Knowledge graph-based factory process parameter intelligent matching method and system

Abstract

The invention provides a factory process parameter intelligent matching method and system based on a knowledge graph, and the method comprises the steps: obtaining a process information set, carrying out the knowledge association analysis of the process information set, and generating a process knowledge association structure; extracting a process correlation feature group from the process knowledge correlation structure, wherein the process correlation feature group comprises operation parameter correlation features and parameter collaboration features; executing a process parameter matching reasoning operation based on the process associated feature group, and generating a candidate parameter matching scheme of a target production stage; matching rationality verification is carried out on the candidate parameter matching scheme, a final process parameter matching result of the target production stage is generated, the production abnormity risk caused by new parameter configuration is reduced, and a reliable basis is provided for factory intelligent decision making.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, in particular to a factory process parameter intelligent matching method and system based on a knowledge graph. BACKGROUND

[0002] In the field of industrial intelligence, factory process parameter matching technology is a key link to realize production process optimization. The traditional method mainly relies on manual experience summary or fixed rule base for parameter configuration, extracts key parameter set points by manually analyzing operation manuals, and establishes fixed parameter matching rules combined with historical configuration records. However, due to the dynamic coupling characteristics of the operation process and parameter setting in the production process which are not structured modeled, the parameter matching process cannot effectively capture the parameter coordination relationship across processes; at the same time, the existing scheme is limited to simple rule matching for historical production data, which is difficult to cope with complex production scenarios with multiple variable linkages. The parameter configuration scheme generated under this mechanism often has the contradiction between single parameter local optimization and system imbalance, and is easy to cause process fluctuations when the equipment is linked or the material characteristics change, which becomes a technical bottleneck restricting the intelligent upgrading of the factory. SUMMARY

[0003] The present application provides a factory process parameter intelligent matching method and system based on a knowledge graph.

[0004] In a first aspect, the present application embodiment provides a factory process parameter intelligent matching method based on a knowledge graph, comprising: obtaining a process information set, the process information set containing operation process descriptions of different production stages and parameter setting records of corresponding stages; performing knowledge association analysis on the process information set to generate a process knowledge association structure; extracting a process association feature group from the process knowledge association structure, the process association feature group containing operation parameter association features and parameter coordination features; performing process parameter matching reasoning operation based on the process association feature group to generate a candidate parameter matching scheme for the target production stage; performing matching reasoning verification on the candidate parameter matching scheme to generate the final process parameter matching result of the target production stage.

[0005] In a second aspect, the present application embodiment provides a computer system, comprising: a memory, the memory storing a computer program; a processor for loading the computer program to realize the factory process parameter intelligent matching method based on a knowledge graph as described above.

[0006] The factory process parameter intelligent matching method provided by the application can realize semantic correlation relationship modeling between operation flow description and parameter setting record by converting fragmented process information set into machine understandable process knowledge correlation structure, eliminate knowledge fragmentation problem caused by artificial experience dependence; the process correlation feature group constructed based on operation parameter correlation feature and parameter collaboration feature can simultaneously capture single parameter operation dependence and multi-parameter linkage rule in parameter matching reasoning process, significantly improve adaptability of candidate parameter matching scheme in complex production environment, and effectively prevent system imbalance risk caused by parameter isolated optimization; the matching rationality verification mechanism based on historical production record is introduced, cross-time and space experience comparison is conducted on the candidate scheme by using verification knowledge correlation structure, so that the final process parameter matching result has current production demand adaptability and historical operation stability verification, reduces production abnormal risk caused by new parameter configuration, and provides reliable basis for factory intelligent decision. BRIEF DESCRIPTION OF DRAWINGS

[0007] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0008] Figure 1 is a flow chart of a factory process parameter intelligent matching method based on a knowledge graph provided by an embodiment of the present application.

[0009] Figure 2 is a composition schematic diagram of a computer system provided by an embodiment of the present application. DETAILED DESCRIPTION

[0010] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments only constitute some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0011] Please refer to Figure 1 , Figure 1 is a flow chart of a factory process parameter intelligent matching method based on a knowledge graph provided by an embodiment of the present application. The method can be executed by a computer system, and includes the following steps: Step S100: acquire a process information set, the process information set contains operation flow description of different production stages and parameter setting record of corresponding stages.

[0012] The process information set is a collection of information at different production stages in the factory production process, wherein the operation flow description is a detailed description of the specific operation steps of each production stage, which clearly shows the sequence and specific content of each operation in the production process, and the parameter setting record records the parameters required for each production stage operation flow, for example, in the production process of mechanical parts processing, the operation flow description may include raw material cutting, part rough machining, part finishing, etc., and the parameter setting record may include cutting speed, cutting depth, feed amount, etc. For the operation flow description, the factory will have detailed standard operation procedure documents, which can be directly extracted as information sources. At the same time, the actual operation process of workers can also be observed and recorded and sorted. For the parameter setting record, it can be obtained from the control system of the production equipment. Modern production equipment has data recording function, which can record various parameters in the equipment running process in real time, and these records can be exported through data interface.

[0013] Step S200: knowledge correlation analysis is performed on the process information set to generate a process knowledge correlation structure, which includes an association relationship identifier between the operation flow and the parameter setting.

[0014] Knowledge correlation analysis is a process of in-depth analysis and processing of the process information set, aiming to mine the internal correlation between the operation flow description and the parameter setting record. The process knowledge correlation structure is a structured representation form, which clearly shows the correspondence between the operation flow and the parameter setting through the association relationship identifier, so that the connection between the operations and parameters of different production stages is clearer and more explicit. The association relationship identifier can be a code or a label, which is used to uniquely identify the association between the operation flow and the parameter setting, facilitating query and use.

[0015] As an implementation, step S200, knowledge correlation analysis is performed on the process information set to generate a process knowledge correlation structure, which can be implemented as steps S210-S250 as follows: Step S210: semantic decomposition is performed on the operation flow description in the process information set to obtain an operation element set containing operation step identifiers and operation target identifiers.

[0016] Semantic decomposition is to analyze and process the natural language text in the operation flow description to extract key semantic information. The operation step identifier is a unique identification identifier for each specific step in the operation flow, which can accurately distinguish different operation steps. The operation target identifier is a unique identifier for the target expected to be achieved by each operation step, reflecting the purpose and intention of the operation. The operation element set is a set composed of operation step identifiers and operation target identifiers, which summarizes the key elements in the operation flow and provides a basis for subsequent analysis and processing.

[0017] As an implementation form, in step S210, the operation flow description in the process information set is semantically decomposed to obtain an operation element set containing operation step identifiers and operation target identifiers, which can be implemented as steps S211-S215 as follows: Step S211: performing natural language word segmentation on the operation flow description to obtain a word segmentation unit set containing operation verbs and operation objects.

[0018] Natural language word segmentation is to divide the text of the operation flow description according to the boundaries of words, and decompose it into independent words. The word segmentation unit set is a set composed of words containing operation verbs and operation objects. The operation verb specifies the specific operation behavior, and the operation object indicates the object to which the operation is directed. For example, in the operation flow description of “using a lathe to perform rough machining on steel”, after natural language word segmentation, the word segmentation unit set may contain “use” (operation verb), “lathe” (operation object), “on”, “steel” (operation object), “perform”, “rough machining” (operation verb), etc.

[0019] Natural language word segmentation can be implemented using various word segmentation algorithms, such as rule-based word segmentation algorithms and statistical-based word segmentation algorithms. Rule-based word segmentation algorithms perform word segmentation according to predefined rules, such as matching words in a dictionary and dividing when a matching word is encountered. Statistical-based word segmentation algorithms statistically determine the boundaries of words by learning from a large amount of text data and according to the probability of word occurrence. In actual applications, appropriate word segmentation algorithms can be selected according to specific requirements and data characteristics.

[0020] Step S212: mapping the word segmentation unit set to step identifiers based on a preset operation step dictionary, mapping the word segmentation units containing operation verbs and operation objects to operation step identifiers.

