific machine tool, specific process parameters, and a specific supplier of raw materials. This information constitutes the specific manufacturing task. The identified manufacturing task is assigned to the corresponding production resources, and a production plan and schedule are arranged. This involves coordinating and communicating with production equipment, personnel, and raw material suppliers to ensure the smooth execution of the manufacturing task. For example, based on the requirements of the manufacturing task, machine tool operators are assigned, the time and quantity for purchasing raw materials are determined, and a production schedule is developed.

[0080] This technical solution provides a hyper-heuristic manufacturing service optimization method based on fitness landscape analysis. By controlling the selection of low-level heuristic operators through high-level strategies, it can adapt to new problem domains without redesign when the complexity of manufacturing tasks changes. First, a new heuristic spatial landscape feature vector extraction method based on dominance relationships is proposed. Then, an adaptive learning strategy based on landscape feature vectors is designed. This strategy dynamically evaluates the performance of low-level heuristic operators by analyzing the local landscape features of the problem space in real time, and selects low-level heuristic algorithms accordingly. Therefore, this method comprehensively considers multiple factors of manufacturing services and utilizes fitness landscape analysis during the search process to characterize and guide the search process. This method overcomes the shortcomings of existing technologies, possesses a certain degree of adaptability, and can achieve more efficient and accurate manufacturing service selection.

[0081] In an optional embodiment of the present invention, step 12 may include:

[0082] Step 121 , processing the initial set of manufacturing service preferred solutions according to the preset low-level heuristic operator queue to determine a plurality of intermediate solution sets;

[0083] Step 122, determining fitness landscape vectors corresponding to the plurality of intermediate solution sets based on the plurality of intermediate solution sets;

[0084] Step 123: Determine a low-level heuristic operator performance ranking table based on the fitness landscape vectors corresponding to the multiple intermediate solution sets.

[0085] In this embodiment, the initial set of manufacturing service preferred solutions, denoted as S0, is a set of solutions of size N that are preferred for manufacturing services; each solution (i.e., each solution) is a decision variable represented by an array; the preset low-level heuristic operator queue can be represented as H = {h1, h2, ..., h num}, this is a low-level heuristic (LLH) low-level heuristic operator queue, denoted as LLHs, h1,h2,...,h num A metaheuristic algorithm that can solve the manufacturing service optimization problem is called a low-level heuristic operator. Its performance ranking table, denoted as R, is derived by ranking each LLH under different fitness landscape characteristics. Specifically, taking P problem sets as an example, all LLHs are first independently run T times on each problem set. Each run generates a solution set, resulting in a total of P × T solution sets. For each of these solution sets, the corresponding fitness landscape vector (FLV) is calculated. Based on the characteristics of this vector, the performance ranking table R of the low-level heuristic operator is obtained.

[0086] In an optional embodiment proposed by the present invention, step 13 may include:

[0087] Step 131 , using the initial set of manufacturing service preferred solutions as an input solution set, and processing it using a low-level heuristic operator queue to obtain a solution set containing multiple solution sets;

[0088] Step 132 : Evaluate the solution set to obtain a first solution set, where the first solution set is the solution set with the highest evaluation value among the multiple solution sets.

[0089] In this embodiment, first, the set H = {h1, h2, ..., h num All LLHs in} are run once with the same probability, and the initial set S0 of manufacturing service optimization solutions is used as the input solution set during the run; then the hypervolume index is used to evaluate all the solution sets obtained from the run, and the solution set S with the best evaluation value is selected. currThe starting solution set for the iteration is used. Specifically, each solution in the initial set is processed sequentially using the low-level heuristic operators in the low-level heuristic operator queue. Each time a low-level heuristic operator is applied, the solution is modified or optimized according to the low-level heuristic operator's rules, generating a new solution. After multiple iterations, each initial solution generates a new solution, which, together with the initial solution, constitutes a solution set. All generated solution sets are compared to select the solution set with the best performance. Each solution in this solution set represents the optimal manufacturing service solution obtained through the operation of different low-level heuristic operators. Based on the evaluation criteria, the solutions in each solution set are calculated and evaluated to obtain the performance value of each solution on various indicators. The performance values of all solutions in each solution set are summarized and calculated to obtain a comprehensive evaluation value for the solution set. This evaluation value reflects the overall quality of the solution set. All solution sets are sorted according to the comprehensive evaluation value, and the solution set with the highest evaluation value is selected as the first solution set. This solution set represents the optimal set of preferred manufacturing service solutions under the current evaluation criteria.

