A Production Line Workstation Partitioning Method Based on Rule Engine and Intelligent Optimization Algorithm

Abstract

This invention provides a production line workstation partitioning method based on a rule engine and intelligent optimization algorithm, relating to the field of intelligent manufacturing technology. First, a workstation partitioning business rule set is constructed, including assembly process generation rules, workstation cycle time calculation rules, and workstation update rules. The source data file is parsed to obtain the workstation partitioning cycle time constraints, assembly information model, assembly sequence, and assembly process set. The input data is parsed using a rule engine, and after rule matching and execution, candidate workstation partitioning results are generated. Finally, a genetic-particle swarm optimization hybrid algorithm is used to iteratively solve the objective optimization function, outputting the workstation partitioning optimization result.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology, and in particular to a method for dividing production line workstations based on a rule engine and intelligent optimization algorithm. Background Technology

[0002] A workstation is a fixed working position on a production line designed to complete a specific process; it is the basic spatial unit for organizing production activities. Dividing the production line into workstations means breaking down the entire product manufacturing process into multiple task points that can be completed at independent stations, with each workstation responsible for a portion of the work. The division of production line workstations is the foundation of lean manufacturing; its rationality directly affects product production efficiency, quality, and cost, and is key to improving production line balance and resource utilization.

[0003] Currently, the division of production line workstations is basically at the human-computer interaction stage. Process designers need to meet the requirements of various process specifications and comprehensively consider factors such as factory and workshop space, equipment costs, number of personnel, maintainability, and operability. Design quality heavily relies on the designer's experience, resulting in problems such as low design efficiency and unstable quality. With the development of the manufacturing industry, product complexity is increasing and the iteration speed is accelerating, further amplifying the above problems.

[0004] Therefore, a new method is urgently needed to achieve high-quality and efficient division of production line workstations. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a production line workstation partitioning method based on a rule engine and intelligent optimization algorithm;

[0006] A production line workstation partitioning method based on rule engine and intelligent optimization algorithm, such as Figure 1 As shown, it includes the following steps:

[0007] Step S1: Construct a set of business rules for workstation division, including assembly process generation rules, workstation cycle time calculation rules, and workstation update rules.

[0008] The assembly process generation rules are used to generate a set of feasible assembly processes for part A;

[0009] The station cycle time calculation rule is used to calculate the station cycle time after the current station completes the assembly of part A.

[0010] The station update rule is used to determine the assembly station of part A and update the assembly information of the station.

[0011] The assembly information includes assembly parts, feeding parts, cycle time, tools, labor time, and process content attributes.

[0012] Step S2: Parse the source data file to obtain the cycle time constraints of the workstation division, the assembly information model, the assembly sequence, and the set of assembly processes;

[0013] Step S2.1: Product attribute analysis;

[0014] Import the logistics list BOM and process requirement documents, and use the RAG engine to extract the manual, semi-automatic and automatic station cycle time limits of the product through prompt words to form the cycle time constraints of the station division.

[0015] Import the product's 3D model and extract the names, assembly relationships, assembly positions, mass, dimensions, process requirements, centroid coordinates, reference parts, coordinate systems, and assembly parent nodes of each component from the CAD parsing library to form an assembly information model. The assembly information model is a list of data dictionaries, which includes the field names, data types, and remarks corresponding to the assembly information.

[0016] Step S2.2: Product assembly sequence planning;

[0017] With the optimization goals of ease of product assembly, reliability, and continuity, an optimized assembly sequence is generated using an assembly information model and a genetic algorithm, specifically including the assembly order and assembly direction.

[0018] Step S2.3: Feasible assembly procedures for each component of the product;

[0019] An assembly process library is constructed. Based on the assembly information model, a set of feasible assembly processes for each component of the product to be assembled is generated using a rule engine. The processes in the assembly process library specifically include name, process, tool, automation attribute, cycle time, loading position, and reverse contact type attribute.

