Mechanical part intelligent cost evaluation system and method based on knowledge graph

Abstract

The invention provides a mechanical part intelligent cost evaluation system and method based on a knowledge graph, and relates to the field of mechanical manufacturing, and the method comprises the steps: converting discrete process parameters, equipment capability and other process elements into multi-dimensional knowledge nodes through semantic modeling, and constructing a hypergraph structure based on a process constraint relation and a cost conduction rule, a full-link knowledge network covering'feature-process-resource-cost 'is formed; performing semantic representation on a plurality of process paths by adopting a hybrid embedding strategy, dynamically capturing implicit coupling effects among different process paths through an attention mechanism, and strengthening a framework optimization path selection strategy; and finally, the cost-driven weight of each path is quantified through a path probability evaluation model based on a classifier, adaptive recommendation of a process chain global optimal solution is realized, and the problem of dimension disasters in dynamic process combination optimization of a traditional method is solved.

Description

Technical Field

[0001] This application relates to the field of mechanical manufacturing, and more particularly, in the embodiments of this application, it relates to an intelligent cost evaluation system and method for mechanical parts based on a knowledge graph. Background Art

[0002] With the rapid development of intelligent manufacturing and industrial 4.0 Internet technologies, cost evaluation in the field of mechanical manufacturing is facing a paradigm shift from experience-driven to data-driven. Traditional cost evaluation methods (such as parametric modeling method, analogy estimation method) are mostly based on linear regression or expert experience databases. Their core defect is that they simplify complex process characteristics into limited key parameters, making it difficult to effectively model the dynamic coupling relationship between processes, equipment, and resources. Process characteristics cover multiple aspects such as process flow, process parameters, equipment type, tooling fixtures, and operator skills. In the multi-variety, small-batch production mode, the high heterogeneity of process characteristics (such as the dependence of special processing technology on high-precision tooling, special heat treatment requirements for composite materials) makes the cost prediction error rate of traditional methods relatively large.

[0003] The current application of knowledge graph technology in the manufacturing industry mainly focuses on the field of product design knowledge management. There are two key bottlenecks in its application paradigm: on the one hand, process knowledge modeling mostly stays at the static attribute level, lacking comprehensive analysis of continuous information such as the dynamic range of process parameters and the elasticity of equipment capabilities, and unable to accurately reflect the impact of processes on costs; on the other hand, the existing knowledge reasoning mechanism is difficult to support the analysis of cost conduction paths across process chains, especially when dealing with complex scenarios such as multi-process parallelism and resource competition.

[0004] Therefore, an optimized intelligent cost evaluation system for mechanical parts is desired. Summary of the Invention

[0005] To solve the above technical problems, this application is proposed. The embodiments of this application provide an intelligent cost evaluation system and method for mechanical parts based on a knowledge graph. First, through semantic modeling, discrete process elements such as process parameters and equipment capabilities are transformed into multi-dimensional knowledge nodes, and a hypergraph structure is constructed based on process constraint relationships and cost conduction laws, forming a full-link knowledge network covering "feature - process - resource - cost". Then, a hybrid embedding strategy is used to semantically represent multiple process paths, and the implicit coupling effect between different process paths is dynamically captured through an attention mechanism to strengthen the framework optimization path selection strategy. Finally, a path probability evaluation model based on a classifier is used to quantify the cost driving weights of each path, realizing the adaptive recommendation of the global optimal solution of the process chain and breaking through the dimensionality disaster problem in the dynamic process combination optimization of traditional methods.

[0006] According to one aspect of the present application, there is provided an intelligent cost evaluation system for mechanical parts based on a knowledge graph, which includes:

[0007] A data input module for inputting mechanical part features and process features;

[0008] A knowledge graph construction module for constructing a mechanical part cost evaluation knowledge graph oriented to process features based on mechanical part features and process features;

[0009] A module for obtaining features of parts to be evaluated, which is used to obtain the feature data of mechanical parts to be evaluated;

[0010] A module for reasoning the path of the mechanical part to be evaluated, which is used to perform path reasoning in the mechanical part cost evaluation knowledge graph oriented to process features according to the feature data of the mechanical part to be evaluated to obtain a set of alternative process paths and a set of analysis results of cost driving factors corresponding to the set of alternative process paths;

[0011] An optimal process path recommendation module for inputting the set of alternative process paths into an optimal process recommendation module to obtain an optimal process path, including: the optimal process recommendation module uses a semantic-level feature analysis method to perform intelligent evaluation of the optimal path and recommend the optimal process for the set of alternative process paths to obtain the optimal process path;

[0012] A cost prediction value module for determining the cost prediction value of the mechanical part to be evaluated based on the analysis result of the cost driving factors of the optimal process path.

[0013] According to another aspect of the present application, there is provided an intelligent cost evaluation method for mechanical parts based on a knowledge graph, which includes:

[0014] Input mechanical part features and process features;

[0015] Based on mechanical part features and process features, construct a mechanical part cost evaluation knowledge graph oriented to process features;

[0016] Obtain the feature data of the mechanical part to be evaluated;

[0017] According to the feature data of the mechanical part to be evaluated, perform path reasoning in the mechanical part cost evaluation knowledge graph oriented to process features to obtain a set of alternative process paths and a set of analysis results of cost driving factors corresponding to the set of alternative process paths;

[0018] Input the set of alternative process paths into an optimal process recommendation module to obtain an optimal process path, including: the optimal process recommendation module uses a semantic-level feature analysis method to perform intelligent evaluation of the optimal path and recommend the optimal process for the set of alternative process paths to obtain the optimal process path;

[0019] Based on the analysis results of cost drivers for the optimal manufacturing process path, determine the cost prediction value of the mechanical part to be evaluated. Compared with the prior art, an intelligent cost evaluation system and method for mechanical parts based on a knowledge graph provided by the present application first converts discrete manufacturing process elements such as process parameters and equipment capabilities into multi-dimensional knowledge nodes through semantic modeling, and constructs a hypergraph structure based on process constraint relationships and cost conduction laws to form a full-link knowledge network covering "feature - process - resource - cost". Then, a hybrid embedding strategy is adopted to semantically represent multiple process paths, and the implicit coupling effect between different manufacturing process paths is dynamically captured through an attention mechanism to strengthen the framework for optimizing the path selection strategy. Finally, a path probability evaluation model based on a classifier is used to quantify the cost driver weights of each path, realizing the adaptive recommendation of the global optimal solution of the process chain and breaking through the curse of dimensionality problem in the dynamic process combination optimization of traditional methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] By describing the embodiments of the present application in more detail with reference to the accompanying drawings, the above and other objects, features, and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification, and are used to explain the present application together with the embodiments of the present application, and do not constitute a limitation to the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0021] Figure 1 FIG. is a system block diagram of an intelligent cost evaluation system for mechanical parts based on a knowledge graph according to an embodiment of the present application.

[0022] Figure 2 FIG. is a schematic diagram of data flow of an intelligent cost evaluation system for mechanical parts based on a knowledge graph according to an embodiment of the present application.

[0023] Figure 3 FIG. is a block diagram of an optimal manufacturing process path recommendation module in an intelligent cost evaluation system for mechanical parts based on a knowledge graph according to an embodiment of the present application.

[0024] Figure 4 FIG. is a block diagram of a process path semantic expression strengthening unit in an intelligent cost evaluation system for mechanical parts based on a knowledge graph according to an embodiment of the present application.