[0090] In an optional embodiment of the present invention, step 14 may include:

[0091] Step 141, determining the fitness landform feature vector corresponding to the first solution set according to the first solution set;

[0092] Step 142: determining low-level heuristic operator performance ranking data in the low-level heuristic operator performance ranking table according to the fitness landscape feature vector corresponding to the first solution set;

[0093] Step 143 : Determine a target low-level heuristic operator for the manufacturing service optimization solution based on the low-level heuristic operator performance ranking data.

[0094] In this embodiment, according to S curr The corresponding FLV is obtained in R, and the performance ranking of LLH under the current local fitness landscape is obtained. According to this ranking, the LLH with the best performance under the current fitness landscape is obtained, which is recorded as h next ; Specifically, for each solution in the first solution set, collect its performance data on each evaluation indicator; based on the collected data, calculate the fitness characteristics of the first solution set on each evaluation indicator; arrange the calculated fitness characteristics in the order of the evaluation indicators to form a fitness landscape feature vector.

[0095] Match the fitness landscape feature vector of the first solution set with the data in the performance ranking table of low-level heuristic operators. Consider the performance of low-level heuristic operators under different fitness landscapes, or screen low-level heuristic operators based on key indicators in the feature vector, such as cost fitness and quality fitness. For each low-level heuristic operator, evaluate its performance under fitness landscapes similar or related to the first solution set. This can be achieved by examining the performance scores under corresponding or similar features in the performance ranking table of low-level heuristic operators. Based on the evaluation results, extract the performance ranking data of the low-level heuristic operators related to the fitness landscape feature vector of the first solution set. This data includes the ranking and performance score of the low-level heuristic operators.

[0096] The extracted low-level heuristic operator performance ranking data is analyzed to understand the performance of each low-level heuristic operator under fitness landscapes similar or related to the first solution set. Based on the analysis results, a greedy strategy is used to prioritize the low-level heuristic operator with the best performance. The continuous improvement of this low-level heuristic operator is monitored, and random low-level heuristic operator selection is automatically triggered when a certain threshold is reached. The final target low-level heuristic operator will be used for further processing and optimization of the manufacturing service optimization solution in subsequent steps.

[0097] In an optional embodiment proposed by the present invention, step 16 may include:

[0098] Step 1611, processing the second solution set to determine a central solution and a non-dominated solution set;

[0099] Step 1612: determining an intermediate solution set based on the central solution and the non-dominated solution set;

[0100] Step 1613: determining a distance vector based on the intermediate solution set, wherein the distance vector is formed by the Euclidean distance between any two solutions in the intermediate solution set;

[0101] Step 1614: Divide each solution in the intermediate solution set into domains according to the distance vector to determine a plurality of domain groups;

[0102] Step 1615 , determining a discrete set based on the multiple domain groups;

[0103] Step 1616: Determine the fitness landscape vector corresponding to the second solution set based on the discrete set.

[0104] In this embodiment, the central solution S of the second solution set S is first calculated. c .

[0105]

[0106] Its essence is to obtain the central solution S by calculating the central value of the decision variable in each dimension. c . Among them S i Represents the i-th (i=1,2,…,N) solution in the solution set S.

[0107] Perform non-dominated sorting on the solution set to obtain the non-dominated solution set S of the solution set S nd If it is not the dominating solution set S nd If the scale is too large, it will significantly increase the computational cost of the random walk. In order to effectively improve the speed of the random walk, it is necessary to prune the non-dominated solution set so that the size of the simplified solution set N′ satisfies N / 2≤N′≤3N / 4.