[0020] Step S3: The input data is parsed by the rule engine, and after rule matching and execution, candidate workstation division results are generated;

[0021] Step S3.1: Transform the parsed input data into fact objects that the rule engine can recognize, and load them into the working memory of the rule engine;

[0022] The input data includes the cycle time constraints of workstation division, product assembly information model, product assembly sequence, and product assembly method;

[0023] Step S3.2: The rule engine matches all applicable rules based on the loaded fact objects and introduces the Rete algorithm to optimize the rule structure;

[0024] The optimized rule structure is as follows: through incremental inference based on model matching, when a fact object is inserted into the working memory, it propagates along the Rete network. First, it is filtered through all Alpha nodes that match the fact object type, and then it enters the Beta node to connect with existing partial matches from other paths. If the connection is successful, a new and more complete partial match is formed and then reaches the Terminal node. When a partial match propagates to the Terminal node, it means that all the conditions of a rule are met, and the rule is activated.

[0025] Step S3.3: The rule engine calls the solver and enables the pruning strategy to solve for the actions that should be performed after the conditions are met.

[0026] Step S3.4: Repeat steps S3.1-S3.3 until the output of the workstation division results that conform to the process specifications is obtained.

[0027] Step S4: Use the genetic-particle swarm optimization hybrid algorithm to iteratively solve the objective optimization function and output the workstation partitioning optimization results;

[0028] Step S4.1: Configure the parameters of the genetic-particle swarm optimization (GSO) algorithm, including population size and number of iterations. In the GSO algorithm, the population is a set of candidate assembly methods for the product. Integer encoding is used, and each individual represents an assembly method for the product. The assembly method consists of the assembly steps of all parts of the product.

[0029] Step S4.2: Use the rule engine to divide the individuals in the population into workstations;

[0030] Step S4.3: Define and calculate the optimization objective function;

[0031] The optimization objective function is F = w1n wp +w2n p +w3n s ;

[0032] In the formula n wp n p and n s The numbers represent the number of workstations, workers, and tools after partitioning, respectively; w1, w2, and w3 are the weight coefficients of the workstation quantity index, worker quantity index, and tool quantity index, respectively, satisfying w1+w2+w3=1; calculate the optimization objective function for all individuals in the population, update the individual historical best value and the population historical best value, thereby obtaining the individual historical best assembly method and the population historical best assembly method under the conditions of the individual historical best value and the population historical best value;

[0033] Step S4.4: Cross the individual to be crossed with the individual's historical best assembly method;

[0034] Randomly extract a fragment of the individual's historical best assembly method and record the position of the fragment in the individual's historical best assembly method. Delete the element at the corresponding position in the individual to be crossed and insert the extracted fragment into the corresponding position in the individual to be crossed to complete the crossing operation.

[0035] Step S4.5: Cross the individuals to be crossed with the population's historical best assembly method;

[0036] Randomly extract a fragment of the population's historical best assembly method and record the position of the fragment in the population's historical best assembly method. Delete the element at the corresponding position in the individuals to be crossed and insert the extracted fragment into the corresponding position of the individuals to be crossed to complete the crossover operation.

[0037] Step S4.6: Randomly select n parts from the product and change their assembly process;

[0038] Step S4.7: Repeat steps S4.2-S4.6 until the maximum number of iterations is reached, and output the optimized workstation division results, including the process content, automation attributes, workstation cycle time, manual time and tools used for each workstation.

[0039] On the other hand, this application proposes a computer-readable storage medium storing executable instructions that, when executed, cause a processor to perform the aforementioned production line workstation partitioning method based on a rule engine and intelligent optimization algorithm.

[0040] Thirdly, this application proposes a computer program product, including a computer program or instructions, which, when executed by a processor, implements the aforementioned production line workstation division method based on a rule engine and intelligent optimization algorithm.