[0025] Figure 5 FIG. is a block diagram of a feature phase reshaping gain factor calculation sub-unit in an intelligent cost evaluation system for mechanical parts based on a knowledge graph according to an embodiment of the present application.

[0026] Figure 6 FIG. is a flowchart of an intelligent cost evaluation method for mechanical parts based on a knowledge graph according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] Various exemplary embodiments, features, and aspects of the present application will be described in detail below with reference to the accompanying drawings. Like reference numerals in the drawings denote functionally identical or similar elements. Although various aspects of the embodiments are shown in the drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0028] As used herein, the term "exemplary" means "serving as an example, embodiment, or illustration." Any embodiment described herein as "exemplary" is not necessarily to be construed as superior or better than other embodiments.

[0029] In addition, for a better description of the present application, numerous specific details are given in the following detailed implementation. Those skilled in the art should understand that the present application can also be implemented without some specific details. In some instances, methods, means, elements, and circuits well known to those skilled in the art are not described in detail so as to highlight the gist of the present application.

[0030] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present application, "a plurality" means two or more unless otherwise specifically defined.

[0031] With the development of intelligent manufacturing and industrial Internet technologies, the assessment of mechanical manufacturing costs is undergoing a transformation from experience-driven to data-driven. However, traditional methods such as parametric modeling and analogy estimation methods are based on linear regression or expert experience that simplifies key parameters and are difficult to effectively model the complex dynamic relationships among processes, equipment, and resources. Especially in multi-variety and small-batch production, the high heterogeneity of process characteristics leads to an increase in cost prediction errors. Although knowledge graphs have been applied in the product design field of manufacturing, their modeling of process knowledge only stays at the static attribute level, lacks comprehensive analysis of continuous information, and the existing reasoning mechanisms are difficult to support the analysis of complex cross-process-chain cost conduction paths.

[0032] In view of the above technical problems, in the technical solution of the present application, an intelligent cost evaluation system for mechanical parts based on a knowledge graph is proposed, which can construct a dynamic knowledge graph architecture for process characteristics. That is, by introducing the idea of process parameter elastic domain, the characteristics of mechanical parts are coupled with the process parameters of continuous processes for modeling, and a hypergraph structure is used to express the multi-process collaborative constraint relationship, breaking through the expression defects of traditional methods in describing the high-order associations of different entities. Specifically, the system converts discrete process elements such as process parameters and equipment capabilities into multi-dimensional knowledge nodes through semantic modeling, and constructs a hypergraph structure based on process constraint relationships and cost conduction laws to form a full-link knowledge network covering "feature - process - resource - cost". In the reasoning stage, a hybrid embedding strategy is adopted to semantically represent multiple process paths, and the attention mechanism is used to dynamically capture the implicit coupling effects between different process paths, and the path selection strategy is optimized by combining the semantic expression reinforcement framework. Finally, the cost driving weights of each path are quantified through a path probability evaluation model based on a classifier, and an adaptive recommendation of the global optimal solution of the process chain is realized, breaking through the dimensionality disaster problem in the dynamic process combination optimization of traditional methods. The present application proposes an intelligent cost evaluation system for mechanical parts based on a knowledge graph. Figure 1 FIG. is a system block diagram of an intelligent cost evaluation system for mechanical parts based on a knowledge graph according to an embodiment of the present application. Figure 2 FIG. is a schematic diagram of data flow of an intelligent cost evaluation system for mechanical parts based on a knowledge graph according to an embodiment of the present application. As Figure 1 and Figure 2 shown, an intelligent cost evaluation system 100 for mechanical parts based on a knowledge graph according to an embodiment of the present application includes: a data input module 110 for inputting mechanical part features and process features; a knowledge graph construction module 120 for constructing a mechanical part cost evaluation knowledge graph for process characteristics based on mechanical part features and process features; a to-be-evaluated part feature acquisition module 130 for acquiring feature data of a to-be-evaluated mechanical part; a to-be-evaluated mechanical part path reasoning module 140 for performing path reasoning in the mechanical part cost evaluation knowledge graph for process characteristics according to the feature data of the to-be-evaluated mechanical part to obtain a set of alternative process paths and a set of cost driving factor analysis results corresponding to the set of alternative process paths; an optimal process path recommendation module 150 for inputting the set of alternative process paths into the optimal process recommendation module to obtain an optimal process path; and a cost prediction value module 160 for determining a cost prediction value of the to-be-evaluated mechanical part based on the cost driving factor analysis result of the optimal process path.

[0033] In the above-mentioned intelligent cost evaluation system 100 for mechanical parts based on a knowledge graph, a data input module 110 is used to input mechanical part features and manufacturing process features. In an embodiment of the present application, the mechanical part features include geometric features, material features, and technical requirements; the manufacturing process features include process type, process parameter range, required equipment, tooling, and cycle time. It should be understood that among the mechanical part features, geometric features cover information such as the basic dimensions and shapes of parts, which are crucial for determining the specific process types to be adopted during machining. For example, complex curved surfaces may require five-axis CNC machine tools for machining, while simple planar or cylindrical surfaces can be completed using relatively basic three-axis machine tools. In addition, the material features indicate the types of materials used for the parts. Different materials not only affect the machining difficulty but also relate to the material cost. For instance, due to its high hardness and poor thermal conductivity, titanium alloy often requires special tools during machining and has a slower machining speed, thus increasing the machining cost; on the contrary, aluminum alloy usually has a lower machining cost due to its good machining performance. In terms of technical requirements, the quality standards and other specific technical conditions that the parts need to meet during production are clearly defined, and these are all factors that cannot be ignored during cost evaluation. Some parts with high-precision requirements may require multiple finishing processes and even special treatments such as ultra-precision grinding or electrical discharge machining, which undoubtedly increases the complexity and cost of the process chain. At the same time, inputting the manufacturing process features is to more accurately simulate various possibilities in the actual production environment. Among them, different process types correspond to different cost structures. For example, there are significant differences in raw material utilization rates between casting processes and forging processes. The former can effectively reduce waste generation but may require higher mold costs. The process parameter range involves various control parameters during machining, such as cutting speed, feed rate, etc. Changes in these parameters will directly affect machining efficiency and quality, and thus affect costs. Specifically, a higher cutting speed can improve machining efficiency but may also lead to increased tool wear, thereby increasing the tool change frequency and corresponding costs; conversely, reducing the cutting speed can extend the tool life but will result in longer machining time, which is also not conducive to cost control. In addition, the required equipment and tooling are also important components of the manufacturing process features, and they directly determine whether the established process steps can be successfully completed. The purchase costs, maintenance costs, and usage efficiencies of different equipment will all have an impact on the final cost. For example, compared with ordinary machine tools, high-precision CNC machine tools, although having a larger initial investment, can improve production efficiency by optimizing machining paths while ensuring machining accuracy, which helps to reduce costs in the long run. Finally, as a key indicator for measuring production efficiency, cycle time reflects the number of parts that can be produced per unit time and directly affects the overall production capacity and economic benefits of the production line.By comprehensively collecting and accurately inputting the above mechanical part features and manufacturing process features, a solid foundation can be laid for the subsequent construction of a knowledge graph for mechanical part cost assessment oriented to manufacturing process features, ensuring that the established knowledge graph can truly reflect various cost driving factors in the whole process of part manufacturing, thereby providing strong support for achieving accurate cost prediction.