[0021] The preset operation step dictionary is a predefined dictionary that contains various possible operation steps and their corresponding identifiers. Step identifier mapping is to match the word segmentation units containing operation verbs and operation objects in the word segmentation unit set with the entries in the operation step dictionary, and map the successfully matched word segmentation units to the corresponding operation step identifiers. Through this mapping, the operation steps described in natural language can be converted into identifiers that can be recognized and processed by computers, facilitating subsequent analysis and processing.

[0022] As an implementation form, in step S212, the word segmentation unit set is mapped to step identifiers based on a preset operation step dictionary, which can be implemented as steps S2121-S2124 as follows: Step S2121: Perform part-of-speech tagging processing on the set of segmented units, identify core segmented units containing operation verbs and operation objects, remove segmented units irrelevant to the production process according to the preset production art language filtering rules, and generate a set of standardized operation units.

[0023] Part-of-speech tagging processing is to tag the part-of-speech of each segmented unit in the set of segmented units, such as noun, verb, adjective, etc. Through part-of-speech tagging, core segmented units containing operation verbs and operation objects can be more accurately identified. The preset production art language filtering rules are rules defined in advance according to the characteristics and requirements of the production process, which are used to filter segmented units related to the production process and remove those irrelevant to the production process, such as mood words, conjunctions, etc. The set of standardized operation units is obtained after part-of-speech tagging and filtering processing, and the segmented units in the set are all core units related to the production process and have a unified format and standard, which is convenient for subsequent processing and analysis.

[0024] Part-of-speech tagging processing can be implemented using part-of-speech tagging tools, which are based on machine learning algorithms and learn from a large amount of text data to perform part-of-speech tagging. For example, part-of-speech tagging tools such as NLTK (Natural Language Toolkit), HanLP, etc. The preset production art language filtering rules can be customized according to specific production processes, for example, defining a production art language dictionary that only retains segmented units in the set of segmented units that match terms in the dictionary.

[0025] Step S2122: Perform bidirectional semantic matching on the set of standardized operation units using the operation step dictionary, directly map the standardized operation units to operation step identifiers when they are identical, calculate the semantic similarity value between the standardized operation units and the dictionary entries when there is a semantic deviation, and filter dictionary entries with a similarity value exceeding a preset baseline value for mapping, and generate an initial set of operation step identifiers.

[0026] Bidirectional semantic matching refers to matching from the standardized operation units to the operation step dictionary entries and from the operation step dictionary entries to the standardized operation units simultaneously. When the standardized operation units are identical to the entries in the operation step dictionary, they are directly mapped to the corresponding operation step identifiers. When there is a semantic deviation, the semantic similarity value between the standardized operation units and the dictionary entries needs to be calculated. Semantic similarity value is an index that measures the degree of semantic similarity between two words or phrases, which can be calculated using various methods, such as word vector-based similarity calculation method, semantic network-based similarity calculation method, etc. The preset baseline value is a threshold value set in advance to filter dictionary entries with high semantic similarity. The initial set of operation step identifiers is obtained after bidirectional semantic matching and filtering, which contains the operation step identifiers obtained through preliminary mapping.

[0027] The similarity calculation method based on word vector represents words as vectors, and measures the semantic similarity of words by calculating the cosine value between vectors. The similarity calculation method based on semantic network calculates the similarity according to the position and relationship of words in the semantic network. For example, the Word2Vec algorithm is used to convert words into word vectors, and then the cosine value between the word vectors of the standardized operation unit and the dictionary entry is calculated. The dictionary entry with a cosine value greater than a preset reference value is taken as the matching result.

[0028] Step S2123: Perform context logic verification on the initial operation step identifier set according to the preset process logic constraint rule, identify the logical coherence feature between the current operation step identifier and the adjacent operation step identifier, perform semantic correction on the operation step identifier based on the process logic constraint rule when a logical conflict is detected, and generate a corrected operation step identifier set.

[0029] The preset process logic constraint rule is a rule defined in advance according to the logic and requirements of the production process, which is used to verify the logical coherence between operation steps. The context logic verification is to check each operation step identifier in the initial operation step identifier set, analyze the logical relationship between it and the adjacent operation step identifier, and judge whether it conforms to the process logic constraint rule. The logical coherence feature includes the sequence of operation steps, causal relationship, etc. When a logical conflict is detected, the operation step identifier needs to be corrected according to the process logic constraint rule to ensure the correct logical relationship between operation steps. The corrected operation step identifier set is the set obtained after context logic verification and semantic correction, and the operation step identifiers in the set have good logical coherence.

[0030] For example, in the production process of mechanical part processing, the process logic constraint rule stipulates that rough machining should be performed before finish machining. If the finish machining step appears before the rough machining step in the initial operation step identifier set, a logical conflict is detected, and the order of the operation step identifier is adjusted according to the process logic constraint rule to conform to the normal production logic.

[0031] Step S2124: Perform sequence reorganization on the corrected operation step identifier set according to the time sequence relationship of the production process, establish the time dependency relationship chain between the operation step identifiers, and generate an operation step identifier sequence.

[0032] The time sequence relationship of the production procedure refers to the time sequence relationship of each operation step in the production process. The sequence reorganization is to reorder the operation step identifiers in the modified operation step identifier set according to the time sequence relationship of the production procedure, so that they are arranged in the correct time sequence. The time dependency relationship chain is a structure for representing the time sequence and dependency relationship between the operation step identifiers. By establishing the time dependency relationship chain, the position and role of each operation step in the production process can be determined. The operation step identifier sequence is an ordered set obtained after sequence reorganization, reflecting the actual execution sequence of the operation steps in the production process.

[0033] For example, in the production process of mechanical part processing, the operation step identifier sequence according to the time sequence relationship of the production procedure may be raw material cutting, part rough machining, part finishing, etc. These operation step identifiers are arranged in time sequence to form a complete production process.

[0034] Step S213: Based on the preset operation target dictionary, the target identifier mapping is performed on the word segmentation unit set to map the word segmentation unit containing the operation purpose description to the operation target identifier.

[0035] The preset operation target dictionary is a predefined dictionary containing various possible operation targets and their corresponding identifiers. The target identifier mapping is to match the word segmentation unit containing the operation purpose description in the word segmentation unit set with the entries in the operation target dictionary, and map the successfully matched word segmentation unit to the corresponding operation target identifier. Through this mapping, the natural language description of the operation target can be converted into an identifier that can be recognized and processed by a computer, facilitating subsequent analysis and processing.

[0036] For example, in the operation process of mechanical part processing, the operation purpose description may be "improve the precision of the part", which is mapped to the corresponding operation target identifier such as "high-precision processing target" through matching with the preset operation target dictionary.

[0037] Step S214: De-duplicate the operation step identifier and the operation target identifier, and retain the operation step identifier and the operation target identifier with unique identifier.

[0038] De-duplication refers to removing duplicate identifiers in the operation step identifier and operation target identifier set, and only retaining elements with unique identifiers. In actual processing, due to the complexity and diversity of operation process description, there may be duplicate operation step identifiers and operation target identifiers, which will increase the complexity of subsequent processing and may lead to inaccurate results. Through de-duplication processing, the quality and processing efficiency of the data can be improved. De-duplication can be achieved by comparing the string values of the identifiers, and only retaining one of the same identifiers.

[0039] Step S215: combining the removed duplicate operation step identifiers and operation target identifiers to obtain an operation element set.

[0040] The deduplicated operation step identifiers and operation target identifiers are combined to form a unified operation element set. This set contains the key elements of the operation process, namely the operation steps and operation targets.

[0041] Step S220: extracting parameter attributes from the parameter setting records in the process information set to obtain a parameter element set including a parameter type identifier and a parameter scope identifier.

[0042] Parameter attribute extraction analyzes and processes parameter setting records within the process information set to extract key parameter attributes. The parameter type identifier uniquely identifies the parameter type, such as temperature, pressure, or speed. The parameter scope identifier identifies the scope of a parameter's effect within the production process, for example, whether a parameter affects the entire production process or only a specific production step. A parameter element set, consisting of a parameter type identifier and a parameter scope identifier, summarizes the key attribute information within the parameter setting record.

