Industrial design material intelligent adaptation type selection system and integrated application method of multi-scene structure device

Abstract

The invention provides an industrial design material intelligent adaptation type selection system and an integrated application method of a multi-scene structure device, and is applied to the technical field of data processing. According to the method, full-process intelligent processing is carried out around industrial design material type selection, standardization preprocessing and quantitative verification are carried out on design requirement parameters, consistency is evaluated through a deviation coefficient, and a standard data set is generated; based on a deep learning matching model, full-automatic or auxiliary material matching is realized, a candidate list and an adaptation score are output, and a material-demand corresponding data set is constructed; an XGBoost network is improved based on TensorFlow, the precision is improved through feature optimization and the like, and an adaptive sequence and a feasibility probability are generated; determining a final recommendation result through a multi-stage weight fusion strategy; and finally, supporting interactive adjustment based on a visual platform, and outputting a model selection report and a parameter configuration file which can be exported.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an integrated application method of an intelligent material adaptation and selection system for industrial design and a multi-scenario structural device. Background Technology

[0002] Currently, there are still many technical bottlenecks in the selection of materials for industrial design: First, the requirements parameters lack standardized processing, with inconsistent parameter formats and insufficient quantification of key indicators in different design scenarios, resulting in low accuracy of parameter matching and verification, and easy deviations in material and requirement adaptation. Second, the selection process relies heavily on the designer's experience, and is subject to cognitive biases such as availability heuristic bias and anchoring effect, with a decision error rate as high as 42%, and it is difficult to cope with the dimensional challenges of multi-objective optimization and cannot efficiently balance the constraints of multiple scenarios. Third, existing intelligent selection technologies often suffer from insufficient data quality, high algorithm complexity, and low computational efficiency, and lack systematic application of feature engineering optimization and sample balancing, making it difficult to meet the adaptation accuracy requirements of high-end manufacturing. Fourth, the selection results lack a dynamic adjustment mechanism, and cannot optimize weight allocation in real time according to changes in scenario characteristics or material performance, and the visualization interaction and report export functions are incomplete, making it difficult to support designers in efficiently completing parameter adjustments and solution confirmation.

[0003] The current state of the industry, including heightened risks in the supply chain of critical materials and stricter environmental policy constraints, places higher demands on the scientific and efficient selection of materials. Traditional trial-and-error and experience-based methods suffer from drawbacks such as high time costs, significant resource waste, and poor compatibility, and can no longer meet the needs of modern industrial design for rapid response, precise matching, and low-cost material selection.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0006] According to one aspect of this application, an integrated application method for an intelligent material adaptation and selection system for industrial design and a multi-scenario structural device is provided, comprising: preprocessing industrial design requirement parameters, uniformly converting them into a standardized data format and quantifying key indicators; completing parameter matching verification with product functional requirements as the core reference and scenario usage conditions as constraints; evaluating parameter consistency through deviation coefficients; and generating a standardized requirement parameter dataset; processing the standardized requirement parameter dataset based on a deep learning matching model, supporting fully automated intelligent matching or designer-assisted matching with minor parameter adjustments; generating a candidate material list and adaptability score; and then extracting core performance indicators and scenario adaptation key points and synchronously associating them with material characteristic parameters to generate a material... - Demand-based dataset; The material-demand mapping dataset is processed using an improved XGBoost evaluation network based on the TensorFlow framework. Feature engineering optimization, sample balancing, and multi-level loss functions are employed to improve adaptation accuracy, generating a comprehensive material adaptation ranking result and multi-dimensional feasibility probability values. The final material adaptation recommendation result is generated through a combination of early-stage functional weight allocation, mid-stage scenario characteristic weighted fusion, or late-stage dynamic weight adjustment voting / gradient boosting fusion strategy based on material performance. The adaptation indicators, performance parameters, and recommendation results are processed using a client-side visual interactive platform, supporting designers to interactively modify parameters and scenario constraints, and generating exportable material selection reports and parameter configuration files.

[0007] According to another aspect of this application, an electronic device includes: a first processor; and a memory for storing executable instructions of the first processor; wherein the first processor is configured to execute the above-described integrated application method in an industrial design material intelligent adaptation and selection system and a multi-scenario structural device by executing the executable instructions.

[0008] According to another aspect of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a second processor, implements the above-described integrated application method in an intelligent material matching and selection system for industrial design and a multi-scenario structural device.

[0009] This application provides an intelligent material matching and selection system for industrial design and an integrated application method for multi-scenario structural devices. First, it standardizes the format, quantifies indicators, performs hierarchical verification, conducts consistency assessment, and corrects anomalies for multi-dimensional heterogeneous design requirement parameters, generating a standardized requirement parameter dataset. Then, through collaborative decision-making using a deep learning matching model and an adaptation rule base, and after classification retrieval, adaptation verification, and scoring calculation, a material-requirement correspondence dataset is generated. Relying on an improved XGBoost evaluation network under the TensorFlow framework, combined with feature engineering optimization, sample balancing, and multi-level loss functions, the adaptation accuracy is improved, outputting a comprehensive material adaptation ranking and multi-dimensional feasibility probabilities. Through a multi-stage fusion strategy involving early functional weight allocation, mid-term scenario characteristic weighted fusion, and late-stage dynamic weight adjustment, the final recommendation result is determined. Finally, based on a visual interactive platform supporting interactive modification of parameters and scenario constraints, an exportable selection report and parameter configuration file are generated, achieving an organic combination of machine intelligence and human experience.