[0034] In the above-mentioned intelligent cost evaluation system 100 for mechanical parts based on a knowledge graph, a knowledge graph construction module 120 is used to construct a knowledge graph for cost evaluation of mechanical parts oriented to process characteristics based on mechanical part characteristics and process characteristics. In an embodiment of the present application, the knowledge graph for cost evaluation of mechanical parts oriented to process characteristics includes entities and relationships. The entities include mechanical parts, materials, processes, equipment, tooling, process parameters, cost elements, and cost indicators; the relationships include part - uses material - material, part - goes through process - process, process - uses equipment - equipment, process - requires tooling - tooling, process - controls parameters - process parameters, part - generates cost - cost element, cost element - affects cost indicator - cost indicator, and process - affects cost element - cost element. It should be understood that based on mechanical part characteristics and process characteristics, a knowledge graph for cost evaluation of mechanical parts oriented to process characteristics can be constructed. This knowledge graph consists of a series of entities and their mutual relationships, forming a full-link knowledge network covering "characteristics - process - resources - cost". Among them, the entity part covers multiple aspects such as mechanical parts, materials, processes, equipment, tooling, process parameters, cost elements, and cost indicators. Each entity represents an indispensable link or factor in the manufacturing process, and they together constitute the basic framework of production activities. For example, as the core entity, mechanical parts, information such as their geometric characteristics, material properties, and technical requirements are crucial for determining subsequent processing procedures; while materials determine the physical properties of parts and their processing difficulties. The process entity is not only the specific operation steps to transform raw materials into finished products but also involves a series of complex decision-making processes. Each process has its specific operation scope and requirements, such as control parameters like temperature and pressure. The equipment entity reflects the hardware facilities required for each process. Different types of equipment not only affect processing accuracy but also relate to acquisition and maintenance costs. Tooling is auxiliary equipment used to ensure processing accuracy and improve production efficiency. The process parameter entity captures the technical details and control variables involved in each process step and is one of the key factors affecting product quality and cost. The cost element entity covers various direct and indirect cost sources, such as raw material costs and equipment depreciation, providing basic data for calculating the total cost. Cost indicators measure the economic benefits in the entire production process, such as the total cost per unit product and profit margin. To more deeply reflect various cost driving factors in the manufacturing process, the knowledge graph also includes the construction of relationships between entities. For example, the "part - uses material - material" relationship clarifies the specific material types selected for a specific part, which is crucial for understanding material costs and their impact on processing difficulties. The "part - goes through process - process" relationship shows the different process steps that a part undergoes in the entire manufacturing process, helping to identify key processes and optimize the process path. The "process - uses equipment - equipment" relationship reflects the types of equipment required for each process, which is of great significance for reasonably allocating resources and reducing equipment idle rates.The "process - required tooling - tooling" relationship indicates the auxiliary tools needed to complete a certain process, which is also essential for improving processing efficiency and ensuring product quality. The "process - control parameters - process parameters" relationship reveals the various technical specifications that must be followed during the process execution, which is the key to ensuring the consistency of product quality. The "part - incurred cost - cost element" relationship shows the various cost expenditures caused by part production, laying the foundation for accurately calculating the total cost. The "cost element - affecting cost indicator - cost indicator" relationship demonstrates how to adjust and optimize cost indicators according to the changes in various cost elements to achieve the maximization of economic benefits. The "process - affecting cost element - cost element" relationship explains how different process steps affect the cost composition, helping to identify potential cost-saving opportunities.

[0035] In the above-mentioned intelligent cost assessment system 100 for mechanical parts based on the knowledge graph, the part feature acquisition module 130 for the part to be evaluated is used to acquire the feature data of the mechanical part to be evaluated. It should be understood that the feature data of mechanical parts usually includes basic information such as the geometric shape, material, size, weight, etc. of the part, and these data determine the processes, equipment, and resources required for the part in different processing processes. By obtaining the detailed features of the part to be evaluated, the process requirements related to the part can be accurately identified and matched with the existing process paths, thereby providing data support for the subsequent processing process of the part. During this process, the geometric features of the part not only affect the selection of processing processes but are also closely related to the processing quality and accuracy. Different material types directly determine the mechanical properties, durability, and processing methods of the part, while size and weight are key factors for evaluating resource consumption and efficiency during part processing. By collecting these data, the attributes of the part can be better understood, and a more accurate basis can be provided for subsequent cost prediction. In addition, obtaining the feature data of mechanical parts can effectively enhance the adaptability of the system to different part characteristics, enabling the cost assessment model to automatically adjust the calculation method according to the specific requirements of the part, thereby ensuring the accuracy and personalization of cost assessment. Through accurate feature data input, not only can the prediction accuracy be improved, but also in the face of a complex processing environment, high processing capabilities can be maintained, avoiding evaluation deviations caused by missing or incorrect feature data.

[0036] In the above-mentioned intelligent cost evaluation system 100 for mechanical parts based on a knowledge graph, the path reasoning module 140 for the mechanical parts to be evaluated is used to perform path reasoning in the knowledge graph for cost evaluation of mechanical parts oriented to process characteristics according to the characteristic data of the mechanical parts to be evaluated, so as to obtain a set of alternative process paths and a set of analysis results of cost driving factors corresponding to the set of alternative process paths. It should be understood that the knowledge graph organically combines mechanical part characteristics (such as geometric shape, material composition, and technical requirements) with process characteristics (such as process type, process parameter range, required equipment, tooling, and cycle time) to form a multi-dimensional view that comprehensively covers the entire process from part design to finished product output. This structured representation method enables the interaction relationships between each process link to be clearly presented, providing strong support for subsequent path reasoning. For example, by recording the geometric characteristics of parts in detail, it is possible to determine which machining methods are most suitable for parts of a specific shape; understanding material characteristics and technical requirements helps to select the most appropriate machining process and predict possible technical problems and their solutions based on this. In this context, path reasoning becomes the key bridge connecting characteristic data with actual production processes. By deeply exploring the constructed knowledge graph, multiple possible process paths can be simulated, thus forming a set containing multiple alternative options. Each alternative process path represents a potential processing flow, which is constructed based on different process combinations, equipment configurations, and operating parameters and other elements. These paths not only reflect the logical relationships between different process steps but also reveal the impact mechanism of each step on the overall cost. For example, when considering the machining of a high-strength alloy part, multiple processes such as rough machining, semi-finishing, heat treatment, and finishing are required. And in each process, there are also multiple feasible operation methods and parameter settings, such as the selection of cutting speed and feed rate. By reasoning about different combinations of these variables, a series of alternative paths can be generated, and each path has its unique cost structure and quality control strategy. Further, the main purpose of performing path reasoning is to identify those process paths that can achieve the optimal cost-benefit. With the support of the knowledge graph, by carefully analyzing the key nodes in each alternative path, the main cost driving factors can be found. For example, some process steps may lead to a high tool wear rate, thus increasing the tool replacement frequency and related costs; while other steps may result in a long processing cycle, affecting the overall efficiency of the production line. By quantifying the specific contributions of these factors to the total cost, clear guidance can be provided for decision-makers to help them make the best choice among many possibilities.

[0037] In the above-mentioned knowledge graph-based intelligent cost evaluation system 100 for mechanical parts, the optimal process path recommendation module 150 is configured to input a set of alternative process paths into the optimal process recommendation module to obtain an optimal process path, including: the optimal process recommendation module uses a semantic-level feature analysis method to perform intelligent evaluation of the optimal path and recommend the optimal process for the set of alternative process paths to obtain the optimal process path.