[0041] The beneficial effects of adopting the above technical solution are as follows:

[0042] This invention provides a production line workstation partitioning method based on a rule engine and intelligent optimization algorithm. The invention uses a rule engine to generate standardized workstation partitioning results, thereby transforming the production line workstation partitioning into a combinatorial optimization problem, and uses a genetic-particle swarm optimization algorithm to iteratively solve it. This can reduce the dependence on the experience and ability of designers in the workstation partitioning process, improve design quality, and shorten the design cycle. Attached Figure Description

[0043] Figure 1 Flowchart of the production line workstation division method based on rule engine and intelligent optimization algorithm in this embodiment of the invention;

[0044] Figure 2 The flowchart of the rule engine generating workstation division results provided in this embodiment of the invention;

[0045] Figure 3The flowchart of the iterative solution of the genetic-particle swarm optimization algorithm provided in this embodiment of the invention. Detailed Implementation

[0046] The specific implementation methods of this application will be further described in detail below with reference to the accompanying drawings and embodiments.

[0047] Example 1:

[0048] A production line workstation partitioning method based on rule engine and intelligent optimization algorithm, such as Figure 1 As shown, it includes the following steps:

[0049] Step S1: Construct a set of business rules for workstation division, including assembly process generation rules, workstation cycle time calculation rules, and workstation update rules.

[0050] The assembly process generation rules are used to generate a set of feasible assembly processes for part A;

[0051] For example in this embodiment: if the assembly relationship of part A is interference, and the assembly axis of other parts that are the same as part A coincides with A, and the assembly direction of other parts that are the same as part A is opposite to A, and the mass of part A is less than the lifting tool use threshold, and the size of part A is less than the lifting tool use threshold, then press fitting method 1 is adopted.

[0052] The station cycle time calculation rule is used to calculate the station cycle time after the current station completes the assembly of part A.

[0053] For example, in this embodiment: if part A is assembled using pressing method 1, and the reference part being pressed is loaded into the current station, and there are no other parts that need to be pressed in the current station, and the automation attribute of the current station is manual or semi-automatic, then the cycle time of the current station after part A is assembled = cycle time before assembly + cycle time of pressing method 1 - cycle time of displacement.

[0054] The station update rule is used to determine the assembly station of part A and update the assembly information of the station.

[0055] The assembly information includes assembly parts, feeding parts, cycle time, tools, labor time, and process content attributes.

[0056] This embodiment includes, for example: if the automation attribute of the press-fitting method used for part A is semi-automatic, and the automation attribute of the current station is semi-automatic, and the cycle time of the current station after assembling part A is greater than the cycle time value of the semi-automatic station, and the cycle time of the current station after assembling part A is less than or equal to the cycle time value of the semi-automatic station plus the press head reset time, then part A can be assembled in the current station.

[0057] Step S2: Parse the source data file to obtain the cycle time constraints of the workstation division, the assembly information model, the assembly sequence, and the set of assembly processes;

[0058] Step S2.1: Product attribute analysis;

[0059] Import the logistics bill of materials (BOM) and process requirement documents. Using the RAG engine, extract the cycle time limits for manual, semi-automatic, and automatic workstations based on prompts to form cycle time constraints for workstation division. In this embodiment, RAGFlow is used, an open-source enterprise-grade RAG (search-enhanced generation) engine developed by the InfiniFlow team.

[0060] Import the product 3D model in the native format of STEP or CAD software, and extract the name, assembly relationship, assembly position, mass, dimensions, process requirements, centroid coordinates, reference parts, coordinate system, and assembly parent node of each component through the CAD parsing library to form an assembly information model; the assembly information model is a list of data dictionaries, which includes the field names, data types, and remarks of the corresponding assembly information.

[0061] In this embodiment, the data dictionary includes the following assembly information:

[0062] Step S2.2: Product assembly sequence planning;

[0063] With the optimization goals of ease of product assembly, reliability, and continuity, an optimized assembly sequence is generated using an assembly information model and a genetic algorithm, specifically including the assembly order and assembly direction.