[0010] Standardized preprocessing addresses the issues of inconsistent requirement parameter formats and insufficient quantification. Through dual validation and tiered anomaly handling, it ensures data standardization and integrity, laying the foundation for accurate selection. The collaborative application of deep learning matching and an improved XGBoost evaluation network, combined with a multi-stage weight fusion strategy, significantly improves material matching accuracy and efficiency, reducing decision-making error rates. Supporting both fully automated and interactive adjustment modes, it meets the needs of efficient selection while providing designers with flexible intervention space to adapt to personalized selection requirements in different scenarios. The visualization platform and standardized report and configuration file export functions enable traceability and data reusability in the selection process, reducing communication costs and contributing to the standardization and efficiency of industrial design processes.

[0011] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0012] Figure 1 This document illustrates a flowchart of an integrated application method for an intelligent material adaptation and selection system for industrial design and a multi-scenario structural device, as provided in an embodiment of this application. Detailed Implementation

[0013] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0014] The following is combined with Figure 1This application describes an integrated application method of an intelligent material adaptation and selection system and a multi-scenario structural device in industrial design, based on exemplary embodiments of this application. It should be noted that the application scenarios described below are merely illustrative for the purpose of understanding the spirit and principles of this application, and the embodiments of this application are not limited in any way. Rather, the embodiments of this application are applicable to any suitable scenario.

[0015] In one implementation, Figure 1 The schematic diagram illustrates a process flow of an integrated application method for an intelligent material adaptation and selection system and a multi-scenario structural device for industrial design, according to an embodiment of this application, including: S101 preprocesses the industrial design requirement parameters, converts them into a standardized data format, quantifies key indicators, uses product functional requirements as the core reference and scenario usage conditions as constraints to complete parameter matching verification, evaluates parameter consistency through deviation coefficients, and generates a standardized requirement parameter dataset.

[0016] In one implementation, considering the multi-dimensional and heterogeneous nature of industrial design requirements parameters, which include functional indicators, scenario constraints, and cost thresholds, the data format is first standardized through unified format conversion. Then, mechanical properties, weather resistance, processing feasibility, cost range, and environmental protection level are quantified to clarify the quantification standards and precision ranges of each parameter. Industrial design requirements parameters exhibit multi-dimensional and heterogeneous characteristics. Functional indicators are often described in words, such as "possessing good wear resistance," scenario constraints are mostly numerical ranges, such as "operating temperature -10℃ to 50℃," and cost thresholds are mostly range statements, such as "cost per gram not exceeding 5 yuan." The presentation formats of different types of parameters differ significantly, directly affecting the uniformity and efficiency of subsequent data processing. Therefore, it is necessary to establish a unified data format specification covering three core parameters to achieve structured standardization of heterogeneous data.

[0017] Functional indicators are presented in a three-part structure: "Indicator Name - Quantification Dimension - Target Value". The "Indicator Name" clarifies the core functional requirements (e.g., wear resistance, compressive strength); the "Quantification Dimension" defines the measurement standard (e.g., wear amount, compressive strength); and the "Target Value" specifies the functional achievement boundary, such as wear amount ≤ 0.05mm / 1000 cycles. Scenario constraints are presented in a standardized manner: "Usage Environment - Constraint Type - Limitation Range". "Usage Environment" distinguishes between indoor, outdoor, high-temperature, and humid application scenarios; "Constraint Type" specifies influencing factors such as temperature, humidity, and ultraviolet radiation; and "Limitation Range" provides a specific numerical range, such as outdoor scenario - ultraviolet radiation intensity ≥ 3000μW / cm². Cost thresholds are entered using a standardized format: "Budget Type - Amount Range - Unit". "Budget Type" is divided into categories such as R&D samples and mass production; "Amount Range" specifies the upper and lower limits of the cost; and "Unit" is uniformly set to yuan / g to ensure the comparability of cost parameters.

[0018] To transform unstructured data into a structured format, a semantic parsing module from natural language processing (NLP) is introduced. This module integrates part-of-speech tagging, entity recognition, and semantic mapping. First, keywords are extracted from the unstructured requirement description; for example, "impact" and "moderate intensity" are extracted from "needs to withstand moderate impact." Then, a pre-defined semantic mapping dictionary transforms the ambiguous expression into precise quantification dimensions and target values ​​(e.g., "moderate impact" corresponds to "tolerance value -10-20 kJ / m²"). Finally, structured data conforming to a unified format is output. This process completely solves the problems of chaotic requirement parameter formats and ambiguous expressions, ensuring that all input data has a consistent structure and clear semantics, laying the foundation for subsequent quantification processing and matching verification.

[0019] The quantification of core indicators is a key prerequisite for achieving intelligent material adaptation. It is necessary to formulate clear and feasible quantitative standards and accuracy ranges for five core dimensions: mechanical properties, weather resistance, processing feasibility, cost range, and environmental protection level, to ensure that each parameter is calculable and comparable.