[0108] Targeting S nd Every solution S in nd,i , where i = 1, 2, ..., N', the new solution set S' i Calculated according to the following formula:

[0109]

[0110] in, is a control parameter, and its calculation formula is:

[0111]

[0112] Where k∈(1, 2, ..., NP) is the number of times the current statistical calculation is performed. NP refers to the dimension of the decision variable. When (S nd,i -S c )>0, parameter α max and α min The calculation formula is:

[0113]

[0114] Where, E min and E max is the lower and upper bound of each dimension of the decision variables of all individuals in the current solution set. nd,i -S c )<0, E min and E max In the formula, exchange, that is, α max E in the formula max Use E min Replace, and α min E in the formula min Use E max replace.

[0115] Perform multi-point random walk processing on S′ to obtain a new intermediate solution set S of size λ spMulti-point random walk is a random process. Specifically, for each solution in S′, a position is randomly selected within the upper and lower limits of the decision variable, and the decision variable at that position is mutated to generate a new intermediate solution set S. sp .

[0116] For the intermediate solution set S sp For all solutions in , calculate the Euclidean distance between each two and save it in the distance vector D. The Euclidean distance between two solutions is calculated as:

[0117]

[0118] Where n is the dimension of the decision variable, d(S1,S2) is the Euclidean distance between two solutions, S 1,i 、S 2,i is the intermediate solution set S sp Any two solutions in .

[0119] According to the distance vector D, sp Each solution in is divided into a neighborhood of size K. Take K = 2 and the number of random walks 4 as an example. The random walk process is as follows Figure 2 As shown in Figure 3 As shown, in this process, the two initial solutions P i,0 and P j,0 Four random walks are performed to obtain four new solutions, resulting in a total of 10 solutions. Figure 3 Figure 2 shows the results of neighborhood partitioning for the 10 solutions obtained after the random walk. Specifically, for each of the 10 solutions, the two closest solutions are selected and grouped together. Ultimately, this method yields a total of six neighborhood groups.

[0120] Calculate the discrete set δ of non-dominated relations between solution sets in each neighborhood group after neighborhood partitioning. Solution s i The corresponding discrete set δ(s i ,s i,j )∈{-1,0,1} can be solved by s i and its neighborhood solution s i,j Calculation, the formula is:

[0121]

[0122] Use each solution δ(s in the computational neighborhood i ,s i,j ), we get the array U(δ), the formula is:

[0123]

[0124] Among them, U(δi ) is the i-th element in the array U(δ).

[0125] The fitness landscape vector corresponding to the second solution set is obtained by using U(δ) through histogram distribution processing. The formula is:

[0126] HSFLV(S)=[UP(0),UP(0),...,UP(k)]

[0127] Among them, HSFLV(S) is the fitness landscape vector corresponding to the solution set S, and UP(k) is the fitness value of the set S on the kth evaluation indicator or dimension.

[0128] In an optional embodiment of the present invention, step 16 may further include:

[0129] Step 1621, obtaining the quality vector and selection probability vector of the target low-level heuristic operator;

[0130] Step 1622: Process the first solution set according to the target low-level heuristic operator to determine a reward value;

[0131] Step 1623: Update the quality vector according to the reward value to determine an updated quality vector;

[0132] Step 1624: Process the selection probability vector according to the updated quality vector to determine the ranking weight of the target low-level heuristic operator.

[0133] In this embodiment, a set of low-level heuristics of size num is first given H = {h1, h2, ..., h num}. Then the probability vector selected at decision point t is W t =(w 1,t ,w 2,t …,w i,t )(where 0≤w i,t ≤1, i=1,2,…,num). The mass vector of LLH is Q t =(q 1,t ,…,q i,t ).

[0134] When applying the low-level heuristic operator h a Afterwards, calculate the reward value r a,t Specifically, if h a The solution set obtained after the above is not as good as the current optimal solution set in terms of performance, then r a,t = 0. Otherwise, update r using the following formula a,t

[0135]

[0136] Among them, E(h a ) is the application h a The evaluation value of the obtained solution set; E best is the evaluation value of the current optimal solution set.