[0064] Step S2.3: Feasible assembly procedures for each component of the product;

[0065] An assembly process library is constructed. Based on the assembly information model, a set of feasible assembly processes for each component of the product to be assembled is generated using a rule engine. The processes in the assembly process library specifically include name, process, tool, automation attribute, cycle time, loading position, and reverse contact type attribute.

[0066] Currently, Drools, LiteFlow, and Aviator are all mainstream open-source rule engines, with Drools, which is used in this embodiment, having the most mature ecosystem.

[0067] The assembly process library in this embodiment includes, but is not limited to, the following:

[0068] Step S3: The input data is parsed using a rule engine. After rule matching and execution, candidate workstation partitioning results are generated. In this embodiment, for the product assembly method {p1,p2,...,p...} m}, where m is the number of parts in the product, pi This refers to the assembly process of the i-th component in the product assembly sequence, where i ∈ [1, m]. See the detailed process below. Figure 2 .

[0069] Step S3.1: Transform the parsed input data into fact objects that the rule engine can recognize, and load them into the working memory of the rule engine;

[0070] The input data includes the cycle time constraints of workstation division, product assembly information model, product assembly sequence, and product assembly method;

[0071] Step S3.2: The rule engine matches all applicable rules based on the loaded fact objects and introduces the Rete algorithm to optimize the rule structure;

[0072] The optimized rule structure is as follows: By using incremental reasoning based on model matching, the problem of assembly process combination explosion caused by "looping through all rules and checking all facts one by one" during the division of complex product workstations is avoided. When a fact object is inserted into the working memory, it propagates along the Rete network. First, it is filtered through all Alpha nodes that match the fact object type, and then it enters the Beta node to connect with existing partial matches from other paths. If the connection is successful, a new and more complete partial match is formed and then reaches the Terminal node. When a partial match propagates to the Terminal node, it means that all the conditions of a rule are met, and the rule is activated.

[0073] Step S3.3: The rule engine calls the solver and enables the pruning strategy to solve for the actions that should be performed after the conditions are met.

[0074] In this embodiment, the specific action to be performed is whether to place the part's process at the current workstation and update the workstation information, or to create a new workstation.

[0075] Step S3.4: Repeat steps S3.1-S3.3 until the output of the workstation division results that conform to the process specifications is obtained.

[0076] Step S4: Using a hybrid genetic-particle swarm optimization algorithm, iteratively solve the objective function and output the workstation partitioning optimization results. The iterative solution process is detailed below. Figure 3 ;

[0077] Step S4.1: Configure the parameters of the genetic-particle swarm optimization (GSO) algorithm, including population size and number of iterations. In the GSO algorithm, the population is a set of candidate assembly methods for the product. Integer encoding is used, and each individual represents an assembly method for the product. The assembly method consists of the assembly steps of all parts of the product.

[0078] Step S4.2: Use the rule engine to divide the individuals in the population into workstations, as shown in step S3.

[0079] Step S4.3: Define and calculate the optimization objective function;

[0080] The optimization objective function is F = w1n wp +w2n p +w3n s ;

[0081] In the formula n wp n p and n s The numbers represent the number of workstations, workers, and tools after partitioning, respectively; w1, w2, and w3 are the weight coefficients of the workstation quantity index, worker quantity index, and tool quantity index, respectively, satisfying w1+w2+w3=1; calculate the optimization objective function for all individuals in the population, update the individual historical best value and the population historical best value, thereby obtaining the individual historical best assembly method and the population historical best assembly method under the conditions of the individual historical best value and the population historical best value;

[0082] Step S4.4: Cross the individual to be crossed with the individual's historical best assembly method;

[0083] Randomly extract a fragment of the individual's historical best assembly method and record the position of the fragment in the individual's historical best assembly method. Delete the element at the corresponding position in the individual to be crossed and insert the extracted fragment into the corresponding position in the individual to be crossed to complete the crossing operation.