[0043] For example, in the parameter setting record of mechanical parts processing, the parameter type identifier may include cutting speed, cutting depth, etc., and the parameter scope identifier may indicate whether these parameters are effective in the rough processing stage or the fine processing stage.

[0044] Parameter attribute extraction can be achieved by parsing and analyzing parameter setting records. First, the parameter type identifier is determined based on the parameter name and description. Then, the parameter scope identifier is determined based on the operation process description and production process characteristics. For example, if the parameter setting record clearly states that a parameter is used during the rough machining process, its parameter scope identifier can be determined as "rough machining stage."

[0045] Step S230: establishing a first association relationship between the operation step identifier in the operation element set and the parameter type identifier in the parameter element set, where the first association relationship is used to represent the parameter type corresponding to the preset operation step.

[0046] Establishing the first association relationship involves associating the operation step identifiers in the operation element set with the parameter type identifiers in the parameter element set, clarifying the parameter type corresponding to each operation step. Preset operation steps refer to the individual operation steps predefined in the operation process. The establishment of the first association relationship can be achieved by analyzing the inherent connection between the operation process description and the parameter setting record. For example, in the production process of mechanical parts processing, the operation step "parts rough processing" may correspond to parameter types such as "cutting speed" and "cutting depth".

[0047] The first association relationship can be established in various ways. One way is to match rules to determine the association between operation step identifiers and parameter type identifiers according to predefined rules. For example, the rule "if the operation step is part machining, then the corresponding parameter types include cutting speed and cutting depth" is defined. Another way is to use a machine learning algorithm to learn from a large number of operation process descriptions and parameter setting records to automatically discover the association between operation steps and parameter types.

[0048] Step S240: Establish a second association relationship between operation target identifiers in the operation element set and parameter scope identifiers in the parameter element set, and the second association relationship is used to represent the parameter scope corresponding to the preset operation target.

[0049] The second association relationship is established by associating operation target identifiers in the operation element set with parameter scope identifiers in the parameter element set, and explicitly specifying the parameter scope corresponding to each operation target. The preset operation target refers to each operation target predefined in the operation process. The establishment of the second association relationship helps better understand the relationship between operation targets and parameters, and provides a basis for reasonable parameter setting. For example, in the production process of mechanical part machining, the operation target "improve part precision" may correspond to the parameter scope of the "finishing stage".

[0050] The method of establishing the second association relationship is similar to that of establishing the first association relationship, and can use rule matching or a machine learning algorithm. By analyzing the relationship between operation target descriptions and parameter setting records, the association between operation target identifiers and parameter scope identifiers is determined. For example, according to the rule "if the operation target is to improve part precision, then the corresponding parameter scope is the finishing stage".

[0051] Step S250: Construct a process knowledge association structure based on the first association relationship and the second association relationship, and the process knowledge association structure represents the association between operation elements and parameter elements through node-edge connection.

[0052] The process knowledge association structure is a graph structure that represents the association between operation elements and parameter elements through nodes and edges. Nodes represent operation elements (operation step identifiers and operation target identifiers) and parameter elements (parameter type identifiers and parameter scope identifiers), and edges represent their association, i.e., the first association relationship and the second association relationship. Through this graph structure, the complex association between operation processes and parameter settings can be intuitively displayed, facilitating query, analysis and reasoning.

[0053] The construction of the process knowledge association structure can be implemented using a graph database, such as Neo4j, JanusGraph, etc. The operation elements and parameter elements are added as nodes to the graph database, and then edges are created according to the first association relationship and the second association relationship to connect the corresponding nodes. For example, a node representing the "rough machining of parts" operation step identifier is created in the graph database, and another node representing the "cutting speed" parameter type identifier is created, and an edge is created to connect the two nodes according to the first association relationship, indicating the association between them.

[0054] Step S300: Extracting a process association feature group from the process knowledge association structure, the process association feature group including operation parameter association features and parameter synergy features.

[0055] The process association feature group is a set of key features extracted from the process knowledge association structure, used to describe the association between the operation process and the parameter setting. The operation parameter association features reflect the association strength and distribution between different operation steps and parameter types, and the parameter synergy features reflect the association strength and distribution between different operation targets and parameter action ranges. By extracting the process association feature group, the mutual relationship between the operation and the parameter in the production process can be better understood, providing an important basis for process parameter matching reasoning.

[0056] As an implementation, step S300, extracting a process association feature group from the process knowledge association structure, can be implemented as steps S310-S340 as follows: Step S310: Quantifying the feature of the node-edge connection relationship in the process knowledge association structure, calculating the first association strength value between the operation step identifier and the parameter type identifier, and calculating the second association strength value between the operation target identifier and the parameter action range identifier.

[0057] Feature quantification is to quantify the association relationship represented by the node-edge connection relationship in the process knowledge association structure, using specific numerical values to represent the strength of the association. The first association strength value is used to measure the degree of association between the operation step identifier and the parameter type identifier, and the second association strength value is used to measure the degree of association between the operation target identifier and the parameter action range identifier. By calculating these two association strength values, the association between the operation and the parameter can be more accurately described.

[0058] As an implementation, step S310, quantifying the feature of the node-edge connection relationship in the process knowledge association structure, calculating the first association strength value between the operation step identifier and the parameter type identifier, can be implemented as steps S311-S316 as follows: Step S311: Counting the number of connection edges between the operation step identifier and the parameter type identifier in the process knowledge association structure.

[0059] The number of connection edges is an index for measuring the closeness of the association between the operation step identifier and the parameter type identifier. In the process knowledge association structure, the connection edge represents that there is an association relationship between the operation step identifier and the parameter type identifier. The number of connection edges can intuitively understand the association between the operation step and the parameter type. The more the number of connection edges, the closer the association between them. For example, in the process knowledge association structure of mechanical part machining, there may be multiple connection edges between the “part rough machining” operation step identifier and the “cutting speed” parameter type identifier, which indicates that the cutting speed parameter is closely related to the operation step in the rough machining process.

[0060] Step S312: Count the total number of connection edges corresponding to the operation step identifier in the process knowledge association structure.

[0061] The total number of connection edges corresponding to the operation step identifier reflects the association between the operation step and all parameter type identifiers. It is an index for measuring the importance of the operation step in the entire process knowledge association structure. By counting the total number of connection edges, the extensive association degree between the operation step and other parameter types can be understood.

[0062] For example, the “part rough machining” operation step identifier may have an association with multiple parameter type identifiers, and counting the total number of connection edges corresponding to it can obtain a comprehensive association information.

[0063] Step S313: Calculate the ratio of the number of connection edges to the total number of connection edges to obtain the initial association strength value between the operation step identifier and the parameter type identifier.

[0064] The initial association strength value is calculated by the ratio of the number of connection edges to the total number of connection edges, which reflects the relative association strength between the operation step identifier and a certain parameter type identifier. The larger this ratio, the stronger the association between the operation step and the parameter type.

[0065] For example, if the number of connection edges between the “part rough machining” operation step identifier and the “cutting speed” parameter type identifier is 5, and the total number of connection edges corresponding to the “part rough machining” operation step identifier is 10, then the initial association strength value between them is 5 / 10=0.5.

[0066] Step S314: Obtain the historical execution times of the operation step identifier and the historical usage times of the parameter type identifier in the process information set.

[0067] The operation step identifier corresponding historical execution times refer to the total number of times the operation step has been executed in past production processes, which reflects the frequency of the operation step in production. The parameter type identifier corresponding historical usage times refer to the total number of times the parameter type has been used in past production processes, which reflects the importance and commonality of the parameter type in production.

[0068] These historical data can be obtained by querying the production record database of the factory. For example, in the production record of mechanical part processing, the execution times of the "rough machining of parts" operation step and the usage times of the "cutting speed" parameter type can be queried.