[0020] Mechanical properties focus on the structural load-bearing requirements of materials in industrial design, quantifying them from three key dimensions: tensile strength, compressive strength, and impact strength. Tensile strength is quantified within a range of 50-1000 MPa, covering common materials from plastics (50-100 MPa) to high-strength alloys (800-1000 MPa); compressive strength is quantified within a range of 80-1500 MPa, adapting to different pressure-bearing requirements in structural components and supports; impact strength is quantified within a range of 5-50 kJ / m², distinguishing the performance differences between brittle materials (5-10 kJ / m²) and tough materials (30-50 kJ / m²). All three values ​​are accurate to two decimal places, with units standardized as MPa (tensile / compressive) and kJ / m² (impact), ensuring consistent data measurement.

[0021] Weather resistance is quantified and graded according to three dimensions: high and low temperature tolerance range, humidity adaptability range, and UV aging resistance, addressing the stability requirements of materials under different usage environments. The high and low temperature tolerance range is set at -20℃ to 150℃, with an accuracy controlled within ±1℃, suitable for indoor normal temperature (10℃-30℃), outdoor extreme environments (-20℃-60℃), and industrial high-temperature scenarios (80℃-150℃). The humidity adaptability range is 30%-90%RH, with an accuracy of ±5%RH, covering different humidity environments such as dry (30%-50%RH) and humid (60%-80%RH). UV aging resistance is graded from 1 to 5, with level 1 being low resistance (suitable for indoor light-protected scenarios) and level 5 being high resistance (suitable for outdoor long-term exposure scenarios), meeting the anti-aging requirements in different scenarios.

[0022] Processing feasibility is assessed based on the material's suitability for manufacturing, using a 0-10 score scale across three dimensions: processing difficulty, molding cycle, and process adaptability, with an accuracy of 0.1 points. Processing difficulty is scored according to the complexity of processes such as cutting, injection molding, and welding, with 0-3 points indicating easy processing, 4-7 points indicating medium difficulty, and 8-10 points indicating high difficulty. Molding cycle is scored based on the duration of a single production cycle, with short cycles (≤10 minutes) receiving high scores (8-10 points) and long cycles (≥60 minutes) receiving low scores (0-3 points). Process adaptability is scored based on the degree of compatibility between the material and common industrial processes, with higher scores indicating a greater number of compatible processes, ensuring that the quantitative results accurately reflect the material's production feasibility.

[0023] The cost range is quantified by unit weight cost, with an accuracy of 0.01 yuan / g, and a range of 0.1-100 yuan / g, covering low-cost plastics (0.1-1 yuan / g), medium-cost metals (5-20 yuan / g), and high-cost specialty materials (50-100 yuan / g), meeting the selection needs of different budget sizes. Environmental protection levels are strictly divided into 1-5 levels according to relevant national standards (such as GB / T26572-2011), with Level 1 being the highest environmental standard (degradable, no release of harmful substances), and Level 5 being the basic environmental standard (meeting the minimum requirements for safe use). The accuracy is Level 1, ensuring the compliance and authority of environmental indicators.

[0024] The above quantification rules are executed through an independent indicator quantification module. This module has a built-in quantification range mapping table and precision control logic for each indicator. It maps the structured requirement parameters to the quantification range according to the corresponding standards, while automatically checking the data precision, eliminating outliers that exceed the precision requirements, and clarifying the data boundaries of each parameter, thus providing accurate quantitative data support for subsequent verification and matching.

[0025] Using core product functional requirements as the primary reference and scenario usage conditions as rigid constraints, a layered verification process of basic parameter screening and core indicator adaptation is conducted to check the compliance of parameters with functional requirements and scenario constraints one by one. To ensure a high degree of compatibility between requirement parameters and product functions and scenario constraints, a two-layer verification mechanism of "basic parameter screening - core indicator adaptation" is constructed, progressively screening compliant parameters and eliminating conflicting and invalid data. The basic parameter screening layer, as the first line of defense in verification, focuses on the completeness and format compliance of parameters. In the verification process, the system automatically traverses all structured requirement parameters, checking one by one whether functional indicators conform to the format of "indicator name - quantitative dimension - target value", whether scenario constraints completely include the three elements of "usage environment - constraint type - limited range", and whether cost thresholds clearly define "budget type - amount range - unit". At the same time, it checks whether there is any missing key information for each parameter, such as functional indicators without clearly defined target values ​​or scenario constraints without defined limited ranges; and whether the format is disordered, such as mixed units or incorrect numerical range descriptions. For parameters with the above problems, the system directly marks them as invalid data and removes them to ensure that all parameters entering the subsequent stages have a complete structure and standardized format.

[0026] Building upon the basic screening, the core indicator adaptation layer conducts in-depth adaptation and verification of the quantified indicator data with product functions and scenario constraints. Using the product's core functional requirements as the primary reference, the system first extracts the product's core functional needs (structural components must be "load-bearing and pressure-resistant," and shell components must be "wear-resistant and scratch-resistant"), then verifies whether the corresponding quantified indicators meet the minimum functional requirements—for example, the tensile strength of load-bearing structural components must be ≥300MPa, and the compressive strength ≥500MPa; the impact strength of wear-resistant shell components must be ≥20kJ / m², and the wear amount ≤0.05mm / 1000 cycles. Simultaneously, using scenario usage conditions as rigid constraints, the system verifies whether the parameters match the environmental requirements of the application scenario—for example, outdoor products must meet high-temperature resistance ≥60℃ and UV aging resistance ≥3 levels; medical equipment accessories must meet environmental protection level ≥2 and humidity adaptability 30%-60%RH. During the verification process, if a parameter meets both the functional requirements and the scenario constraints, it is deemed compliant; if there is a single dimension or two dimensions that are not met (such as tensile strength meeting the standard but high temperature resistance being insufficient), it is marked as an adaptation conflict, triggering the subsequent exception handling process to ensure that the parameters ultimately retained are bidirectionally adapted to the product requirements and have no compliance contradictions.