[0137] Update the quality vector Q t Middle Q a,t+1 , while maintaining the q-values of other low-level heuristic operators, the update formula is:

[0138]

[0139] According to the updated mass vector, the LLH with the maximum current q value is obtained and recorded as h a* . Then add low-level heuristic operator h a* Choose the probability value w a*,t+1 , the specific formula is:

[0140] w a*,t+1 =(1-β)×w a*,t +β×w max (0≤β≤1)

[0141] Reduce the selection probability value w of other low-level heuristic operators i,t+1 , the specific formula is

[0142]

[0143] Among them, α and β are two control parameters with a value range of [0,1]. In adaptive tracking, in order to prevent some low-level heuristic operators from being no longer selected due to the convergence of their selection probability values, two parameters w are introduced. min and w max Make the selection probability in the interval [w min ,w max ]. Among them, 0 <w min <w max <1 and satisfies w max =1-(num-1)×w min .

[0144] In an optional embodiment of the present invention, step 17 may include:

[0145] Step 171, determining an iterative low-level heuristic operator according to the ranking weight of the target low-level heuristic operator;

[0146] Step 172, processing the second solution set according to the iterative low-level heuristic operator to determine a current optimal solution set;

[0147] Step 173 : performing comparison and judgment processing on the current optimal solution set to determine the target manufacturing service preferred solution set.

[0148] In this embodiment, the ranking weights of the target low-level heuristic operators are first analyzed to understand the relative importance of each low-level heuristic operator in the optimization process. Based on the results of the weight analysis, the low-level heuristic operator with the higher weight is selected as the iterative low-level heuristic operator. The final iterative low-level heuristic operator is used for subsequent processing and optimization of the second solution set.

[0149] Iterative low-level heuristic operators are applied to each solution in the second solution set to optimize and improve the solution. The specific modes of action of the iterative low-level heuristic operators may include operations such as crossover, mutation, and environmental selection, aiming to improve the fitness of the solutions. After processing by the iterative low-level heuristic operators, the solutions in the second solution set are changed, generating new solutions or improving existing solutions. The updated solution set will contain these new or improved solutions. During the iterative process, the best-performing solution set is selected from the updated solution set based on some evaluation criteria (such as fitness value, cost, quality, etc.) and becomes the current optimal solution set.

[0150] Establish clear evaluation criteria, encompassing multiple dimensions such as cost, quality, time, and efficiency, reflecting the key requirements and objectives of the manufacturing task. Based on these criteria, compare and analyze the solutions in the current optimal solution set. Compare the performance of each solution across various metrics to understand their strengths and weaknesses and differences. After this comparison and analysis, select the solutions that perform best across all or key metrics to form the target manufacturing service optimal solution set. This solution set represents the optimal set of optimal manufacturing service solutions for the current manufacturing task requirements and objectives.

[0151] A specific embodiment of the method for processing a manufacturing task provided by an embodiment of the present invention is as follows:

[0152] In a real-world industrial manufacturing scenario, an automobile manufacturer needs to select the optimal service combination from numerous parts suppliers to complete the processing and manufacturing of auto parts. This task involves multiple subtasks, such as stamping, welding, and painting, each of which has multiple potential service providers to choose from. The company needs to comprehensively consider factors such as cost, service quality, and delivery time to select the optimal service combination to meet production needs. First, they research the subtasks that parts suppliers can complete and number them. Then, they divide the auto parts processing and manufacturing task into smaller subtasks.

[0153] Before processing manufacturing tasks, it is necessary to abstract the manufacturing service optimization into a multi-objective optimization problem. The two sub-problems of service selection and sub-task arrangement are encoded. The encoding diagram is as follows: Figure 4 shown.

[0154] Figure 4Middle left (i.e. Figure 4 The a part in the middle) represents the execution order of subtasks of different tasks. Figure 4 Part (b) represents the candidate services assigned to the subtasks. The number of codes in the left half equals the number of subtasks. The numerical code corresponding to each gene represents a different task number, and the number of times the number appears represents the number of subtasks. When traversing the chromosome from left to right, the kth occurrence of a task number represents the kth subtask of that task. If the first half of the code is [3, 2, 3, 2, 1, 1, 2, 3], then the first number 3 represents the first subtask of task 3, and the fourth number 2 represents the second subtask of task 2.

[0155] The second half of the chromosome contains the same number of codes as the first, indicating the matching service for each subtask. The numbers represent the kth subtask of the nth task. If the chromosome in the second half is [1, 2, 2, 1, 3, 4, 2, 1], then the 1 in the first position indicates that the first subtask of task 3 is assigned to service provider 1, and the 2 in the third position indicates that the second subtask of task 3 is assigned to service provider 2. The service provider for each subtask is selected from the service provider set.