[0069] Step S315: Calculate the product of the historical execution times and the historical usage times to obtain the association weight correction value.

[0070] The association weight correction value is obtained by multiplying the operation step identifier corresponding historical execution times and the parameter type identifier corresponding historical usage times. This value takes into account the use of operation steps and parameters in historical production processes, and is used to correct the initial association strength value. The more historical execution times and historical usage times, the more important the operation step and parameter in production, and the greater the association weight correction value.

[0071] Step S316: Multiply the initial association strength value and the association weight correction value to obtain the first association strength value between the operation step identifier and the parameter type identifier.

[0072] Multiplying the initial association strength value and the association weight correction value obtains the final first association strength value. This value takes into account the direct association strength between operation steps and parameter types and their use in historical production processes, and more accurately reflects the actual association degree between operation steps and parameter types.

[0073] Step S320: Construct an operation parameter association feature based on the first association strength value, which contains the parameter type association strength distribution corresponding to different operation steps.

[0074] The operation parameter association feature is constructed according to the first association strength value, which describes the association strength distribution between different operation steps and parameter types. By analyzing the operation parameter association feature, we can understand which parameter types are strongly associated with each operation step and which are weakly associated, thereby providing a basis for reasonable parameter setting.

[0075] For example, in mechanical part processing, the operation parameter association feature may show that the "rough machining of parts" operation step has a high association strength with the "cutting speed" and "cutting depth" parameter types, and a low association strength with other parameter types.

[0076] The operation parameter correlation feature can arrange and sort the first correlation strength values according to the operation step identifiers and the parameter type identifiers, forming a correlation strength distribution matrix. The rows of the matrix represent operation step identifiers, the columns represent parameter type identifiers, and the elements in the matrix represent corresponding first correlation strength values.

[0077] Step S330: Construct a parameter synergy feature based on the second correlation strength values, the parameter synergy feature containing parameter action range correlation strength distributions corresponding to different operation targets.

[0078] The parameter synergy feature is constructed according to the second correlation strength values, which describes the correlation strength distribution between different operation targets and parameter action ranges. By analyzing the parameter synergy feature, it can be understood which operation targets are more strongly associated with which parameter action ranges, and which are more weakly associated, which helps to reasonably select and set parameters when achieving operation targets.

[0079] For example, in mechanical part processing, the parameter synergy feature may show that the "improve part precision" operation target has a higher correlation strength with the "finishing stage" parameter action range, and a lower correlation strength with other parameter action ranges.

[0080] The method of constructing the parameter synergy feature is similar to that of constructing the operation parameter correlation feature. The second correlation strength values are arranged and sorted according to the operation target identifiers and the parameter action range identifiers, forming a correlation strength distribution matrix.

[0081] Step S340: Combine the operation parameter correlation feature and the parameter synergy feature to obtain a process correlation feature group.

[0082] The operation parameter correlation feature and the parameter synergy feature are combined together to form a process correlation feature group. This group contains comprehensive correlation information between the operation process and the parameter settings, and is an important basis for subsequent process parameter matching reasoning.

[0083] Combination can be achieved by merging the operation parameter correlation feature matrix and the parameter synergy feature matrix to form a larger feature matrix. This matrix integrates the correlation information between operation steps, operation targets, parameter types, and parameter action ranges.

[0084] Step S400: Perform process parameter matching reasoning operation based on the process correlation feature group to generate candidate parameter matching schemes for the target production stage.

[0085] The process parameter matching reasoning operation is to infer a suitable parameter matching scheme for the target production stage according to the association information in the process association feature group. The target production stage refers to the specific stage that needs to be produced at present, such as the rough machining stage and the finishing machining stage in mechanical part machining. The candidate parameter matching scheme is a set of parameter setting schemes that may be suitable for the target production stage, providing a reference for actual production.

[0086] As an implementation manner, step S400, a process parameter matching reasoning operation is performed based on the process association feature group to generate a candidate parameter matching scheme for the target production stage, which can be specifically implemented as steps S410-S450: Step S410: Obtain a target operation flow description of the target production stage, and perform semantic decomposition on the target operation flow description to obtain a target operation element set, which contains a target operation step identifier and a target operation target identifier.

[0087] The target operation flow description is a detailed description of the specific operation flow of the target production stage, which is the basis for parameter matching reasoning. Semantic decomposition is an analysis and processing of the target operation flow description to extract key information therefrom. The target operation element set is a set composed of a target operation step identifier and a target operation target identifier, which summarizes the key elements in the operation flow of the target production stage.

[0088] For example, when the target production stage of mechanical part machining is "part finishing", the target operation flow description can include steps such as "using high-precision tools for fine cutting", and through semantic decomposition, the "part finishing" target operation step identifier and the "improve part precision" target operation target identifier can be obtained.

[0089] The method of semantic decomposition is similar to the method of semantic decomposition of the operation flow description in step S210, including steps such as natural language word segmentation, part-of-speech tagging, step identifier mapping, and target identifier mapping.

[0090] Step S420: Extract operation parameter association features corresponding to the target operation step identifier from the process association feature group.

[0091] According to the target operation step identifier, operation parameter association features corresponding thereto are extracted from the process association feature group. These features reflect the association strength distribution between the target operation step and each parameter type, providing a basis for determining the parameter type suitable for the operation step.

[0092] For example, for the "part finishing" target operation step identifier, the association strength information of the parameter types such as "cutting speed" and "cutting depth" is extracted from the process association feature group.

[0093] Step S430: extracting, from the process-related feature group, a parameter synergy feature corresponding to the target operation target identifier.

[0094] According to the target operation target identifier, a parameter synergy feature corresponding thereto is extracted from the process-related feature group. These features reflect the correlation strength distribution between the target operation target and the parameter action range, providing a basis for determining the parameter action range suitable for the operation target.

[0095] For example, for the "improve part precision" target operation target identifier, the correlation strength information between it and the "finishing stage" parameter action range is extracted from the process-related feature group.

[0096] Step S440: feature fusion of the operation parameter correlation feature and the parameter synergy feature to generate a target correlation feature combination.

[0097] Feature fusion is the integration of operation parameter correlation features and parameter synergy features to obtain a comprehensive feature representation. The target correlation feature combination is a feature set obtained after feature fusion, which contains comprehensive correlation information between operation steps, operation targets, parameter types and parameter action ranges, and is more conducive to process parameter matching reasoning.

[0098] As an implementation, step S440, feature fusion of the operation parameter correlation feature and the parameter synergy feature to generate a target correlation feature combination, can be implemented as steps S441-S444 as follows: Step S441: feature dimension alignment of the operation parameter correlation feature to make the parameter type correlation strength distribution corresponding to different operation step identifiers have the same dimension; and feature dimension alignment of the parameter synergy feature to make the parameter action range correlation strength distribution corresponding to different operation target identifiers have the same dimension.

[0099] Feature dimension alignment is a step of preprocessing the operation parameter correlation feature and the parameter synergy feature, the purpose of which is to make the features corresponding to different operation steps and operation targets have the same dimension, facilitating subsequent feature fusion. Since different operation steps and operation targets may be associated with different numbers of parameter types and parameter action ranges, the dimensions of the feature matrices are inconsistent. Through feature dimension alignment, the dimensions of the feature matrices can be unified.

[0100] For example, for the operation parameter correlation feature matrix, some operation steps may be associated with 5 parameter types, while some operation steps may be associated with 3 parameter types. Through feature dimension alignment, the dimensions of the feature matrices corresponding to all operation steps are unified to the same number.

[0101] The feature dimension alignment can be achieved by padding or cropping. For a feature matrix with insufficient dimensions, it can be padded with default values; for a feature matrix with excessive dimensions, it can be cropped.

[0102] Step S442: Based on the preset feature fusion rule, the dimension-aligned operation parameter relevance features and parameter synergy features are added dimension by dimension to generate preliminary fusion features.