[0027] A deviation coefficient evaluation mechanism is introduced to verify parameter consistency, while a parameter integrity threshold is set for dual verification. A tiered processing mechanism is established, with automatic correction for minor anomalies and manual intervention for severe anomalies. To further ensure data quality, a dual verification mechanism of "consistency-integrity" is introduced, along with tiered anomaly handling rules. Differentiated processing methods are adopted for anomalies of different degrees to ensure data reliability and integrity. Consistency verification is achieved through a deviation coefficient evaluation mechanism, which uses the "standard deviation / mean" calculation method. This algorithm can effectively reflect the dispersion of parameter groups and quantify the consistency level between parameters. The system first classifies the requirement parameters according to functional indicators, scenario constraints, and cost thresholds. For parameters of the same type in each category (such as all mechanical performance indicators and all temperature constraint parameters), a deviation coefficient is calculated. A default acceptable threshold of 0.3 is set—when the deviation coefficient is ≤0.3, it indicates that the parameter group has low dispersion and good consistency; when the deviation coefficient is >0.3, it indicates that there are significant differences between parameters and the consistency does not meet the standard (for example, if the tensile strength requirements for different parts of the same product are 100MPa and 500MPa respectively, the deviation coefficient is far greater than 0.3, which is a logical contradiction).

[0028] Integrity verification is controlled by parameter integrity thresholds, setting the standard for acceptance as the percentage of missing core indicators not exceeding 5% (i.e., integrity threshold ≥ 95%). The system automatically counts the number of missing core indicators (five categories including mechanical properties and weather resistance) and calculates the percentage of missing indicators. If the percentage of missing indicators is ≤ 3%, the integrity is considered good; if 3% < percentage of missing indicators ≤ 5%, it is marked as slightly missing; if the percentage of missing indicators is > 5%, the integrity is considered unacceptable.

[0029] Based on the results of dual verification, a tiered anomaly handling mechanism is established. Minor anomalies include single-parameter deviation coefficients between 0.3 and 0.5, and missing core indicators accounting for 1% to 3%. These anomalies are automatically corrected by the system: for minor consistency anomalies, discrete parameters are adjusted by filling in the parameter group mean and interpolating neighboring parameters to reduce the deviation coefficient to within the acceptable threshold; for minor integrity anomalies, missing indicators are supplemented by mapping typical parameters from similar scenarios. For example, if an outdoor product lacks the UV aging resistance parameter, it is supplemented according to the default level 3 for outdoor scenarios. Severe anomalies include deviation coefficients > 0.5 and missing core indicators accounting for > 3%. These anomalies may involve conflicting requirement logic or missing key information. The system automatically triggers a manual intervention process, pushing a list of abnormal parameters and problem descriptions to the designer, such as "tensile strength parameter deviation coefficient 0.6, indicating a logical contradiction" or "weather resistance indicator missing percentage 6%, requiring supplementary key data." The designer verifies and corrects these anomalies based on the actual product requirements, ensuring accurate handling of abnormal data and avoiding impact on subsequent selection results.

[0030] Through a comprehensive process encompassing format unification, indicator quantification, hierarchical verification, consistency assessment, and anomaly correction, a standardized requirement parameter dataset is generated, ensuring data standardization, parameter completeness, and material selection suitability. Upon initiation, the process first transforms heterogeneous raw requirements into structured data through format unification, providing a unified data structure foundation for subsequent quantification. The indicator quantification stage, based on the structured data, quantifies and calibrates core indicators according to preset standards, outputting calculable quantitative data to support hierarchical verification. The hierarchical verification stage performs compliance screening on the quantitative data, eliminating invalid formats and conflicting parameters, retaining valid data that meets functional and scenario requirements. The consistency and completeness assessment stage verifies the quality of valid data and identifies data anomalies. The anomaly correction stage automatically or manually addresses anomalies to rectify data quality issues. The output of each stage directly serves as the input for the next stage, forming a closed-loop mechanism of "input-processing-output-reprocessing," ensuring that each step is based on high-quality data.

[0031] After a complete processing cycle, a standardized requirement parameter dataset is generated. This dataset contains four core components: quantitative data of functional indicators (e.g., tensile strength 350MPa, impact strength 25kJ / m²), scenario constraint ranges (e.g., outdoor use environment - constraint type temperature - limit range - 20℃-60℃), cost threshold ranges (e.g., budget type mass production - amount range 2-5 yuan / g), and parameter verification records (e.g., format verification passed, consistency deviation coefficient 0.25, no abnormal correction records). The dataset also possesses three core characteristics: data standardization (all parameters have uniform format and consistent accuracy, with no heterogeneous expressions), parameter completeness (the proportion of missing core indicators is ≤3%, and no key information is omitted), and material selection adaptability (all parameters meet product functional requirements and scenario constraints, with no compliance conflicts). It can provide high-quality and highly reliable data input for subsequent intelligent matching of deep learning models and material evaluation, ensuring the accuracy and efficiency of the entire material selection process.