[0038] Figure 3 It is a block diagram of the optimal process path recommendation module in the knowledge graph-based intelligent cost evaluation system for mechanical parts according to an embodiment of the present application. As Figure 3 shown, in the embodiment of the present application, the optimal process path recommendation module 150 includes: an alternative process path semantic embedding encoding unit 151, configured to perform semantic embedding encoding on each alternative process path in the set of alternative process paths to obtain a set of alternative process path semantic embedding encoding vectors; a process path semantic expression strengthening unit 152, configured to perform process path semantic expression strengthening on each alternative process path semantic embedding encoding vector in the set of alternative process path semantic embedding encoding vectors to obtain a set of alternative process path semantic strengthening encoding vectors; an optimal process path determination unit 153, configured to perform intelligent evaluation of the optimal path based on the set of alternative process path semantic strengthening encoding vectors to determine the optimal process path.

[0039] Specifically, the alternative process path semantic embedding encoding unit 151 is configured to perform semantic embedding encoding on each alternative process path in the set of alternative process paths to obtain a set of alternative process path semantic embedding encoding vectors. It should be understood that considering the complex non-linear relationship between continuous features such as process parameter ranges and equipment capacity flexibility and discrete features such as tooling fixtures involved in the machining process of mechanical parts, and the static triple representation method of the existing knowledge graph cannot effectively capture the dynamic parameter interaction relationship in the process chain. That is to say, traditional cost evaluation methods are difficult to handle the high-dimensional heterogeneity and dynamic coupling of process features. Therefore, in the technical solution of the present application, semantic embedding encoding is performed on each alternative process path in the set of alternative process paths to obtain a set of alternative process path semantic embedding encoding vectors. Through the method of semantic embedding encoding, different alternative process paths can be embedded and mapped into a common semantic space to construct a deep semantic representation of different process paths. In this way, implicit logics such as the temporal dependence relationship between processes (such as the requirement of the heat treatment process for the accuracy of the previous machining process) and resource competition constraints (such as the scheduling conflict of sharing high-precision CNC machine tools by multiple processes) can be transformed into computable mathematical expressions, providing analyzable semantic primitives for subsequent intelligent reasoning.

[0040] Figure 4 It is a block diagram of a process path semantic expression enhancement unit in an intelligent cost evaluation system for mechanical parts based on a knowledge graph according to an embodiment of the present application. As Figure 4 shown, in the embodiment of the present application, the process path semantic expression enhancement unit 152 is used to enhance the process path semantic expression of each alternative process path semantic embedding coding vector in the set of alternative process path semantic embedding coding vectors to obtain a set of alternative process path semantic enhanced coding vectors, including: an alternative process path semantic feature compression subunit 1521, which is used to perform feature phase reconstruction and information compression processing on the alternative process path semantic embedding coding vector to obtain a set of alternative process path semantic feature compressed local phase coding vectors; a feature effective component statistical number calculation subunit 1522, which is used to calculate the alternative process path semantic feature effective component statistical number of each alternative process path semantic feature compressed local phase coding vector in the set of alternative process path semantic feature compressed local phase coding vectors; a feature phase reshaping gain factor calculation subunit 1523, which is used to calculate the alternative process path semantic feature phase reshaping gain factor of each alternative process path semantic feature local phase coding vector based on the alternative process path semantic feature effective component statistical number of each alternative process path semantic feature compressed local phase coding vector; a feature phase significance reshaping subunit 1524, which is used to perform feature phase significance reshaping on the set of alternative process path semantic feature local phase coding vectors based on the alternative process path semantic feature phase reshaping gain factor of each alternative process path semantic feature local phase coding vector to obtain an alternative process path semantic enhanced coding vector. It should be understood that due to the complex constraint relationships involved in the multi-process collaboration of mechanical part processing (such as the parameter matching of turning and heat treatment) and the equipment capacity flexibility (such as the adaptation interval of the spindle speed of a numerically controlled machine tool), although different process paths are mapped to low-dimensional vectors in the initial semantic embedding stage, due to the high-order interaction characteristics between process nodes (such as a five-axis machining center simultaneously affecting machining accuracy and energy consumption cost), there are often feature dimension redundancy and semantic mixing phenomena inside the vectors. Therefore, it is difficult for each alternative process path semantic embedding coding vector in the set of alternative process path semantic embedding coding vectors to accurately represent the dynamic coupling effect hidden in the process path. Based on this, in the technical solution of the present application, the process path semantic expression of each alternative process path semantic embedding coding vector in the set of alternative process path semantic embedding coding vectors is further enhanced to obtain a set of alternative process path semantic enhanced coding vectors.

[0041] In particular, in the process of strengthening the semantic expression of the process path, by introducing a feature phase reconstruction mechanism, one-dimensional convolutional coding is used to capture implicit features and reconstruct the feature phase of the local structure of the semantic embedding coding vector of each alternative process path. This can effectively capture the phase correlation pattern between the dynamic range of process parameters and the equipment capacity threshold. For example, the sensitive interval of high-speed cutting process to sudden changes in equipment power can be identified through a specific convolutional kernel. In the information compression stage, irrelevant noise interference in the process path (such as the replacement frequency of non-critical tooling) is eliminated through feature rectification, and the effective components strongly related to cost drivers (such as the doubling effect of high-precision measurement processes on labor costs) are retained. Based on the gain operator calculation of the holomorphic structure metric decomposition, the cost-sensitive features in the process path (such as the clamping time cost depending on special fixtures) are calibrated for phase direction symmetry, so that the strengthened semantic vector not only highlights the key cost nodes in the process chain (such as the argon consumption of the laser cutting process), but also maintains the consistency of feature distribution during multi-path comparison. The effect of this semantic expression strengthening is reflected in two aspects: First, by decoupling and reconstructing the implicit cost conduction chain in the process path (such as the increase in subsequent finishing time caused by the deviation of heat treatment temperature), the classifier can accurately quantify the influence weight of the elastic domain of process parameters on the total cost; Second, using the reshaping of feature phase significance to eliminate the misjudgment of process node similarity in the traditional vector space (such as the confusion between a common milling machine and a machining center when the surface roughness meets the standard), which helps to improve the semantic feature expression ability and discrimination of different process paths, providing a basis for subsequent process path recommendation.

[0042] In an embodiment of the present application, the alternative process path semantic feature compression subunit 1521 is configured to perform feature phase reconstruction and information compression processing on the alternative process path semantic embedding coding vector to obtain a set of alternative process path semantic feature compression local phase coding vectors, including: performing feature phase reconstruction based on one-dimensional convolutional coding on the alternative process path semantic embedding coding vector to obtain a set of alternative process path semantic feature local phase coding vectors; performing information compression on each alternative process path semantic feature local phase coding vector in the set of alternative process path semantic feature local phase coding vectors to obtain a set of alternative process path semantic feature compression local phase coding vectors.