[0103] The preset feature fusion rule is a rule predefined for fusing operation parameter relevance features and parameter synergy features. The dimension-by-dimension addition is an addition operation on each dimension of the two dimension-aligned feature matrices to obtain a new feature matrix, i.e., preliminary fusion features. For example, the elements of the operation parameter relevance feature matrix and the parameter synergy feature matrix on each corresponding dimension are added to obtain a preliminary fusion feature matrix.

[0104] The preset feature fusion rule can be defined according to specific requirements and application scenarios. For example, different weights can be set for the operation parameter relevance features and the parameter synergy features for weighted addition.

[0105] Step S443: Feature normalization is performed on the preliminary fusion features to adjust the feature value range to a preset normalization interval.

[0106] Feature normalization is to adjust the feature value range of the preliminary fusion features to a preset normalization interval, such as [0, 1]. The purpose of normalization is to eliminate the dimensional differences between different features, making the feature values comparable, and facilitating subsequent feature screening and reasoning.

[0107] Feature normalization can be achieved using various methods, such as min-max normalization, Z-score normalization, etc. Min-max normalization is to linearly map the feature values to the [0, 1] interval.

[0108] Step S444: Feature screening is performed on the normalized preliminary fusion features to retain dimensions with feature values exceeding a preset threshold to generate a target association feature combination.

[0109] Feature screening is to select important feature dimensions from the normalized preliminary fusion features, retain dimensions with feature values exceeding a preset threshold, and remove dimensions with small feature values that have little impact on the reasoning results. The preset threshold is a pre-set value for judging the importance of feature dimensions. The target association feature combination is a feature set obtained after feature screening, which only contains important feature dimensions, reduces feature redundancy, and improves the efficiency and accuracy of reasoning.

[0110] As an implementation, in step S444, feature screening is performed on the normalized preliminary fusion features, which can be implemented as steps S4441-S4445 as follows: Step S4441: Obtain the correlation record data of different dimension features and key quality indicators in the historical production records, calculate the statistical correlation indicators of the dimension feature values and the quality indicator values, and generate feature importance weight distribution data.

[0111] The historical production records contain correlation record data of different dimension features and key quality indicators, which reflect the influence of each feature dimension on product quality. The statistical correlation indicators are indicators for measuring the correlation between the feature dimension values and the quality indicator values, such as Pearson correlation coefficient, etc. By calculating the statistical correlation indicators, the importance weight of each feature dimension can be obtained, and the feature importance weight distribution data can be generated.

[0112] For example, in mechanical part processing, the key quality indicator may be the precision of the part. By analyzing the correlation between different parameter types and the precision of the part in the historical production records, the feature importance weight corresponding to each parameter type can be obtained.

[0113] Step S4442: Perform weighted enhancement on the preliminary fusion features according to the feature importance weight distribution data, multiply each dimension value of the preliminary fusion features by the feature importance weight value of the corresponding dimension, and generate a weighted feature vector.

[0114] Weighted enhancement is a step of processing the preliminary fusion features according to the feature importance weight distribution data, multiplying each dimension value of the preliminary fusion features by the feature importance weight value of the corresponding dimension, and obtaining a weighted feature vector. Through weighted enhancement, the feature dimensions that have a greater impact on product quality can be highlighted, and the discrimination of the features can be improved.

[0115] Step S4443: Calculate the arithmetic mean and standard deviation of all dimension values of the weighted feature vector, and construct a dynamic screening reference value based on the arithmetic mean and the standard deviation.

[0116] The arithmetic mean is the sum of all dimension values of the weighted feature vector divided by the number of dimensions, which reflects the average level of the weighted feature vector. The standard deviation is an indicator for measuring the degree of deviation of each dimension value in the weighted feature vector from the arithmetic mean. The dynamic screening reference value is a reference value for screening feature dimensions constructed according to the arithmetic mean and the standard deviation, which will be dynamically adjusted according to the actual situation of the weighted feature vector.

[0117] Step S4444: Perform dimension filtering on the weighted feature vector according to the dynamic screening reference value, retain the feature dimensions with weighted values exceeding the dynamic screening reference value, and generate key features.

[0118] Dimension filtering is a step of filtering the weighted feature vector according to a dynamic filtering benchmark value, only retaining the feature dimensions with a weighted value exceeding the dynamic filtering benchmark value, and removing those feature dimensions with a lower weighted value. Through dimension filtering, the redundancy of the features can be further reduced to obtain key features.

[0119] Step S4445: performing feature vector reconstruction on the key features, recombining the feature values of the retained dimensions in the original feature dimension order to generate an optimized target associated feature combination.

[0120] Feature vector reconstruction is a step of post-processing the key features, recombining the retained feature dimensions in the original feature dimension order to obtain an optimized target associated feature combination. Since the order of the feature dimensions may be disturbed during the dimension filtering process, the original order of the features can be restored through feature vector reconstruction, facilitating subsequent use.

[0121] Step S450: performing parameter matching reasoning based on the target associated feature combination to generate a candidate parameter matching scheme containing a parameter type recommendation identifier and a parameter action range recommendation identifier.

[0122] Parameter matching reasoning is reasoning suitable parameter types and parameter action ranges for the target production stage according to the associated information in the target associated feature combination. The parameter type recommendation identifier is an identifier of a recommended parameter type suitable for the target production stage, and the parameter action range recommendation identifier is an identifier of a recommended parameter action range for the target production stage. The candidate parameter matching scheme is a scheme composed of the parameter type recommendation identifier and the parameter action range recommendation identifier, providing candidate choices for parameter setting of the target production stage.

[0123] For example, when the target production stage of mechanical part processing is "part finishing", according to the target associated feature combination, the recommended parameter type identifier can include "high-precision cutting speed", "small cutting depth", etc., and the recommended parameter action range identifier can be "finishing stage". Parameter matching reasoning can be implemented through rule matching or machine learning algorithm. Rule matching is to filter out the parameter types and parameter action ranges that meet the conditions from the target associated feature combination according to the pre-defined rules. Machine learning algorithm can automatically discover the association between the target associated feature combination and the parameter types and parameter action ranges through learning from a large amount of historical data, and perform reasoning and prediction.

[0124] Step S500: performing matching reasoning verification on the candidate parameter matching scheme to generate a final process parameter matching result of the target production stage.

[0125] The matching rationality verification is a process of evaluating and verifying the candidate parameter matching scheme, aiming to ensure the rationality and feasibility of the candidate parameter matching scheme in the target production stage. The final process parameter matching result is a parameter matching scheme suitable for the target production stage obtained after the matching rationality verification, and is an important basis for guiding actual production.

[0126] As an implementation manner, in step S500, the matching rationality verification is performed on the candidate parameter matching scheme to generate the final process parameter matching result of the target production stage. Specifically, the following steps S510-S560 can be implemented: In step S510, a verification data set in the historical production record is obtained, and the verification data set includes an actual operation process description and an actual parameter setting record of the historical production stage.

[0127] The verification data set is a data set obtained from the historical production record for verifying the rationality of the candidate parameter matching scheme. The actual operation process description of the historical production stage records the operation steps actually performed in the past production process, and the actual parameter setting record records the corresponding actual parameter setting condition. By analyzing these historical data, the actual association between the operation and the parameter in the past production process can be understood, thereby providing a reference for verifying the candidate parameter matching scheme. The verification data set can be obtained by querying the production database of the factory and arranging and extracting the historical production record.

[0128] In step S520, the verification data set is subjected to knowledge association analysis to generate a verification knowledge association structure, and the verification knowledge association structure includes an association relationship identifier between the historical operation process and the historical parameter setting.

[0129] The process of knowledge association analysis on the verification data set is similar to the process of knowledge association analysis on the process information set in step S200. Through the knowledge association analysis, the internal association between the historical operation process description and the actual parameter setting record is mined to generate the verification knowledge association structure. The verification knowledge association structure explicitly indicates the corresponding relationship between the historical operation process and the historical parameter setting through the association relationship identifier, thereby providing a basis for subsequent matching degree calculation.