[0156] When decoding, the service provider information on the right side is first parsed, and then the subtask order is determined based on the left side. The above encoding and decoding methods combined with task priority can dynamically select services.

[0157] The offspring generation method in this method uses a method that selects N / 2 pairs of parents with equal probability from the parent solution set to generate offspring. The crossover operation uses two methods: for the service selection problem (right part), the uniform crossover method is used to generate offspring; for the crossover of the subtask part (left part), the POX (Precedence Operation Crossover) crossover method is used, which does not violate the task predecessor relationship. The crossover and mutation example is shown in the figure Figures 5 to 7 shown.

[0158] Three classic multi-objective optimization algorithms were selected as low-level heuristic operators: h1: SPEA2, h2: NSGAII, and h3: IBEA. These algorithms have demonstrated excellent performance and widespread application in the field of multi-objective optimization. To ensure algorithmic effectiveness and comparability, the parameters of these LLHs were uniformly set: the solution set size N was set to 91, and new solution set generation, crossover, and mutation were performed using the methods described above. The crossover probability was set to 0.9, the mutation probability was set to 1 / n (where n is the dimension of the decision variable), and the distribution exponent for both crossover and mutation was set to 20. Furthermore, the evaluation function uses the hypervolume (HV) metric, which measures the volume covered by the approximate solution set in the target space. Its value range is [0, +∞). A larger HV value indicates a higher overall quality and better distribution of the approximate solution set, meaning that these solutions are balanced across all target dimensions and cover a larger target space. The reference point setting is consistent with relevant research. The specific steps for solving the automotive parts processing and manufacturing task are as follows:

[0159] Step 1: Randomly generate a solution set S0 of size N (S0 is the initial solution set, i.e., the 0th generation solution set).

[0160] Step 2: Use the three selected LLHs to run the initial solution set S0 independently. After the run, use the HV index to evaluate all the obtained solution sets and select the solution set S with the best evaluation value. curr As the starting solution set of the HH-FLV algorithm. At the same time, according to S curr The corresponding fitness landscape feature vector (FLV) is obtained from the pre-established performance ranking table R to obtain the performance ranking of each LLH under the current local fitness landscape, and the LLH with the best performance is determined as the initial low-level heuristic operator h for the next iteration. next .

[0161] Step 3: Optimize the initial solution set S through loop iteration curr The steps in the loop are:

[0162] (1) Using the initial solution set S curr and the selected LLH h next Start iteration and get a new solution set S next In this process, the selected LLH optimization solution set S curr , obtain a new solution set S through crossover, mutation and environmental selection next .

[0163] (2) Then use the method of obtaining the local fitness landscape vector to obtain the new solution set S next The corresponding fitness landscape vector. According to the ranking R of each LLH on different fitness landscapes, the ranking R of the current LLH is obtained. land .

[0164] (3) Calculate the adaptive tracking strategy weight R of the current iteration according to the adaptive tracking strategy ap .

[0165] (4) Calculate the final weight R of LLH final , where w1 and w2 are weight coefficients, which must satisfy 0≤w1≤1 and 0≤w2≤1 and w1+w2=1. The final weight R of LLH final The calculation formula is:

[0166] Rfi na l=(w1×R land )+(w2×R ap )

[0167] (5) According to the final weight R final Select the LLH for the next iteration. Record the current optimal solution set in each iteration. Compare the new solution set with the current optimal solution set to determine whether there is any improvement. If the solution set has not improved for count consecutive times, randomly select the LLH for the next iteration. Otherwise, use the greedy strategy according to R final Select the LLH for the next iteration.

[0168] (6) Select the solution set S for the next iteration curr The acceptance strategy only accepts new solutions whose quality is better than or similar to the current optimal solution.

[0169] (7) Determine whether the conditions for loop termination are met. If so, the loop is terminated and the solution set S after the last iteration is output. curr As the final solution set, otherwise jump to step 1, the processing flow is as follows Figure 8 shown.

[0170] The final solution set is decoded to obtain a service combination solution for the processing and manufacturing tasks of automotive parts.