[0043] Specifically, performing feature phase reconstruction based on one-dimensional convolutional coding on the alternative process path semantic embedding coding vector to obtain a set of alternative process path semantic feature local phase coding vectors, which is represented by the alternative process path feature phase reconstruction formula as:

[0044] Conv l×1 (X) = {x1, x2,..., x i ,..., x n}

[0045] Among them, X is the semantic embedding encoding vector of the alternative process path, and Conv l×1 is one-dimensional convolutional encoding processing, l is the feature phase reconstruction step size, and x1, x2, x i , x n are the 1st, 2nd, ith, and nth alternative process path semantic feature local phase encoding vectors in the set of alternative process path semantic feature local phase encoding vectors respectively. It should be understood that the static triple representation method of traditional knowledge graphs is difficult to effectively characterize the non-linear correlation between continuous features such as the dynamic range of process parameters and the elasticity of equipment capabilities and discrete process elements. Due to the high-order interaction characteristics of process nodes in the process path (such as the temporal dependence and resource competition constraints between multiple processes) being hidden in the local structure pattern of the alternative process path semantic embedding encoding vector, a feature phase reconstruction mechanism is required to analyze the relative relationship and change trend between adjacent dimensions inside it, so as to solve the problems of feature dimension redundancy and semantic mixing when heterogeneous information is mapped to a unified semantic space. The local feature pattern (such as the mutation edge of process parameters or the smooth adaptation trend of equipment capabilities) contained in the alternative process path semantic embedding encoding vector is captured through the one-dimensional convolutional sliding window mechanism, and the multi-scale convolutional kernel or stacked convolutional layer strategy is used to extract multi-angle local structure information from different receptive field ranges to construct a feature phase expression that can reflect the dynamic coupling effect of the process chain. The dynamic adaptation relationship between the elastic domain of process parameters and the threshold of equipment capabilities is transformed into an analyzable alternative process path semantic feature local phase encoding, and the implicit conduction logic between process-resource-cost in the process path is refined and characterized through the set of reconstructed alternative process path semantic feature local phase encoding vectors, providing an intermediate expression rich in structural diversity and semantic discriminability for subsequent feature interaction and path optimization, and further supporting the accurate quantification of the cost-driven weights of multiple paths by the classifier.

[0046] Specifically, information compression is performed on each alternative process path semantic feature local phase encoding vector in the set of alternative process path semantic feature local phase encoding vectors to obtain a set of alternative process path semantic feature compressed local phase encoding vectors, which is represented by the alternative process path semantic information compression formula as:

[0047]

[0048] Among them, ‖·‖ is the one-norm of the vector, and v iIt is the i-th alternative process path semantic feature compressed local phase encoding vector in the set of alternative process path semantic feature compressed local phase encoding vectors. It should be understood that there are still problems of dimensional redundancy and noise interference in the set of alternative process path semantic feature local phase encoding vectors after phase reconstruction. Specifically, irrelevant information such as non-critical tooling parameters and equipment idle status in the process path is mixed with core cost-driven features (such as high-precision process time-consuming and special material consumption) in the same vector space, resulting in a decrease in semantic discriminative power. Through the non-linear normalization and feature rectification mechanism, the information density of the reconstructed alternative process path semantic feature local phase encoding vectors can be optimized, redundant dimensions (such as repetitive process parameter fluctuations) can be suppressed, and noise components (such as unnecessary tool change records) can be eliminated, so as to focus on the effective features strongly related to cost in the process chain (such as equipment energy consumption mutation intervals and elastic thresholds of key process parameters). Through vector norm constraint and dynamic scaling factor adjustment, multi-dimensional features can be mapped to a low-dimensional compact space, while retaining the dynamic coupling effect of the process path (such as the influence weight of the temperature control accuracy of the heat treatment process on the subsequent processing cost), improving the distinguishability and robustness of the feature distribution, and providing a high-purity and low-redundancy semantic representation basis for subsequent gain factor calculation and path probability evaluation.

[0049] Specifically, the feature effective component statistical number calculation sub-unit 1522 is used to calculate the alternative process path semantic feature effective component statistical numbers of each alternative process path semantic feature compressed local phase encoding vector in the set of alternative process path semantic feature compressed local phase encoding vectors, and is represented by the alternative process path semantic feature effective component statistical number calculation formula:

[0050]

[0051] Among them, is the feature value at the j-th position in the i-th alternative process path semantic feature compressed local phase encoding vector, count i represents the effective component count, ε is a trainable preset threshold, en i is v iThe statistical count of the effective components of the corresponding alternative process path semantic features. It should be understood that although the redundant noise has been eliminated in the alternative process path semantic feature compressed local phase coding vector after information compression, it is still necessary to quantify the distribution density of the effective components (such as the elastic interval of key process parameters and the mutation threshold of equipment energy consumption) strongly related to cost drivers in each alternative process path semantic feature compressed local phase coding vector, so as to solve the problem of modeling deviation of the dynamic coupling effect of the process chain caused by fuzzy weight allocation in the feature enhancement stage. Through the adjacent dimension gradient mutation detection mechanism (such as determining that the jump amplitude of the eigenvalue exceeds the preset threshold), the number of effective dimensions with significant cost impact factors (such as the time fluctuation of high-precision machining processes and the clamping efficiency parameters of special fixtures) in each alternative process path semantic feature compressed local phase coding vector can be counted, so as to establish the mapping relationship between the feature contribution degree and the gain regulation. The effect is to accurately identify the implicit cost conduction nodes in the process path (such as the non-linear relationship between the temperature control accuracy of the heat treatment process and the qualification rate of subsequent finish machining) by quantifying the statistical distribution of the effective components, provide an interpretable quantitative basis for the calculation of the gain factor based on the holomorphic structure metric decomposition, ensure the significant enhancement of the key cost driver features (such as the weight of equipment depreciation cost and the fluctuation of material utilization rate) in the feature phase reshaping process, and then improve the decision-making robustness of the optimal path recommendation.

[0052] Figure 5 The block diagram of the feature phase reshaping gain factor calculation sub-unit in the intelligent cost assessment system for mechanical parts based on the knowledge graph according to the embodiment of the present application. As Figure 5 shown, in the embodiment of the present application, the feature phase reshaping gain factor calculation sub-unit 1523 is used to calculate the alternative process path semantic feature phase reshaping gain factors of each alternative process path semantic feature local phase coding vector based on the statistical count of the effective components of the alternative process path semantic features of each alternative process path semantic feature compressed local phase coding vector, including: the suppression factor determination secondary sub-unit 1523-1, which is used to determine the suppression factors corresponding to each alternative process path semantic feature compressed local phase coding vector based on the statistical count of the effective components of the alternative process path semantic features of each alternative process path semantic feature compressed local phase coding vector; the alternative process path semantic feature phase reshaping gain factor calculation secondary sub-unit 1523-2, which is used to calculate the alternative process path semantic feature phase reshaping gain factors of each alternative process path semantic feature local phase coding vector based on the suppression factors corresponding to each alternative process path semantic feature compressed local phase coding vector.

[0053] In an embodiment of the present application, the alternative process path semantic feature phase reshaping gain factor calculation secondary subunit 1523-2 is configured to calculate the alternative process path semantic feature phase reshaping gain factor of each alternative process path semantic feature local phase encoding vector based on the suppression factor corresponding to each alternative process path semantic feature compressed local phase encoding vector, including: calculating the initial alternative process path semantic feature phase reshaping gain factor of each alternative process path semantic feature local phase encoding vector based on the suppression factor corresponding to each alternative process path semantic feature compressed local phase encoding vector; and performing feature phase scatter missing correction on the initial alternative process path semantic feature phase reshaping gain factor to obtain the alternative process path semantic feature phase reshaping gain factor.