[0130] For example, in the verification knowledge association structure of mechanical part machining, the association relationship identifier can associate the “part rough machining” historical operation process with the “cutting speed” and “cutting depth” historical parameter settings.

[0131] In step S530, a verification association feature group is extracted from the verification knowledge association structure, and the verification association feature group includes historical operation parameter association features and historical parameter coordination features.

[0132] The process and steps of extracting the verification association feature group from the verification knowledge association structure are similar to the process of extracting the process association feature group from the process knowledge association structure in step S300. The historical operation parameter association feature reflects the association strength and distribution between the historical operation step and the historical parameter type, and the historical parameter synergy feature reflects the association strength and distribution between the historical operation target and the historical parameter action range. By extracting the verification association feature group, the association pattern between the operation and the parameter in the historical production process can be understood, which is used for comparison with the candidate parameter matching scheme.

[0133] For example, in mechanical part processing, the historical operation parameter association feature may show that the association strength between the "part rough machining" historical operation step and the "cutting speed" historical parameter type is high, and the historical parameter synergy feature may show that the association strength between the "improve part precision" historical operation target and the "finishing stage" historical parameter action range is high.

[0134] Step S540: Calculate the first matching degree of the parameter type recommendation identifier in the candidate parameter matching scheme and the historical operation parameter association feature in the verification association feature group.

[0135] The first matching degree is an index for measuring the matching degree between the parameter type recommendation identifier in the candidate parameter matching scheme and the historical operation parameter association feature. It reflects the rationality and applicability of the recommended parameter type in the historical production process. By calculating the first matching degree, it can be evaluated whether the recommendation of the parameter type in the candidate parameter matching scheme conforms to the actual situation of the historical production.

[0136] As an implementation manner, in step S540, the first matching degree of the parameter type recommendation identifier in the candidate parameter matching scheme and the historical operation parameter association feature in the verification association feature group is calculated, which can be implemented as steps S541-S546 as follows: Step S541: Extract the parameter type recommendation identifier set from the candidate parameter matching scheme.

[0137] The parameter type recommendation identifier set is the identifier set of the recommended parameter type in the candidate parameter matching scheme, which contains the identifiers of all the recommended parameter types for the target production stage.

[0138] For example, in the candidate parameter matching scheme of mechanical part processing, the parameter type recommendation identifier set may include "high-precision cutting speed", "small cutting depth", etc.

[0139] Step S542: Extract the historical operation parameter association feature from the verification association feature group, which contains the parameter type association strength distribution corresponding to the historical operation step identifier.

[0140] The historical operation parameter association feature reflects the association strength and distribution between the historical operation step and the historical parameter type. By extracting the historical operation parameter association feature from the verification association feature group, the association information between the operation step and the parameter type in the historical production process can be obtained, which is used for comparison with the parameter type recommendation identifier in the candidate parameter matching scheme.

[0141] For example, in mechanical part processing, the historical operation parameter association feature may show the association strength distribution of the "part rough machining" historical operation step and the "cutting speed", "cuting depth" and other historical parameter types.

[0142] Step S543: Perform matching analysis on the parameter type recommendation identifier set and the historical operation parameter association feature, and count the number of parameter type identifiers in the parameter type recommendation identifier set that exist in the historical operation parameter association feature.

[0143] The matching analysis is to compare the parameter type recommendation identifier set with the historical operation parameter association feature, and find out the parameter type identifiers in the parameter type recommendation identifier set that exist in the historical operation parameter association feature. Counting the number of existing parameter type identifiers can intuitively understand the degree of coincidence of the recommended parameter types in the candidate parameter matching scheme and the parameter types actually used in the historical production.

[0144] Step S544: Calculate the ratio of the number of existing parameter type identifiers to the total number of parameter type recommendation identifier set to obtain the initial matching degree.

[0145] The initial matching degree is calculated by the ratio of the number of existing parameter type identifiers to the total number of parameter type recommendation identifier set, which reflects the preliminary matching degree of the recommended parameter types in the candidate parameter matching scheme and the parameter types actually used in the historical production. The larger the ratio, the more matched the recommended parameter types and the historical actual use parameter types.

[0146] Step S545: Obtain the association strength value corresponding to the parameter type identifier that exists in the historical operation parameter association feature, and calculate the average value of the association strength value as the strength correction value.

[0147] The association strength value reflects the association closeness between the historical operation step and the parameter type. By obtaining the association strength value corresponding to the parameter type identifier that exists in the historical operation parameter association feature, and calculating the average value thereof, the strength correction value is obtained. The strength correction value is used to correct the initial matching degree, considering the importance and association of the parameter type in the historical production.

[0148] As an implementation, in step S545, the parameter type identifier corresponding to the correlation strength value existing in the historical operation parameter correlation feature is obtained, which can be implemented as steps S5451-S5455 as follows: Step S5451: Analyze the parameter adjustment record data in the historical production database, extract the parameter value change trajectory of the target parameter type identifier in the continuous production cycle, calculate the product of the fluctuation amplitude index and the fluctuation frequency index of the parameter value change trajectory, and generate the parameter stability correction factor.

[0149] The parameter adjustment record data records the adjustment of the parameter in the historical production process, including the adjustment time, the value before and after adjustment, etc. The target parameter type identifier refers to the identifier of the parameter type existing in the historical operation parameter correlation feature. The parameter value change trajectory is the value change of the target parameter type in the continuous production cycle, and the stability of the parameter can be understood by analyzing this trajectory. The fluctuation amplitude index is an index that measures the magnitude of the parameter value change, and the fluctuation frequency index is an index that measures the frequency of the parameter value change. The parameter stability correction factor is obtained by multiplying the fluctuation amplitude index and the fluctuation frequency index, which reflects the stability degree of the parameter.

[0150] For example, in mechanical part processing, for the "cutting speed" parameter type, by analyzing the parameter adjustment record data in the historical production database, the value change trajectory in the continuous production cycle is obtained. If the fluctuation amplitude is large and the fluctuation frequency is high, the parameter stability is poor, and the value of the parameter stability correction factor will be small.

[0151] Step S5452: Obtain the device running state record data associated with the parameter type identifier, analyze the device energy consumption rate data and product qualification rate data when the parameter adjustment operation occurs, calculate the ratio of the energy consumption rate data and the qualification rate data, and generate the energy efficiency balance correction factor.

[0152] The device running state record data records the running state of the production equipment in the historical production process, including the energy consumption, product qualification rate and other information of the equipment. By analyzing the relationship between the device energy consumption rate data and the product qualification rate data when the parameter adjustment operation occurs, the influence of parameter adjustment on the energy efficiency and product quality of the equipment can be evaluated. The energy efficiency balance correction factor is obtained by the ratio of the energy consumption rate data and the qualification rate data, which reflects the balance degree of the parameter adjustment between energy efficiency and quality.

[0153] For example, in mechanical part processing, when adjusting the "cutting speed" parameter, observe the energy consumption rate and product qualification rate of the equipment. If the energy consumption rate is large and the qualification rate is small, the balance between energy efficiency and quality of the parameter adjustment is poor, and the value of the energy efficiency balance correction factor will be small.

[0154] Step S5453: Perform multiplication operation on the original correlation strength value corresponding to the parameter type identifier in the historical operation parameter correlation feature and the parameter stability correction factor to generate a primary correction strength value.

[0155] The primary correction strength value is obtained by multiplying the original correlation strength value corresponding to the parameter type identifier in the historical operation parameter correlation feature and the parameter stability correction factor. The influence of the stability of the parameter on the correlation strength is considered, and the original correlation strength value is preliminarily corrected.

[0156] Step S5454: Perform weighted harmonic calculation on the primary correction strength value and the energy efficiency balance correction factor to fuse the primary correction strength value and the energy efficiency balance correction factor according to the preset harmonic weight coefficient to generate a final correlation strength value.