[0032] S102 processes standardized requirement parameter datasets based on deep learning matching models, supports fully automated intelligent matching or matching assisted by designers adjusting a small number of parameters, generates a candidate material list and adaptability score, extracts core performance indicators and key points of scenario adaptation and synchronously associates material characteristic parameters to generate a material-requirement correspondence dataset.

[0033] In one implementation, based on the indicator dimensional features of a standardized demand parameter dataset and the characteristics of a material library, a deep learning matching model and an adaptation rule base negotiate to determine an intelligent matching strategy. The matching retrieval scope and similarity threshold are determined through demand parameter priority parsing and material characteristic correlation statistics. The indicator dimensional features of the standardized demand parameter dataset cover the quantitative dimensions, numerical ranges, and accuracy requirements of three categories: function, scenario, and cost. The material library characteristics include core information such as the coverage of material performance parameters, characteristic labeling system, and data update frequency. Together, they constitute the basic data support for intelligent matching. To achieve a balance between data adaptability and industry rationality, a collaborative decision-making mechanism between the deep learning matching model and the adaptation rule base is constructed, forming a closed-loop logic of two-way interaction and mutual verification.

[0034] The deep learning matching model adopts a three-layer architecture: the feature extraction layer extracts high-dimensional features of requirement parameters and material properties through a convolutional neural network, realizing feature upscaling and redundant information removal from the original data; the dimensionality mapping layer uses a fully connected network to map requirement features and material features to the same dimensional space, eliminating the dimensional barriers of heterogeneous data; and the strategy generation layer optimizes the matching logic based on the gradient descent algorithm and outputs a preliminary matching strategy. The adaptation rule base contains three core categories of rules: industry selection specifications covering material selection standards in different fields such as mechanical manufacturing and electronic equipment; material compatibility taboo rules explicitly prohibiting the use of material combinations (such as materials with conflicting chemical properties); and parameter matching priority criteria defining the importance of each parameter in the selection process. During the collaboration process, the model first outputs preliminary matching logic based on data features, and the rule base then performs compliance verification on the logic according to established specifications. If conflicts exist (such as the model-recommended materials violating compatibility taboos), the matching logic is automatically adjusted, ultimately forming an intelligent matching strategy that balances data compatibility and industry practicality.

[0035] The priority analysis of requirement parameters employs the Analytic Hierarchy Process (AHP), constructing a three-layer structure: a target layer (material matching and selection), a criterion layer (function, scenario, cost), and an indicator layer (specific parameters). The weight of each layer is calculated. Functional indicators, directly determining core product performance, have a weight of 50%; scenario constraints affect material stability, with a weight of 30%; and cost thresholds relate to production economics, with a weight of 20%. This weighting forms the parameter priority sequence, ensuring core requirements are met first. Material property correlation analysis uses the Pearson correlation coefficient algorithm. By calculating the linear correlation between requirement parameters and various property parameters in the material library, material properties with a correlation ≥ 0.7 are included in the matching range, avoiding ineffective characteristics that reduce efficiency. The similarity threshold is dynamically differentiated. Core parameters (such as the tensile strength of structural components) directly affect product functionality, with a similarity threshold set ≥ 0.85. Secondary parameters (such as the processing cycle of decorative parts) have a relatively smaller impact, with a threshold set ≥ 0.7, ensuring accurate matching of core requirements while retaining some flexibility for secondary requirements.

[0036] A parameter classification retrieval mechanism is employed to split the standardized requirement parameter dataset, extracting core parameters based on functional indicators, scenario constraints, and cost thresholds. These parameters are then associated with corresponding material property matching tags to generate a preliminary candidate material set. This parameter classification retrieval mechanism precisely splits the standardized requirement parameter dataset based on the differences in the attributes of the requirement parameters, ensuring the targeting and efficiency of the retrieval process. The functional indicator category focuses on the core performance requirements of the product, extracting quantified core parameters such as tensile strength, compressive strength, weather resistance, and processing difficulty. These parameters directly determine whether the material can meet the product's functional requirements. The scenario constraint category revolves around the product's usage environment, extracting constraints such as indoor / outdoor conditions, high temperature, humidity, temperature range, humidity requirements, and UV radiation intensity, reflecting the external conditions the material must adapt to. The cost threshold category focuses on production economics, extracting economic parameters such as unit weight cost, budget range, and bulk purchase discounts to clarify the cost boundaries for material selection.

[0037] To achieve precise correlation between parameters and materials, a standardized material characteristic matching labeling system has been established. Each material corresponds to three types of labels: functional adaptation labels, scenario adaptation labels, and cost adaptation labels. All labels are generated based on the quantitative results of material performance parameters, ensuring the objectivity and traceability of the labels. Functional adaptation labels are generated based on parameters such as mechanical properties and weather resistance. For example, tensile strength ≥300MPa corresponds to a "high-strength label," and UV resistance level ≥4 corresponds to an "anti-aging label." Scenario adaptation labels are generated based on the material's applicable environment. For example, suitable for environments from -20℃ to 60℃ corresponds to a "wide-temperature adaptation label," and humidity resistance ≥85%RH corresponds to a "moisture-proof label." Cost adaptation labels are generated based on unit weight cost. For example, 0.1-1 yuan / g corresponds to a "low-cost label," 2-5 yuan / g corresponds to a "medium-cost label," and above 50 yuan / g corresponds to a "high-cost label."