[0171] The present invention provides a method for processing manufacturing tasks, which controls the selection of low-level heuristic operators through high-level strategies. When the complexity of the manufacturing task changes, it can adapt to the new problem domain without redesign. First, a new heuristic spatial landform feature vector extraction method is proposed based on the dominance relationship, and then an adaptive learning strategy based on the landform feature vector is designed. This strategy dynamically evaluates the performance of low-level heuristic operators by analyzing the local landform features of the problem space in real time, and selects low-level heuristic algorithms accordingly. Therefore, this method comprehensively considers various factors of manufacturing services, uses fitness landform analysis in the search process to characterize and guide the search process, overcomes the shortcomings of existing technologies, has a certain degree of adaptive ability, and can achieve more efficient and accurate manufacturing service selection.

[0172] like Figure 9 As shown, the embodiment of the present invention further provides a manufacturing task processing device 90, comprising:

[0173] A first acquisition module 91 is configured to acquire an initial set of manufacturing service preferred solutions for a manufacturing task and a preset low-level heuristic operator queue for determining the manufacturing service preferred solutions;

[0174] A first determining module 92 is configured to determine a low-level heuristic operator performance ranking table based on the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue, wherein the preset low-level heuristic operator queue includes a plurality of low-level heuristic operators; determine a first solution set of the manufacturing service preferred solution based on the initial set of manufacturing service preferred solutions and the preset low-level heuristic operator queue; determine a target low-level heuristic operator of the manufacturing service preferred solution based on the first solution set and the low-level heuristic operator performance ranking table; and determine a second solution set of the manufacturing service preferred solution based on the first solution set and the target low-level heuristic operator;

[0175] A second acquisition module 93 is configured to acquire a fitness landscape vector corresponding to the second solution set and a ranking weight of the target low-level heuristic operator;

[0176] The second determination module 94 is used to determine the target manufacturing service preferred solution set based on the fitness landscape vector corresponding to the second solution set and the ranking weight of the target low-level heuristic operator; determine the manufacturing task corresponding to the manufacturing service preferred solution based on the target manufacturing service preferred solution set, and schedule the manufacturing task.

[0177] Optionally, the first determining module 92 is specifically configured to:

[0178] Processing the initial set of manufacturing service preferred solutions according to the preset low-level heuristic operator queue to determine a plurality of intermediate solution sets;

[0179] Determining, based on the plurality of intermediate solution sets, fitness landform vectors corresponding to the plurality of intermediate solution sets;

[0180] A low-level heuristic operator performance ranking table is determined according to the fitness landscape vectors corresponding to the multiple intermediate solution sets.

[0181] Optionally, the first determining module 92 is further specifically configured to:

[0182] The initial set of manufacturing service preferred solutions is used as an input solution set, and is processed using a low-level heuristic operator queue to obtain a solution set containing multiple solution sets;

[0183] The solution sets are evaluated to obtain a first solution set, where the first solution set is a solution set with the highest evaluation value among the multiple solution sets.

[0184] Optionally, the first determining module 92 is further specifically configured to:

[0185] According to the first solution set, determining a fitness landform feature vector corresponding to the first solution set;

[0186] Determining low-level heuristic operator performance ranking data in the low-level heuristic operator performance ranking table according to the fitness landscape feature vector corresponding to the first solution set;

[0187] A target low-level heuristic operator for the manufacturing service optimization solution is determined according to the low-level heuristic operator performance ranking data.

[0188] Optionally, the second acquisition module 93 is specifically configured to:

[0189] Processing the second solution set to determine a central solution and a non-dominated solution set;

[0190] Determine an intermediate solution set based on the central solution and the non-dominated solution set;

[0191] Determine a distance vector based on the intermediate solution set, wherein the distance vector is composed of the Euclidean distance between any two solutions in the intermediate solution set;

[0192] Dividing each solution in the intermediate solution set into a domain according to the distance vector to determine a plurality of domain groups;

[0193] determining a discrete set based on the plurality of domain groups;

[0194] According to the discrete set, a fitness landscape vector corresponding to the second solution set is determined.