[0157] The weighted harmonic calculation is a calculation method for fusing the primary correction strength value and the energy efficiency balance correction factor according to the preset harmonic weight coefficient. The harmonic weight coefficient is a coefficient preset for distributing the weights of the primary correction strength value and the energy efficiency balance correction factor. Through the weighted harmonic calculation, the influences of the stability of the parameter and the energy efficiency balance on the correlation strength are comprehensively considered to obtain the final correlation strength value.

[0158] Step S5455: Update the correlation strength distribution data of the parameter type identifier in the historical operation parameter correlation feature based on the final correlation strength value.

[0159] The final correlation strength value calculated is updated to the correlation strength distribution data of the parameter type identifier in the historical operation parameter correlation feature, so that the correlation strength data more accurately reflects the actual correlation of the parameter in the historical production process.

[0160] Step S546: Multiply the initial matching degree and the strength correction value to obtain a first matching degree of the parameter type recommendation identifier and the historical operation parameter correlation feature.

[0161] The initial matching degree is multiplied by the strength correction value to obtain a final first matching degree. This value comprehensively considers the coincidence degree of the recommended parameter type in the candidate parameter matching scheme and the historical actual parameter type, and the correlation strength of the parameter in the historical production, and more accurately reflects the rationality and applicability of the parameter type recommendation.

[0162] Step S550: Calculate a second matching degree of the parameter action range recommendation identifier in the candidate parameter matching scheme and the historical parameter synergy feature in the verification correlation feature group.

[0163] The second matching degree is an index for measuring the matching degree between the parameter action range recommendation identifier in the candidate parameter matching scheme and the historical parameter synergy feature. It reflects the rationality and applicability of the recommended parameter action range in the historical production process. The calculation method is similar to that of the first matching degree. By comparing the association strength distribution of the parameter action range recommendation identifier set and the parameter action range in the historical parameter synergy feature, the number of overlapping parameter action range identifiers is counted, the initial matching degree is calculated, and the association strength value is considered for correction to obtain the second matching degree.

[0164] For example, in mechanical part processing, the recommended parameter action range identifier in the candidate parameter matching scheme is "finishing stage", and the second matching degree is calculated by comparison with the historical parameter synergy feature.

[0165] Step S560: generating a matching rationality evaluation result based on the first matching degree and the second matching degree, and screening the final process parameter matching result of the target production stage according to the matching rationality evaluation result.

[0166] The matching rationality evaluation result is the result of comprehensively evaluating the rationality of the candidate parameter matching scheme according to the first matching degree and the second matching degree. By weighting calculation or other comprehensive evaluation methods on the first matching degree and the second matching degree, a comprehensive evaluation index is obtained. According to this evaluation index, the candidate parameter matching scheme meeting the requirements is screened out as the final process parameter matching result of the target production stage.

[0167] As an implementation manner, in step S560, a matching rationality evaluation result is generated based on the first matching degree and the second matching degree, and the final process parameter matching result of the target production stage is screened according to the matching rationality evaluation result. Specifically, it can be implemented as steps S561-S565 as follows: Step S561: obtaining a preset matching degree weight distribution rule, the matching degree weight distribution rule including a weight value corresponding to the first matching degree and a weight value corresponding to the second matching degree.

[0168] The preset matching degree weight distribution rule is a rule for distributing the weights of the first matching degree and the second matching degree defined in advance. The weight value corresponding to the first matching degree and the weight value corresponding to the second matching degree reflect the importance of parameter type matching and parameter action range matching when evaluating the rationality of the candidate parameter matching scheme.

[0169] Step S562: calculating the product of the first matching degree and the corresponding weight value to obtain the first weighted matching degree; and calculating the product of the second matching degree and the corresponding weight value to obtain the second weighted matching degree.

[0170] The first weighted matching degree is a product of the first matching degree and a corresponding weight value, and the second weighted matching degree is a product of the second matching degree and a corresponding weight value. Through the weighted calculation, the importance of different matching degrees in the evaluation can be highlighted.

[0171] Step S563: Adding the first weighted matching degree and the second weighted matching degree to obtain a comprehensive matching degree.

[0172] The comprehensive matching degree is a result of adding the first weighted matching degree and the second weighted matching degree, and comprehensively considers the parameter type matching and the parameter action range matching. It is a comprehensive index for evaluating the rationality of the candidate parameter matching scheme.

[0173] Step S564: Comparing the comprehensive matching degree with a preset matching degree threshold value. If the comprehensive matching degree is greater than or equal to the matching degree threshold value, the candidate parameter matching scheme is taken as the final process parameter matching result.

[0174] The preset matching degree threshold value is a threshold value preset for judging whether the candidate parameter matching scheme is reasonable. If the comprehensive matching degree is greater than or equal to the matching degree threshold value, it indicates that the candidate parameter matching scheme is relatively reasonable in the matching of the parameter type and the parameter action range, and can be taken as the final process parameter matching result of the target production stage.

[0175] Step S565: If the comprehensive matching degree is less than the matching degree threshold value, the process returns to re-execute the process parameter matching reasoning operation based on the process-associated feature group until a candidate parameter matching scheme with a comprehensive matching degree greater than or equal to the matching degree threshold value is obtained as the final process parameter matching result.

[0176] If the comprehensive matching degree is less than the matching degree threshold value, it indicates that the candidate parameter matching scheme has unreasonable places in the matching of the parameter type and the parameter action range, and the process parameter matching reasoning operation based on the process-associated feature group needs to be re-executed. By continuously adjusting the parameters and rules in the reasoning process, a new candidate parameter matching scheme is tried to be generated until a scheme with a comprehensive matching degree greater than or equal to the matching degree threshold value is obtained, which is taken as the final process parameter matching result.

[0177] It can be understood that, in the above introduction of the embodiments of the present application, various algorithms involved, such as word segmentation algorithm, conflict resolution algorithm and the like, can be known from related content in the prior art. In order to save space, the embodiments of the present application are not expanded too much. In addition, those skilled in the art can supplement details according to common knowledge in the art when implementing the scheme of the present application. For example, according to common knowledge in the art, normalization can be used to eliminate the dimensional conflict before feature fusion, interpolation can be used to eliminate the dimensional difference, threshold setting can be combined with historical data, experience or business scenario requirements, model training can be based on a general model training method, the number of layers in the model structure can be set based on actual needs, and the selection of activation function and the like. The present application no longer introduces the redundant implementation process in too much detail.

[0178] Please refer to Figure 2 , Figure 2 A structural schematic diagram of a computer system provided by the embodiments of the present application is shown in the figure, which at least includes a processor 101, a communication interface 102 and a memory 103. The processor 101, the communication interface 102 and the memory 103 can be connected through a bus or other means. The processor 101 (or Central Processing Unit, CPU) is the computing core and control core of the computer system, which can parse various instructions in the computer system and process various data of the computer system. The communication interface 102 can optionally include a standard wired interface, a wireless interface (such as WI-FI, a mobile communication interface, etc.), and can be used for transmitting and receiving data under the control of the processor 101; the communication interface 102 can also be used for internal data transmission and interaction of the computer system. The memory 103 is a memory device in the computer system, used for storing programs and data. It can be understood that the memory 103 here can include the built-in memory of the computer system, and of course can also include the extended memory supported by the computer system. The memory 103 provides a storage space that stores the operating system of the computer system, and the present application does not limit this.

[0179] In one embodiment, the processor 101 executes the computer program in the memory 103 to perform the knowledge graph-based intelligent matching method of factory process parameters provided by the embodiments of the present application.

Claims

1. A knowledge graph-based intelligent matching method for factory process parameters, characterized in that, The method comprises the following steps: acquiring a process information set, the process information set comprising operation flow descriptions of different production stages and parameter setting records of corresponding stages; performing knowledge correlation analysis on the process information set to generate a process knowledge correlation structure; extracting a process correlation feature group from the process knowledge correlation structure, the process correlation feature group comprising operation parameter correlation features and parameter synergy features; performing process parameter matching reasoning based on the process correlation feature group to generate a candidate parameter matching scheme for a target production stage; verifying the candidate parameter matching scheme to generate a final process parameter matching result for the target production stage.