[0054] Specifically, the feature phase reshaping gain factor calculation subunit 1523 is represented by the feature phase reshaping gain factor calculation formula as:

[0055]

[0056] where n is the number of vectors in the set of alternative process path semantic feature local phase encoding vectors, θ i is the polar angle corresponding to v i λ is the suppression factor corresponding to v i π represents pi, arctan represents the arctangent function, e i is the initial alternative process path semantic feature phase reshaping gain factor, e(v i ) is the initial alternative process path semantic feature phase reshaping gain factor corresponding to v i K i is the alternative process path semantic space holomorphic flatness factor, T i is the alternative process path semantic flatness decomposition metric representation factor, e′ i is iis the semantic feature phase reshaping gain factor for the alternative process path. It should be understood that after the statistical analysis of the effective components, the local phase encoding vector of the semantic features of the alternative process path has not established a computable mapping mechanism between the feature enhancement weight and the dynamic coupling effect of the process chain (such as the non-linear relationship between the elastic range of equipment capabilities and the adaptation threshold of process parameters), resulting in the difficulty of accurately regulating the semantic expression intensity of the implicit cost conduction features in the process path (such as the exponential relationship between the spindle speed of a five-axis machining center and the energy consumption cost) by traditional linear scaling methods. By constructing a suppression factor calculation model under the holomorphic structure, the statistical number of effective components can be transformed into the symmetry constraint conditions of the feature phase direction (such as the geometric convexity preservation requirement of high-value feature dimensions), and the initial alternative process path semantic feature phase reshaping gain factor can be generated based on the non-linear saturation characteristics of the arctangent function. Then, the spatial projection deviation of the high-order interaction features (such as the combined impact of the special tooling replacement frequency on the multi-process pass rate) between the process parameter elastic domain and the equipment capability threshold during the gain allocation process can be eliminated through the feature phase scatter missing correction algorithm. In this way, the adaptive enhancement regulation of the dynamic cost-driven features of the process chain (such as the time fluctuation of key processes and the depreciation weight of high-precision measurement equipment) can be realized, enabling the alternative process path semantic feature phase reshaping gain factor to dynamically adjust the vector weight distribution of the feature phase space according to the effective component density, forming a non-linear enhancement strategy with process constraint perception ability, providing a differentiable mathematical framework for analyzing the multi-path cost game relationship to support subsequent feature saliency reshaping, and ultimately improving the cost conduction logic modeling accuracy of the classifier for the optimal process path.

[0057] And, for the compressed local phase encoding vector v of the semantic features of the alternative process path i The corresponding statistical number en of the effective components of the semantic features of the alternative process path i , when calculating the suppression factor λ corresponding to the compressed local phase encoding vector of the semantic features of the alternative process path i , if each compressed local phase encoding vector v of the semantic features of the alternative process path i is regarded as a set v of the compressed local phase encoding vectors of the semantic features of the alternative process path i (i = 1~n) based on the basic building units of the effective components, then the basic building units, as the core parameter combinations describing the multi-dimensional parameter space, let θ i = en i / ∑ i×1～n en i , and it is also expected that the polar angle representation θ i satisfies the direction equilibrium characteristic, so as to maintain the stability of the key feature distribution in the set space.

[0058] Therefore, if the set space is regarded as a hierarchical analysis model of the multi-dimensional feature space, the consistent expression of the multi-dimensional feature space can be obtained as:

[0059]

[0060] Then, enhance the effectiveness of local phase encoding as a normalization metric method under the hierarchical parsing model based on the multi-dimensional feature space to construct a single-mode interaction relationship into a normalized mapping field:

[0061]

[0062] That is, through the effectiveness enhancement representation of the alternative process path semantic feature phase reshaping gain factor e i to indicate that its individual mode is the optimal expression state under the multi-dimensional feature space consistency framework. In this way, the alternative process path semantic feature phase reshaping gain factor e i can be updated:

[0063]

[0064] Thus, while maintaining the set of alternative process path semantic feature phase reshaping gain factors e i under the feature direction equilibrium constraint of the multi-dimensional feature space stable framework, the loss of feature dimension correlation can be avoided in the subsequent feature coupling enhancement process based on feature phase significance reshaping, and the expression effect of the enhanced feature vector can be improved.

[0065] Specifically, the feature phase significance reshaping subunit 1524 is used to perform feature phase significance reshaping on the set of alternative process path semantic feature local phase encoding vectors based on the alternative process path semantic feature phase reshaping gain factors of each alternative process path semantic feature local phase encoding vector to obtain an alternative process path semantic enhanced encoding vector, which is expressed by the alternative process path feature phase significance reshaping formula as:

[0066]

[0067] where exp is the natural exponential function value with e as the base, a i is the alternative process path semantic feature phase reshaping gain weight, and v cIt is the semantic enhancement coding vector of the alternative process path. It should be understood that there is still a problem of feature distribution balance in the local phase coding vector of the semantic features of the alternative process path after the gain factor calculation. Specifically, the representation weights of process parameter dynamic adaptability (such as the negative correlation curve between cutting speed and tool life) and equipment capacity elasticity (such as the spindle speed threshold of a CNC machine tool) in the feature space do not accurately match the actual cost impact intensity. Through the holomorphic structure metric decomposition mechanism, the gain factor of the semantic feature phase reshaping of the alternative process path is applied to the symmetry calibration of the feature phase direction, dynamically adjusting the vector space projection intensity of the implicit cost conduction features (such as the influence coefficient of the temperature control accuracy of the heat treatment process on the finishing time) in the process chain, and at the same time suppressing the interference signals of non-critical tooling parameters (such as the replacement frequency of general fixtures). By constructing a cost-sensitive biased enhanced semantic space, the key process path nodes (such as the argon consumption of the high-precision laser cutting process) can obtain a geometric convexity expression in the features, and the process-resource coupling effect (such as the marginal balance between equipment utilization rate and energy consumption cost) can be computationally modeled through the reallocation of the phase direction weights. Finally, a high-discriminative semantic expression that can support the analysis of the multi-path cost game relationship is formed, providing a feature base with dynamic cost conduction modeling ability for the classifier to quantify the global optimal solution of the process chain.

[0068] In the embodiment of the present application, the optimal process path determination unit 153 includes: inputting a set of semantic enhancement coding vectors of alternative process paths into an optimal path intelligent evaluator based on a classifier to obtain a set of best path probability values; and taking the alternative process path corresponding to the maximum value in the set of best path probability values as the optimal process path.

[0069] Specifically, the set of semantic enhanced encoding vectors of alternative process paths is input into the optimal path intelligent evaluator based on a classifier to obtain a set of best path probability values. It should be understood that during the machining process of mechanical parts, the cost impact of different process paths not only involves direct economic indicators (such as equipment depreciation fees, energy consumption costs), but also implies implicit constraints such as the dynamic adaptability of process parameters (such as the negative correlation between cutting speed and tool life), and the cost of resolving resource conflicts (such as the scheduling waiting time of high-precision machine tools). Although these factors have been re-weighted by the feature phase reshaping gain operator in the semantic enhanced encoding vectors of alternative process paths, it is still necessary to quantify their combined effect through a probability model. Traditional process path recommendation methods are difficult to balance the non-linear game relationship between multi-dimensional cost driving factors. Based on this, in the technical solution of this application, the set of semantic enhanced encoding vectors of alternative process paths is further input into the optimal path intelligent evaluator based on a classifier to obtain a set of best path probability values. It should be understood that the optimal path intelligent evaluator based on a classifier analyzes the dynamic characteristics of the process chain embedded in the semantic enhanced encoding vectors of alternative process paths through a multi-layer perception mechanism (such as the influence weight of the temperature control accuracy of the heat treatment process on the qualified rate of subsequent machining), and combines the non-linear activation function (such as ReLU) in the fully connected layer to simulate the complex interaction between multi-cost elements (such as the marginal benefit balance between the improvement of equipment utilization rate and the increase in energy consumption cost), and finally maps the high-dimensional semantic features to path selection probability values.