[0195] Optionally, the second obtaining module 93 is further specifically configured to:

[0196] Obtaining a quality vector and a selection probability vector of the target low-level heuristic operator;

[0197] processing the first solution set according to the target low-level heuristic operator to determine a reward value;

[0198] updating the mass vector according to the reward value to determine an updated mass vector;

[0199] The selection probability vector is processed according to the updated quality vector to determine a ranking weight of the target low-level heuristic operator.

[0200] Optionally, the second determining module 94 is specifically configured to:

[0201] determining an iterative low-level heuristic operator according to the ranking weight of the target low-level heuristic operator;

[0202] processing the second solution set according to the iterative low-level heuristic operator to determine a current optimal solution set;

[0203] The current optimal solution set is compared and judged to determine the target manufacturing service preferred solution set.

[0204] It should be noted that this device is a device corresponding to the above method, and all implementation methods in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0205] like Figure 10 As shown, an embodiment of the present invention further provides a computing device 100, including a processor 101, a memory 102, and a program or instruction stored in the memory 102 and executable on the processor 101. When executed by the processor 101, the program or instruction implements each process of the embodiment of the processing method for the manufacturing task described above, and can achieve the same technical effect. To avoid repetition, it will not be described here. It should be noted that the computing device in the embodiment of the present invention includes the mobile electronic device and the non-mobile electronic device described above.

[0206] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0207] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0208] In the embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0209] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0210] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0211] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the various embodiments of the method of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard disk, ROM, RAM, a magnetic disk, or an optical disk.

[0212] In addition, it should be noted that, in the apparatus and method of the present invention, it is obvious that each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent schemes of the present invention. Moreover, the steps of performing the above-mentioned series of processing can naturally be performed in chronological order according to the order of description, but it is not necessary to perform them in chronological order, and some steps can be performed in parallel or independently of each other. For those of ordinary skill in the art, it will be understood that all or any steps or components of the method and apparatus of the present invention can be implemented in any computing device (including processors, storage media, etc.) or a network of computing devices in hardware, firmware, software or a combination thereof, which can be achieved by those of ordinary skill in the art using their basic programming skills after reading the description of the present invention.

[0213] Therefore, the purpose of the present invention can also be achieved by running a program or a group of programs on any computing device. The computing device can be a well-known general-purpose device. Therefore, the purpose of the present invention can also be achieved simply by providing a program product containing program code for implementing the method or device. That is to say, such a program product also constitutes the present invention, and the storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any well-known storage medium or any storage medium developed in the future. It should also be pointed out that in the device and method of the present invention, it is obvious that each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent schemes of the present invention. In addition, the steps of performing the above-mentioned series of processing can naturally be performed in chronological order according to the order of description, but do not necessarily need to be performed in chronological order. Certain steps can be performed in parallel or independently of each other.

[0214] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for processing a manufacturing task, characterized in that: include: Obtaining an initial set of manufacturing service preferred solutions for the manufacturing task and a preset low-level heuristic operator queue for determining the manufacturing service preferred solutions; Determining a low-level heuristic operator performance ranking table according to the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue, wherein the preset low-level heuristic operator queue includes a plurality of low-level heuristic operators; Determining a first solution set of the manufacturing service preferred solutions according to the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue; Determining a target low-level heuristic operator for a manufacturing service optimization solution based on the first solution set and the low-level heuristic operator performance ranking table; determining a second solution set of manufacturing service optimization solutions based on the first solution set and a target low-level heuristic operator; Obtaining a fitness landscape vector corresponding to the second solution set and a ranking weight of the target low-level heuristic operator; Determining a target manufacturing service preferred solution set according to the fitness landscape vector corresponding to the second solution set and the ranking weight of the target low-level heuristic operator; According to the target manufacturing service preferred solution set, manufacturing tasks corresponding to the manufacturing service preferred solutions are determined, and the manufacturing tasks are scheduled.

2. The method for processing a manufacturing task according to claim 1, characterized in that: Determining a low-level heuristic operator performance ranking table based on the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue includes: Processing the initial set of manufacturing service preferred solutions according to the preset low-level heuristic operator queue to determine a plurality of intermediate solution sets; Determining, based on the plurality of intermediate solution sets, fitness landform vectors corresponding to the plurality of intermediate solution sets; A low-level heuristic operator performance ranking table is determined according to the fitness landscape vectors corresponding to the multiple intermediate solution sets.