2. The method of claim 1, wherein, The process knowledge correlation structure is generated by performing knowledge correlation analysis on the process information set, comprising the following steps: performing semantic decomposition on the operation flow descriptions in the process information set to obtain an operation element set comprising operation step identifiers and operation target identifiers; extracting parameter attributes from the parameter setting records in the process information set to obtain a parameter element set comprising parameter type identifiers and parameter scope identifiers; establishing a first correlation relationship between the operation step identifiers in the operation element set and the parameter type identifiers in the parameter element set, the first correlation relationship being used to represent the parameter types corresponding to preset operation steps; establishing a second correlation relationship between the operation target identifiers in the operation element set and the parameter scope identifiers in the parameter element set, the second correlation relationship being used to represent the parameter scopes corresponding to preset operation targets; constructing a process knowledge correlation structure based on the first correlation relationship and the second correlation relationship, the process knowledge correlation structure representing the correlation relationships between operation elements and parameter elements in a node-edge connection manner.

3. The method of claim 2, wherein, The process correlation feature group is extracted from the process knowledge correlation structure, comprising the following steps: quantifying the node-edge connection relationships in the process knowledge correlation structure to calculate first correlation strength values between operation step identifiers and parameter type identifiers, and second correlation strength values between operation target identifiers and parameter scope identifiers; constructing operation parameter correlation features based on the first correlation strength values, the operation parameter correlation features comprising parameter type correlation strength distributions corresponding to different operation steps; constructing parameter synergy features based on the second correlation strength values, the parameter synergy features comprising parameter scope correlation strength distributions corresponding to different operation targets; combining the operation parameter correlation features and the parameter synergy features to obtain the process correlation feature group.

4. The method of claim 3, wherein, The process parameter matching reasoning is performed based on the process correlation feature group to generate a candidate parameter matching scheme for a target production stage, comprising the following steps: acquiring a target operation flow description of a target production stage, performing semantic decomposition on the target operation flow description to obtain a target operation element set comprising target operation step identifiers and target operation target identifiers; extracting operation parameter correlation features corresponding to the target operation step identifiers from the process correlation feature group; extract a parameter synergy feature corresponding to the target operation target identifier from the process-related feature group; perform feature fusion on the operation parameter-related feature and the parameter synergy feature to generate a target-related feature combination; perform parameter matching reasoning based on the target-related feature combination to generate a candidate parameter matching scheme containing a parameter type recommendation identifier and a parameter action range recommendation identifier.

5. The method of claim 4, wherein, The candidate parameter matching scheme is verified for matching rationality to generate the final process parameter matching result of the target production stage, including: obtain a verification data set in the historical production record, the verification data set containing actual operation process descriptions and actual parameter setting records in the historical production stage; perform knowledge association analysis on the verification data set to generate a verification knowledge association structure containing association relationship identifiers between historical operation processes and historical parameter settings; extract a verification-related feature group from the verification knowledge association structure, the verification-related feature group containing historical operation parameter-related features and historical parameter synergy features; calculate a first matching degree of the parameter type recommendation identifier in the candidate parameter matching scheme and the historical operation parameter-related feature in the verification-related feature group; calculate a second matching degree of the parameter action range recommendation identifier in the candidate parameter matching scheme and the historical parameter synergy feature in the verification-related feature group; generate a matching rationality evaluation result based on the first matching degree and the second matching degree, and filter the final process parameter matching result of the target production stage according to the matching rationality evaluation result.

6. The method of claim 2, wherein, The operation process description in the process information set is semantically decomposed to obtain an operation element set containing operation step identifiers and operation target identifiers, including: perform natural language word segmentation on the operation process description to obtain a word segmentation unit set containing operation verbs and operation objects; map the word segmentation unit set to step identifiers based on a preset operation step dictionary, and map the word segmentation unit containing operation verbs and operation objects to operation step identifiers; map the word segmentation unit set to target identifiers based on a preset operation target dictionary, and map the word segmentation unit containing operation purpose descriptions to operation target identifiers; remove duplicates from the operation step identifiers and the operation target identifiers to retain operation step identifiers and operation target identifiers with unique identifiers; combine the operation step identifiers and the operation target identifiers after deduplication to obtain an operation element set.

7. The method of claim 3, wherein, The node-edge connection relationship in the process knowledge association structure is quantified to calculate a first association strength value between the operation step identifier and the parameter type identifier, including: statistically count the number of connection edges between the operation step identifier and the parameter type identifier in the process knowledge association structure; statistically count the total number of connection edges corresponding to the operation step identifier in the process knowledge association structure; calculate the ratio of the number of connection edges to the total number of connection edges to obtain an initial association strength value between the operation step identifier and the parameter type identifier; Obtain the historical execution times corresponding to the operation step identifiers and the historical usage times corresponding to the parameter type identifiers in the process information set; Calculate the product of the historical execution times and the historical usage times to obtain an association weight correction value; Multiply the initial association strength value by the association weight correction value to obtain a first association strength value between the operation step identifiers and the parameter type identifiers.

8. The method of claim 4, wherein, The feature fusion of the operation parameter association feature and the parameter synergy feature generates a target association feature combination, including: Aligning the feature dimensions of the operation parameter association feature to make the parameter type association strength distributions corresponding to different operation step identifiers have the same dimensions, and aligning the feature dimensions of the parameter synergy feature to make the parameter action range association strength distributions corresponding to different operation target identifiers have the same dimensions; Based on a preset feature fusion rule, add the dimension-aligned operation parameter association feature and the parameter synergy feature dimension by dimension to generate a preliminary fusion feature; Normalize the preliminary fusion feature to adjust the feature value range to a preset normalization interval; Filter the normalized preliminary fusion feature to retain dimensions with feature values exceeding a preset threshold to generate a target association feature combination.

9. The method of claim 5, wherein, The calculation of the first matching degree of the parameter type recommendation identifier in the candidate parameter matching scheme and the historical operation parameter association feature in the verification association feature group includes: Extract a parameter type recommendation identifier set from the candidate parameter matching scheme; Extract a historical operation parameter association feature from the verification association feature group, which includes a parameter type association strength distribution corresponding to a historical operation step identifier; Perform matching analysis on the parameter type recommendation identifier set and the historical operation parameter association feature to count the number of parameter type identifiers in the parameter type recommendation identifier set that exist in the historical operation parameter association feature; Calculate the ratio of the number of existing parameter type identifiers to the total number of parameter type recommendation identifiers to obtain an initial matching degree; Obtain the association strength value corresponding to the existing parameter type identifier in the historical operation parameter association feature, and calculate the average value of the association strength value as a strength correction value; Multiply the initial matching degree by the strength correction value to obtain the first matching degree of the parameter type recommendation identifier and the historical operation parameter association feature; The generation of the matching rationality evaluation result based on the first matching degree and the second matching degree, and the selection of the final process parameter matching result of the target production stage according to the matching rationality evaluation result, includes: Obtain a preset matching degree weight distribution rule, which includes a weight value corresponding to the first matching degree and a weight value corresponding to the second matching degree; Calculate the product of the first matching degree and the corresponding weight value to obtain a first weighted matching degree; Calculate the product of the second matching degree and the corresponding weight value to obtain a second weighted matching degree; Add the first weighted matching degree and the second weighted matching degree to obtain a comprehensive matching degree; Compare the comprehensive matching degree with a preset matching degree threshold value, and if the comprehensive matching degree is greater than or equal to the matching degree threshold value, the candidate parameter matching scheme is taken as a final process parameter matching result. If the comprehensive matching degree is less than the matching degree threshold value, a process parameter matching reasoning operation based on the process association feature group is returned to be re-executed until a candidate parameter matching scheme with a comprehensive matching degree greater than or equal to the matching degree threshold value is obtained as a final process parameter matching result.

10. A computer system, characterized by Comprise: A memory in which a computer program is stored; A processor for loading the computer program to realize the knowledge graph-based factory process parameter intelligent matching method according to any one of claims 1-9.

Images