[0038] During the search process, the system employs a multi-dimensional tag matching algorithm to achieve precise association: First, the split requirement parameters are converted into corresponding matching tags (e.g., requirement parameters "outdoor use, tensile strength ≥350MPa, cost 2-5 yuan / g" are converted into "UV resistant tag, high strength tag, medium cost tag"). Then, the system iterates through the material tags in the material library, filtering out materials with completely matching tags or matching core tags, forming a preliminary candidate material set. This set covers all materials that meet the basic compatibility conditions, laying the foundation for subsequent refined verification.

[0039] The initial candidate material set undergoes suitability verification. For materials with substandard performance parameters, a supplementary search service is initiated. A rule adjustment mechanism is triggered for materials with conflicting scenario compatibility. Samples with costs exceeding a threshold are subject to priority re-ranking. This process generates an optimized candidate material list that includes material selection strategies and compatibility correction schemes. The initial candidate material set must undergo comprehensive compatibility verification, establishing rigid verification standards around three core dimensions: performance, scenario, and cost. This ensures that the selected materials meet both data matching requirements and practical application needs. The verification process employs a closed-loop workflow of "verification-correction-supplementation," synchronously recording key information to ensure the traceability of verification results.

[0040] Performance parameter verification employs a threshold comparison method, comparing each material performance parameter with the quantitative standards of the required parameters one by one to clarify the judgment criteria: In mechanical properties, tensile strength and compressive strength must reach the lower limit of the required threshold, and impact strength must be within the range set in the requirements; in weather resistance, the high and low temperature tolerance range must completely cover the temperature range of the required scenario, and the UV aging resistance level must not be lower than the required level; in processing feasibility, the processing difficulty score must be ≤ the required threshold, and the molding cycle must be ≤ the maximum allowable cycle. If any core performance parameter of the material fails to meet the above requirements, it is judged as substandard, and the system automatically initiates a supplementary search service, using a similarity algorithm (calculating the similarity of material characteristic vectors) to search the material library for alternative materials with similar characteristics and meeting the performance requirements, ensuring that there are no shortcomings in performance dimensions.

[0041] The scenario adaptation verification employs a scenario constraint matching algorithm to construct a scenario-material property mapping relationship library. This clarifies the core requirements for materials in different scenarios. For example, outdoor scenarios require materials to have UV resistance and high / low temperature resistance, while medical scenarios require materials to have biocompatibility and environmental friendliness. During verification, the core constraints of the required scenario are first extracted, and then the material properties are checked to ensure they are fully compatible. If conflicts exist, a rule adjustment mechanism is triggered to re-search for materials that match the scenario adaptation tags, ensuring a high degree of compatibility between the materials and the usage environment.

[0042] Cost compatibility verification employs a range comparison method, precisely comparing the unit weight cost of materials with a threshold demand cost: if the material cost is within the threshold range, it is deemed cost compliant; if it exceeds the threshold, its compatibility priority is reduced proportionally, for example, exceeding the threshold by 10% lowers the priority by 1 level, and exceeding it by 50% or more lowers it by 3 levels. Materials with the lowest priority are directly removed from the candidate set to ensure the economic efficiency of the selected solution.

[0043] During the verification process, the system simultaneously records two core pieces of information: the material selection strategy, which specifies the concrete thresholds for performance screening, the core rules for scenario matching, and the calculation logic for cost ranking; and the adaptation and correction scheme, which details the retrieval algorithm for alternative materials and the calculation method for priority adjustment coefficients. The final optimized candidate material list must ensure that all materials meet the basic requirements of performance compliance, scenario adaptation, and cost control, providing high-quality materials for subsequent scoring and calculation.

[0044] The optimized candidate material list undergoes a suitability scoring process. A multi-dimensional weighted algorithm quantifies functional matching, scenario suitability, and cost controllability, simultaneously generating a material suitability scoring report and detailed feature associations. The suitability scoring calculation, centered on a multi-dimensional weighted algorithm, constructs a scientifically sound scoring system to quantitatively assess material suitability. It also generates detailed scoring reports and feature associations, providing data support for subsequent decision-making. The scoring system focuses on three core dimensions: function, scenario, and cost, balancing quantitative accuracy with logical rationality.

[0045] The functional matching score employs a parameter similarity weighted summation algorithm. First, the individual similarity of each functional parameter is calculated: for numerical parameters (e.g., tensile strength), similarity = 1 - |material parameter value - requirement parameter value| / requirement parameter value; the smaller the deviation, the closer the similarity is to 1. For graded parameters (e.g., weathering grade), similarity = 1 - |material grade - requirement grade| / requirement grade, ensuring a linear negative correlation between grade differences and similarity. Then, the individual similarity of each parameter is multiplied by its corresponding weight (e.g., tensile strength weight 20%, weathering grade weight 15%), and summed to obtain the total functional matching score (out of 100), which directly reflects the degree to which the material meets the core functions.

[0046] The scenario fit scoring employs a constraint-weighted algorithm based on the number of constraints satisfied. Scenario constraints are first divided into core and secondary constraints: core constraints (such as UV resistance requirements in outdoor scenarios) directly impact material stability, accounting for 60% of the weight; secondary constraints (such as minor moisture resistance requirements) have a relatively smaller impact, accounting for 40% of the weight. The scoring rule is set as follows: A perfect score of 100 points is achieved by covering all scenario constraints; 20 points are deducted for each missing core constraint, and 10 points are deducted for each missing secondary constraint, until all points are deducted. This rule highlights the importance of core scenario requirements while comprehensively considering the impact of secondary constraints.