[0070] Specifically, the alternative process path corresponding to the maximum value in the set of best path probability values is taken as the optimal process path. It should be understood that the path with the largest probability value represents a solution that is most likely to succeed under the current known conditions. This means that it is not only technically feasible but also the most economically reasonable. Of course, the so-called "most likely" does not mean there is no risk at all, but rather that this path has a lower risk and a greater chance of success compared to other paths. Recommending it as the optimal path to the decision-maker not only conforms to scientific principles but also has strong practical application value. Taking the alternative process path corresponding to the maximum value in the set of best path probability values as the optimal process path is not only based on an in-depth understanding and quantitative assessment of various cost drivers but also a scientific decision-making method achieved with the help of advanced mathematical models and algorithms. This method can not only effectively reduce production costs, improve product quality but also enhance the enterprise's adaptability and competitiveness in the complex and ever-changing market environment. By continuously optimizing and improving this process, the enterprise can not only improve its own operational efficiency and economic benefits but also better cope with future challenges and ensure long-term stable development. Specifically, as the best path probability values of each alternative process path are calculated, these probability values are sorted, and the alternative path corresponding to the maximum value is found. This maximum probability value means that this path has achieved cost optimization with the highest possibility in actual production, so it is considered the optimal process path. This optimal process path demonstrates the best balance and the lowest cost among its corresponding cost drivers and production factors, thus providing the most suitable processing plan for the subsequent part manufacturing process. For example, assume that during the evaluation of a specific mechanical part, three alternative process paths are generated. The best path probability value of path A is 0.85, the best path probability value of path B is 0.78, and the best path probability value of path C is 0.91. In this case, the probability value of path C is the largest, so the system selects path C as the optimal process path. According to the process requirements of path C, a precise cost prediction will be further generated, calculating the consumption of various resources and various costs such as time, labor, and equipment usage during the processing. Ultimately, the selection of the optimal process path not only optimizes the cost structure but also ensures high efficiency and high quality during the production process.

[0071] In the above-mentioned intelligent cost assessment system 100 for mechanical parts based on a knowledge graph, the cost prediction value module 160 is used to determine the cost prediction value of the mechanical parts to be evaluated based on the analysis results of the cost drivers of the optimal manufacturing process path. It should be understood that by deeply analyzing the processing technologies and paths adopted for the parts, the key factors affecting the cost can be systematically identified, and the impacts of these factors can be quantified through precise model calculations, and finally the cost prediction value of the mechanical parts to be evaluated can be determined. At this stage, the analysis of cost drivers mainly focuses on the use of various resources in the production process, specifically including equipment consumption, material consumption, tooling use, labor costs, energy consumption, etc. These cost drivers can be quantitatively analyzed through mathematical models. Among them, factors such as the use efficiency of equipment, price fluctuations of materials, and replacement frequency of tooling will directly affect the final cost. By analyzing the resource consumption and operation steps of the optimal manufacturing process path, the system can identify the key cost factors among them. More importantly, the analysis of cost drivers is not static but dynamically changing. Process parameters in the manufacturing process path, such as cutting speed, feed rate, machining depth, etc., will change their impacts on the cost as the part machining progresses. For example, during high-speed cutting, increasing the cutting speed can significantly improve production efficiency, but at the same time, it may also increase the energy consumption of the equipment or reduce the service life of the equipment, thus generating additional maintenance costs. Similarly, the choice of materials will also affect the processing cost. The cutting performance differences of different materials and the fluctuations in material costs will directly affect the overall production cost. By deeply analyzing these dynamically changing cost drivers, the system can more accurately predict the final cost of mechanical parts. The process of determining the cost prediction value is based on the analysis results of cost drivers and finally performs weighted calculations on various factors. Weighted calculations can individually evaluate each cost driver and also take into account the mutual influences and coupling effects between different factors. For example, there may be a mutual dependence relationship between the energy consumption and maintenance costs of the equipment. Certain process operations may cause additional wear of the equipment, thus increasing the long-term maintenance costs. In addition, changes in process parameters may also lead to waste of raw materials and increase production costs. By comprehensively analyzing and quantifying these factors, a corresponding cost value can be calculated for each optimal manufacturing process path, and finally the cost prediction value of the mechanical parts to be evaluated can be obtained. The obtained cost prediction value can not only provide an effective cost control basis for manufacturers, but also provide decision-making support for the optimization of production plans, the rationalization of resource allocation, and the improvement of production efficiency.

[0072] In summary, the intelligent cost assessment system 100 for mechanical parts based on a knowledge graph according to an embodiment of the present application is elucidated. It transforms discrete process parameters, equipment capabilities, and other process elements into multi-dimensional knowledge nodes through semantic modeling, and constructs a hypergraph structure based on process constraint relationships and cost conduction laws to form a full-link knowledge network covering "feature - process - resource - cost". Then, a hybrid embedding strategy is adopted to semantically represent multiple process paths, and the implicit coupling effect between different process paths is dynamically captured through an attention mechanism to strengthen the framework for optimizing the path selection strategy. Finally, a path probability evaluation model based on a classifier is used to quantify the cost driving weights of each path, realizing an adaptive recommendation of the global optimal solution for the process chain and breaking through the curse of dimensionality problem in traditional methods for dynamic process combination optimization.

[0073] As described above, the intelligent cost assessment system 100 for mechanical parts based on a knowledge graph according to an embodiment of the present application can be implemented in various terminal devices. In one example, the intelligent cost assessment system 100 for mechanical parts based on a knowledge graph can be integrated into a terminal device as a software module and / or a hardware module. For example, the intelligent cost assessment system 100 for mechanical parts based on a knowledge graph can be a software module in the operating system of the terminal device, or can be an application program developed for the terminal device; of course, the intelligent cost assessment system 100 for mechanical parts based on a knowledge graph can also be one of many hardware modules of the terminal device.

[0074] Alternatively, in another example, the intelligent cost assessment system 100 for mechanical parts based on a knowledge graph and the terminal device can also be separate devices, and the intelligent cost assessment system 100 for mechanical parts based on a knowledge graph can be connected to the terminal device through a wired and / or wireless network and transmit interaction information in accordance with a predefined data format. Figure 6 It is a flowchart of an intelligent cost assessment method for mechanical parts based on a knowledge graph according to an embodiment of the present application. As Figure 6As shown, the intelligent cost evaluation method for mechanical parts based on a knowledge graph according to an embodiment of the present application includes: S110, inputting mechanical part features and process features; S120, constructing a mechanical part cost evaluation knowledge graph oriented to process features based on the mechanical part features and process features; S130, obtaining feature data of the mechanical part to be evaluated; S140, performing path reasoning in the mechanical part cost evaluation knowledge graph oriented to process features according to the feature data of the mechanical part to be evaluated to obtain a set of alternative process paths and a set of cost driver factor analysis results corresponding to the set of alternative process paths; S150, inputting the set of alternative process paths into an optimal process recommendation module to obtain an optimal process path, including: the optimal process recommendation module uses a semantic-level feature analysis method to perform optimal path intelligent evaluation and optimal process recommendation on the set of alternative process paths to obtain an optimal process path; S160, determining the cost prediction value of the mechanical part to be evaluated based on the cost driver factor analysis result of the optimal process path.

[0075] Here, those skilled in the art can understand that the specific operations of each step in the above intelligent cost evaluation method for mechanical parts based on a knowledge graph have been introduced in detail in the description of the Figures 1 to 5 intelligent cost evaluation system for mechanical parts based on a knowledge graph above, and therefore, the repeated description thereof will be omitted.