[0084] Step S4.5: Cross the individuals to be crossed with the population's historical best assembly method;

[0085] Randomly extract a fragment of the population's historical best assembly method and record the position of the fragment in the population's historical best assembly method. Delete the element at the corresponding position in the individuals to be crossed and insert the extracted fragment into the corresponding position of the individuals to be crossed to complete the crossover operation.

[0086] Step S4.6: Randomly select n parts from the product and change their assembly process;

[0087] Step S4.7: Repeat steps S4.2-S4.6 until the maximum number of iterations is reached, and output the optimized workstation division results, including the process content, automation attributes, workstation cycle time, manual time and tools used for each workstation.

[0088] Example 2:

[0089] This embodiment proposes a computer-readable storage medium that stores executable instructions. When these instructions are executed, if they are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0090] The computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the production line workstation division method based on rule engine and intelligent optimization algorithm described in various embodiments of this application.

[0091] The aforementioned storage media include: flash memory, hard disk, multimedia card, card-type memory (e.g., SD (Secure Digital Memory Card) or DX (Memory Data Register, MDR) memory, random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic storage, disk, optical disk, server, APP (Application) application store, and other media capable of storing program verification codes. These media store computer programs, and when executed by a processor, they can implement the various steps of the production line workstation division method based on rule engines and intelligent optimization algorithms described above.

[0092] Example 3:

[0093] This embodiment proposes a computer program product, including a computer program or instructions, which, when executed by a processor, implements the aforementioned production line workstation division method based on a rule engine and intelligent optimization algorithm.

[0094] Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a computer program product.

[0095] The various embodiments in this application are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0096] The scope of protection of this application is not limited to the embodiments described above. Obviously, those skilled in the art can make various modifications and variations to this disclosure without departing from the scope and spirit of this disclosure. If such modifications and variations fall within the scope of the methods disclosed herein and their equivalents, then the intent of this disclosure also includes such modifications and variations.

Claims

1. A production line workstation partitioning method based on a rule engine and intelligent optimization algorithm, characterized in that, Includes the following steps: Step S1: Construct a set of business rules for workstation partitioning, including assembly process generation rules, workstation cycle time calculation rules, and workstation update rules; Step S2: Parse the source data file to obtain the cycle time constraints of the workstation division, the assembly information model, the assembly sequence, and the set of assembly processes; Step S3: The input data is parsed by the rule engine, and after rule matching and execution, candidate workstation division results are generated; Step S4: Use the genetic-particle swarm optimization hybrid algorithm to iteratively solve the objective optimization function and output the workstation partitioning optimization results.

2. The production line workstation division method based on rule engine and intelligent optimization algorithm according to claim 1, characterized in that, The assembly process generation rules described in step S1 are used to generate a set of feasible assembly processes for part A; The station cycle time calculation rule is used to calculate the station cycle time after the current station completes the assembly of part A. The station update rule is used to determine the assembly station of part A and update the assembly information of the station; wherein the assembly information includes the assembly parts, the loading parts, the cycle time, the tools, the manual time, and the process content attributes.

3. The production line workstation division method based on rule engine and intelligent optimization algorithm according to claim 1, characterized in that, Step S2 specifically includes the following steps: Step S2.1: Product attribute analysis; Import the logistics list BOM and process requirement documents, and use the RAG engine to extract the manual, semi-automatic and automatic station cycle time limits of the product through prompt words to form the cycle time constraints of the station division. Import the product's 3D model and extract the names, assembly relationships, assembly positions, mass, dimensions, process requirements, centroid coordinates, reference parts, coordinate systems, and assembly parent nodes of each component from the CAD parsing library to form an assembly information model. The assembly information model is a list of data dictionaries, which includes the field names, data types, and remarks corresponding to the assembly information. Step S2.2: Product assembly sequence planning; With the optimization goals of convenience, reliability and continuity of product assembly, an optimized assembly sequence is generated using an assembly information model and a genetic algorithm, specifically including assembly order and assembly direction. Step S2.3: Feasible assembly procedures for each component of the product; An assembly process library is constructed. Based on the assembly information model, a set of feasible assembly processes for each component of the product to be assembled is generated using a rule engine. The specific processes in the assembly process library include name, process, tool, automation attribute, cycle time, loading position, and reverse contact type attribute.