[0047] The cost controllability score uses a range fit algorithm. If the material cost is within the demand threshold range, it receives a full score of 100 points. If it exceeds the range, points are deducted proportionally: ≤10% deduct 10 points, 10% < ≤30% deduct 30 points, 30% < ≤50% deduct 60 points, and >50% deduct 100 points. This scoring logic balances cost flexibility with strict constraints on materials that significantly exceed the budget, ensuring the economic efficiency of the selected materials.

[0048] The comprehensive adaptability score is calculated using a multi-dimensional weighted algorithm, with weights of 50% for functional matching, 30% for scenario suitability, and 20% for cost controllability. After scoring, two core documents are generated: a material adaptability score report clearly shows the individual score, overall score, ranking, and reasons for deductions for each material (e.g., "Functional matching 85 points, deducted 10 points due to tensile strength deviation; Cost controllability 90 points, deducted 10 points due to exceeding the budget by 5%"); and a characteristic correlation details documenting the correspondence between each material's performance parameters and the required parameters, adaptability highlights (e.g., "Weather resistance level 5, exceeding the required level by 1, suitable for long-term outdoor use"), and potential differences (e.g., "Processing cycle 8 minutes, slightly longer than the required threshold by 6 minutes"), providing designers with comprehensive decision-making references.

[0049] The system supports fully automated intelligent matching or designer-assisted matching with minor parameter adjustments. It extracts core performance indicators and key points for scenario adaptation, and synchronously associates them with material characteristic parameters to form a final material-requirement mapping dataset containing material information, compatibility scores, and characteristic correlation data. To balance selection efficiency and decision-making flexibility, the system is designed with both fully automated intelligent matching and designer interactive adjustment modes, achieving an organic combination of machine intelligence and human experience, ultimately generating a structured, high-quality material-requirement mapping dataset.

[0050] The fully automated mode uses the scoring results and detailed characteristic correlation as its core basis to automatically filter the top 20 materials with the highest comprehensive scores, ensuring the quality and conciseness of the recommended list. After filtering, the system extracts three types of core data: core performance indicators, including specific values ​​of key parameters such as tensile strength, compressive strength, and weather resistance, which intuitively reflect the material's performance level; key points for scene adaptation, which clearly define the environmental type and constraint compliance of the material, such as "suitable for outdoor scenes, meeting the constraints of high temperature 60℃ and ultraviolet intensity 3000μW / cm²", clearly presenting the advantages of scene adaptation; and complete material characteristic parameters, covering basic attributes (material name, model, manufacturer), production process (adaptive processes such as injection molding and cutting), supply cycle (regular supply cycle, expedited supply plan), and after-sales guarantee (quality warranty period, repair service scope), providing support for subsequent production implementation.

[0051] The designer-interactive adjustment mode focuses on supplementing human experience, providing a visual parameter adjustment interface that supports three core operations: modifying core parameter weights (e.g., for precision instrument products, increasing the functional indicator weight to 60% and decreasing the cost threshold weight to 10%), adjusting cost thresholds (e.g., adjusting the unit weight cost threshold from 2-5 yuan / g to 2-8 yuan / g due to increased R&D budget), and adding scenario constraints (e.g., adding a "chemical corrosion resistance" scenario constraint). The system responds to each adjustment operation in real time, quickly recalculating the suitability score and candidate material list through a real-time response recalculation algorithm, ensuring immediate feedback of adjustment results. This algorithm, based on pre-compiled calculation logic and a caching mechanism, avoids repeatedly traversing the entire dataset, significantly improving recalculation efficiency and ensuring smooth interaction. Designers can further filter based on their professional experience and the adjusted list to meet personalized selection needs.

[0052] Both models ultimately extract core data and generate a material-demand mapping dataset with a unified structure. The dataset contains three core modules: a basic material information module recording material name, model, manufacturer, contact information, and other basic content; a compatibility scoring module covering individual scores, overall scores, and rankings for functional matching, scenario suitability, and cost controllability; and a characteristic correlation data module detailing performance parameter correspondences, scenario suitability details, cost suitability, suitability highlights, and potential differences. This dataset has a well-structured and comprehensive information, meeting the input requirements for subsequent improvements to the XGBoost evaluation network and providing designers with traceable and analyzable complete selection data support.

[0053] S103 uses an improved XGBoost evaluation network based on the TensorFlow framework to process the material-demand correspondence dataset. It employs feature engineering optimization, sample balancing, and multi-level loss functions to improve the fitting accuracy and generate a comprehensive material fit ranking result and multi-dimensional feasibility probability values.

[0054] In one implementation, the adaptation feature complexity of the material-demand mapping dataset is processed, and core evaluation indicators, including the functional matching accuracy threshold and scenario constraint fit conditions, are identified and categorized. Simultaneously, the computing power resources of the TensorFlow framework are quantitatively evaluated, generating feature complexity classification results and computing power resource evaluation data. Key evaluation indicators are identified and quantitative standards are established, while the computing power resource carrying capacity is accurately measured, providing data support for subsequent training strategy formulation. The core evaluation indicators focus on the functional matching accuracy threshold and scenario constraint fit conditions. The functional matching accuracy threshold is divided into levels based on the actual fit deviation between material performance parameters and demand parameters. A high level requires a deviation ≤ 5%, suitable for scenarios with stringent performance requirements, such as precision instruments; a medium level has a deviation of 5% < deviation ≤ 10%, suitable for general industrial products; and a low level has a deviation > 10%, only applicable to non-core components with relaxed performance requirements. Scenario constraint fit conditions are categorized by constraint type. Environmental constraints cover external environmental factors such as temperature, humidity, and UV intensity; operational constraints include working condition factors such as usage intensity, operating frequency, and load pressure. Each type of constraint specifies concrete adaptation requirements. For example, in the environmental constraint category, outdoor scenarios require a UV resistance intensity ≥3000μW / cm², and in the operational constraint category, high-frequency usage scenarios require a fatigue strength ≥10. 6 Second-rate.