[0076] In summary, the intelligent cost evaluation method for mechanical parts based on a knowledge graph according to an embodiment of the present application is elucidated. It transforms discrete process elements such as process parameters and equipment capabilities into multi-dimensional knowledge nodes through semantic modeling, and constructs a hypergraph structure based on process constraint relationships and cost conduction laws to form a full-link knowledge network covering "feature - process - resource - cost". Then, a hybrid embedding strategy is adopted to perform semantic representation on multiple process paths, and the implicit coupling effect between different process paths is dynamically captured through an attention mechanism to strengthen the frame optimization path selection strategy. Finally, a path probability evaluation model based on a classifier is used to quantify the cost driver weights of each path, realizing the adaptive recommendation of the global optimal solution of the process chain and breaking through the dimensionality disaster problem in the dynamic process combination optimization of traditional methods.

Claims

1. An intelligent cost evaluation system for mechanical parts based on a knowledge graph, characterized in that Including: A data input module for inputting mechanical part features and process features; A knowledge graph construction module for constructing a mechanical part cost evaluation knowledge graph oriented to process features based on the mechanical part features and the process features; A module for obtaining features of a mechanical part to be evaluated, which is used to obtain feature data of the mechanical part to be evaluated; A module for reasoning about the path of the mechanical part to be evaluated, which is used to perform path reasoning in the mechanical part cost evaluation knowledge graph oriented to process features according to the feature data of the mechanical part to be evaluated to obtain a set of alternative process paths and a set of analysis results of cost driving factors corresponding to the set of alternative process paths; An optimal process path recommendation module for inputting the set of alternative process paths into an optimal process recommendation module to obtain an optimal process path, including: the optimal process recommendation module uses a semantic-level feature analysis method to perform intelligent evaluation of the optimal path and recommend the optimal process for the set of alternative process paths to obtain the optimal process path; A cost prediction value module for determining a cost prediction value of the mechanical part to be evaluated based on the analysis result of the cost driving factors of the optimal process path.

2. The intelligent cost evaluation system for mechanical parts based on a knowledge graph according to claim 1, wherein The mechanical part features include geometric features, material features and technical requirements; the process features include process type, process parameter range, required equipment, tooling and cycle time.

3. The intelligent cost evaluation system for mechanical parts based on a knowledge graph according to claim 2, wherein The mechanical part cost evaluation knowledge graph oriented to process features includes entities and relationships. The entities include mechanical parts, materials, processes, equipment, tooling, process parameters, cost elements and cost indicators; the relationships include part - adopt material - material, part - go through process - process, process - use equipment - equipment, process - need tooling - tooling, process - control parameter - process parameter, part - generate cost - cost element, cost element - affect cost indicator - cost indicator and process - affect cost element - cost element.

4. The intelligent cost evaluation system for mechanical parts based on the knowledge graph according to claim 1, characterized in that The optimal process path recommendation module includes: An alternative process path semantic embedding coding unit for performing semantic embedding coding on each alternative process path in the set of alternative process paths to obtain a set of alternative process path semantic embedding coding vectors; A process path semantic expression strengthening unit for strengthening the process path semantic expression of each alternative process path semantic embedding coding vector in the set of alternative process path semantic embedding coding vectors to obtain a set of alternative process path semantic strengthened coding vectors; An optimal process path determination unit for performing intelligent evaluation of the optimal path based on the set of alternative process path semantic strengthened coding vectors to determine the optimal process path.

5. The intelligent cost evaluation system for mechanical parts based on the knowledge graph according to claim 4, wherein The process path semantic expression strengthening unit includes: An alternative process path semantic feature compression sub-unit for performing feature phase reconstruction and information compression processing on the alternative process path semantic embedding coding vector to obtain a set of alternative process path semantic feature compression local phase coding vectors; A feature effective component statistic calculation subunit, configured to calculate the feature effective component statistics of each alternative process path semantic feature compressed local phase encoding vector in the set of alternative process path semantic feature compressed local phase encoding vectors; A feature phase reshaping gain factor calculation subunit, configured to calculate the feature phase reshaping gain factor of each alternative process path semantic local phase encoding vector based on the feature effective component statistics of each alternative process path semantic feature compressed local phase encoding vector; A feature phase significance reshaping subunit, configured to perform feature phase significance reshaping on the set of alternative process path semantic local phase encoding vectors based on the feature phase reshaping gain factor of each alternative process path semantic local phase encoding vector to obtain the alternative process path semantic enhanced encoding vectors.

6. The intelligent cost evaluation system for mechanical parts based on a knowledge graph according to claim 5, wherein, The alternative process path semantic feature compression subunit includes: Performing feature phase reconstruction based on one-dimensional convolutional encoding on the alternative process path semantic embedded encoding vector to obtain a set of alternative process path semantic feature local phase encoding vectors; Performing information compression on each alternative process path semantic feature local phase encoding vector in the set of alternative process path semantic feature local phase encoding vectors to obtain the set of alternative process path semantic feature compressed local phase encoding vectors.

7. The intelligent cost evaluation system for mechanical parts based on the knowledge graph according to claim 6, wherein The feature phase reshaping gain factor calculation subunit includes: An inhibition factor determination secondary subunit, configured to determine the inhibition factor corresponding to each alternative process path semantic feature compressed local phase encoding vector based on the feature effective component statistics of each alternative process path semantic feature compressed local phase encoding vector; An alternative process path semantic feature phase reshaping gain factor calculation secondary subunit, configured to calculate the feature phase reshaping gain factor of each alternative process path semantic local phase encoding vector based on the inhibition factor corresponding to each alternative process path semantic feature compressed local phase encoding vector.

8. The intelligent cost evaluation system for mechanical parts based on a knowledge graph according to claim 7, characterized in that, The alternative process path semantic feature phase reshaping gain factor calculation secondary subunit includes: Calculating the initial alternative process path semantic feature phase reshaping gain factor of each alternative process path semantic local phase encoding vector based on the inhibition factor corresponding to each alternative process path semantic feature compressed local phase encoding vector; Performing feature phase dispersion missing correction on the initial alternative process path semantic feature phase reshaping gain factor to obtain the alternative process path semantic feature phase reshaping gain factor.

9. The intelligent cost evaluation system for mechanical parts based on a knowledge graph according to claim 8, wherein The optimal process path determination unit includes: Inputting the set of alternative process path semantic enhanced encoding vectors into an optimal path intelligent evaluator based on a classifier to obtain a set of best path probability values; Taking the alternative process path corresponding to the maximum value in the set of best path probability values as the optimal process path.

10. An intelligent cost evaluation method for mechanical parts based on a knowledge graph, characterized in that, Includes: Inputting mechanical part features and process features; Construct a knowledge graph for mechanical part cost assessment oriented to process features based on the mechanical part features and the process features; Obtain the feature data of the mechanical part to be evaluated; According to the feature data of the mechanical part to be evaluated, perform path reasoning in the knowledge graph for mechanical part cost assessment oriented to process features to obtain a set of alternative process paths and a set of analysis results of cost drivers corresponding to the set of alternative process paths; Input the set of alternative process paths into the optimal process recommendation module to obtain the optimal process path, including: the optimal process recommendation module uses a semantic-level feature analysis method to perform intelligent evaluation of the optimal path and recommend the optimal process for the set of alternative process paths to obtain the optimal process path; Determine the cost prediction value of the mechanical part to be evaluated based on the analysis result of the cost driver of the optimal process path.

Images