3. The method for processing a manufacturing task according to claim 1, characterized in that: Determining a first solution set of the manufacturing service preferred solutions according to the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue includes: The initial set of manufacturing service preferred solutions is used as an input solution set, and is processed using a low-level heuristic operator queue to obtain a solution set containing multiple solution sets; The solution sets are evaluated to obtain a first solution set, where the first solution set is a solution set with the highest evaluation value among the multiple solution sets.

4. The method for processing a manufacturing task according to claim 1, characterized in that: Determining a target low-level heuristic operator for the manufacturing service optimization solution based on the first solution set and the low-level heuristic operator performance ranking table includes: According to the first solution set, determining a fitness landform feature vector corresponding to the first solution set; Determining low-level heuristic operator performance ranking data in the low-level heuristic operator performance ranking table according to the fitness landscape feature vector corresponding to the first solution set; A target low-level heuristic operator for the manufacturing service optimization solution is determined according to the low-level heuristic operator performance ranking data.

5. The method for processing a manufacturing task according to claim 1, characterized in that: Obtaining the fitness landscape vector corresponding to the second solution set includes: Processing the second solution set to determine a central solution and a non-dominated solution set; Determine an intermediate solution set based on the central solution and the non-dominated solution set; Determine a distance vector based on the intermediate solution set, wherein the distance vector is composed of the Euclidean distance between any two solutions in the intermediate solution set; Dividing each solution in the intermediate solution set into a domain according to the distance vector to determine a plurality of domain groups; determining a discrete set based on the plurality of domain groups; According to the discrete set, a fitness landscape vector corresponding to the second solution set is determined.

6. The method for processing a manufacturing task according to claim 1, characterized in that: Obtaining the ranking weight of the target low-level heuristic operator includes: Obtaining a quality vector and a selection probability vector of the target low-level heuristic operator; processing the first solution set according to the target low-level heuristic operator to determine a reward value; updating the mass vector according to the reward value to determine an updated mass vector; The selection probability vector is processed according to the updated quality vector to determine a ranking weight of the target low-level heuristic operator.

7. The method for processing a manufacturing task according to claim 1, characterized in that: Determining a target manufacturing service preferred solution set according to the fitness landscape vector corresponding to the second solution set and the ranking weight of the target low-level heuristic operator includes: determining an iterative low-level heuristic operator according to the ranking weight of the target low-level heuristic operator; processing the second solution set according to the iterative low-level heuristic operator to determine a current optimal solution set; The current optimal solution set is compared and judged to determine the target manufacturing service preferred solution set.

8. A processing device for manufacturing tasks, characterized in that: include: a first acquisition module, configured to acquire an initial set of manufacturing service preferred solutions for a manufacturing task and a preset low-level heuristic operator queue for determining the manufacturing service preferred solutions; A first determination module is configured to determine a low-level heuristic operator performance ranking table based on the initial set of manufacturing service preferred solutions and a preset low-level heuristic operator queue, wherein the preset low-level heuristic operator queue includes a plurality of low-level heuristic operators; determine a first solution set of the manufacturing service preferred solution based on the initial set of manufacturing service preferred solutions and the preset low-level heuristic operator queue; determine a target low-level heuristic operator of the manufacturing service preferred solution based on the first solution set and the low-level heuristic operator performance ranking table; and determine a second solution set of the manufacturing service preferred solution based on the first solution set and the target low-level heuristic operator; a second acquisition module, configured to acquire a fitness landscape vector corresponding to the second solution set and a ranking weight of the target low-level heuristic operator; The second determination module is used to determine the target manufacturing service preferred solution set based on the fitness landscape vector corresponding to the second solution set and the ranking weight of the target low-level heuristic operator; determine the manufacturing task corresponding to the manufacturing service preferred solution based on the target manufacturing service preferred solution set, and schedule the manufacturing task.

9. A computing device, characterized in that include: A processor and a memory storing a computer program, wherein when the computer program is executed by the processor, the method according to any one of claims 1 to 7 is performed.

10. A computer-readable storage medium, characterized in that The device stores instructions, which, when executed on a computer, enable the computer to execute the method according to any one of claims 1 to 7.

Images