4. The production line workstation division method based on rule engine and intelligent optimization algorithm according to claim 1, characterized in that, Step S3 specifically includes the following steps: Step S3.1: Transform the parsed input data into fact objects that the rule engine can recognize, and load them into the working memory of the rule engine; The input data includes the cycle time constraints of workstation division, product assembly information model, product assembly sequence, and product assembly method; Step S3.2: The rule engine matches all applicable rules based on the loaded fact objects and introduces the Rete algorithm to optimize the rule structure; The optimized rule structure is as follows: through incremental inference based on model matching, when a fact object is inserted into the working memory, it propagates along the Rete network. First, it is filtered through all Alpha nodes that match the fact object type, and then it enters the Beta node to connect with existing partial matches from other paths. If the connection is successful, a new and more complete partial match is formed and then reaches the Terminal node. When a partial match propagates to the Terminal node, it means that all the conditions of a rule are met, and the rule is activated. Step S3.3: The rule engine calls the solver and enables the pruning strategy to solve for the actions that should be performed after the conditions are met; Step S3.4: Repeat steps S3.1-S3.3 until the output of the workstation division results that conform to the process specifications is obtained.

5. The production line workstation division method based on rule engine and intelligent optimization algorithm according to claim 1, characterized in that, Step S4 specifically includes the following steps: Step S4.1: Configure the parameters of the genetic-particle swarm optimization (GSO) algorithm, including population size and number of iterations. In the GSO algorithm, the population is a set of candidate assembly methods for the product. Integer encoding is used, and each individual represents an assembly method for the product. The assembly method consists of the assembly steps of all parts of the product. Step S4.2: Use the rule engine to divide the individuals in the population into workstations; Step S4.3: Define and calculate the optimization objective function; The optimization objective function is F = w1n wp +w2n p +w3n s ; In the formula n wp n p and n s The numbers represent the number of workstations, workers, and tools after partitioning, respectively; w1, w2, and w3 are the weight coefficients of the workstation quantity index, worker quantity index, and tool quantity index, respectively, satisfying w1+w2+w3=1; calculate the optimization objective function for all individuals in the population, update the individual historical best value and the population historical best value, thereby obtaining the individual historical best assembly method and the population historical best assembly method under the conditions of the individual historical best value and the population historical best value; Step S4.4: Cross the individual to be crossed with the individual's historical best assembly method; Randomly extract a fragment of the individual's historical best assembly method and record the position of the fragment in the individual's historical best assembly method. Delete the element at the corresponding position in the individual to be crossed and insert the extracted fragment into the corresponding position in the individual to be crossed to complete the crossing operation. Step S4.5: Cross the individuals to be crossed with the population's historical best assembly method; Randomly extract a fragment of the population's historical best assembly method and record the position of the fragment in the population's historical best assembly method. Delete the element at the corresponding position in the individuals to be crossed and insert the extracted fragment into the corresponding position of the individuals to be crossed to complete the crossover operation. Step S4.6: Randomly select n parts from the product and change their assembly process; Step S4.7: Repeat steps S4.2-S4.6 until the maximum number of iterations is reached, and output the optimized workstation division results, including the process content, automation attributes, workstation cycle time, manual time and tools used for each workstation.

6. A computer-readable storage medium storing executable instructions, characterized in that, When the instruction is executed, it causes the processor to perform the production line workstation division method based on a rule engine and intelligent optimization algorithm as described in any one of claims 1-5.

7. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by the processor, they implement the production line workstation division method based on a rule engine and intelligent optimization algorithm as described in any one of claims 1-5.

Images