[0055] Feature complexity quantification is calculated comprehensively across three dimensions: the number of feature dimensions (statistics include the total number of features such as performance, scenario, and cost in the dataset); parameter correlation density (measured by the mean of the Pearson correlation coefficients between features); and numerical distribution dispersion (reflected by the ratio of the standard deviation to the mean, indicating the uniformity of data distribution). Each dimension is scored from 0-40, 0-30, and 0-30 respectively. A combined score ≤30 indicates simple complexity, 30 < score ≤ 60 indicates medium complexity, and a score > 60 indicates complex complexity, generating a feature complexity classification result.

[0056] Meanwhile, a computing power benchmarking algorithm was used to quantitatively evaluate the computing power resources of the TensorFlow framework. The core evaluation metrics included data processing throughput (unit: samples / second), model training concurrency (unit: number of tasks), and memory usage limit (unit: GB). By running a standard test dataset, the number of samples processed per unit time, the maximum number of tasks trained in parallel simultaneously, and the maximum memory consumption per training round were recorded to generate computing power resource evaluation data, clarifying the upper limit of computing power (such as the maximum number of samples that can be processed in a single round and the maximum supported feature dimension) and performance bottlenecks (such as memory limitations when processing high-dimensional data and the throughput reduction problem during concurrent training).

[0057] Based on feature complexity classification results and computational resource assessment data, a supply-demand balancing algorithm is used to dynamically adapt the training strategy. The core of this approach is balancing feature processing requirements with computational resource supply. When feature complexity is high and computational resource is sufficient, a refined training strategy is employed. This strategy includes multiple rounds of iterative optimization (adjusting feature weights and optimizing the loss function in each iteration) and deep mining of high-dimensional features (extracting hidden feature associations through multi-layer networks) to ensure that the adaptation details of complex features are fully captured. When feature complexity is low or computational resource is limited, an efficient training strategy is adopted. This strategy simplifies the feature processing flow (skipping redundant feature cross-cutting and reducing the number of feature dimensionality reduction layers) and shortens the iteration cycle (reducing the total number of iterations and increasing the learning rate), thereby improving training efficiency while maintaining basic adaptation accuracy.

[0058] Based on the principle of full coverage of core features, the mutual information entropy algorithm is used to calculate the correlation between each feature and the fitting result. This algorithm quantifies the importance of features by measuring the information gain between the feature and the fitting result. Features with a correlation degree ≥ 0.8 are defined as core features, such as the tensile strength of structural components and the UV resistance rating of outdoor products. These features directly determine the accuracy of the fitting result and are all included in the model input. Features with a correlation degree between 0.5 and < 0.8 are secondary features, such as the processing cycle of decorative parts and the supply cycle of low-cost products. These are selected as needed according to the training strategy (included for refined training strategies and selectively removed for efficient training strategies). Features with a correlation degree < 0.5 are redundant features, such as the manufacturer's address of materials and non-critical process parameters, and are directly removed. Through this selection logic, the range of model input data is clearly defined, ensuring that key information is not missing while avoiding redundant data from increasing the training burden. The final result is the definition of the model training strategy and the range of input data.

[0059] Based on the defined range of input data, systematic feature engineering optimization was performed on the dataset to improve feature quality and model adaptability through multi-step processing. Principal component analysis (PCA) was used for feature dimensionality reduction. This algorithm maps high-dimensional features to a low-dimensional space through orthogonal transformation, retaining principal components with a cumulative contribution rate ≥90%. While reducing feature dimensionality (lowering model training complexity), it retains the key information of the original features to the greatest extent. Feature normalization was performed, using the min-max normalization method to map all feature parameters to the [0,1] interval, eliminating the influence of numerical differences caused by different units (such as MPa, yuan / g, grade), and ensuring that each feature has equal weight in model training. Feature cross-multiplication algorithms were used to construct combined features, such as cross-multiplication of "tensile strength" and "weather resistance grade" to generate the combined feature "high strength anti-aging", strengthening the correlation information between key parameters, helping the model capture complex adaptation logic, and improving feature extraction capabilities.

[0060] To address the potential issue of imbalanced sample distribution in the dataset (a sample ratio ≥70% for a certain adaptation scenario can lead to biased learning), a sample balancing method combining stratified sampling and synthetic sampling is employed. For minority class samples (<20%), the SMOTE algorithm is used to synthesize new samples by interpolating virtual samples in the feature space of the minority class samples to supplement the sample count. For majority class samples (>30%), proportional undersampling is performed, and some redundant samples are randomly removed to ensure that the sample ratio of each class is controlled within a reasonable range of 20%-30%. This sample balancing process avoids model adaptation accuracy deviations due to sample bias, generating an optimized material-demand mapping dataset that provides a balanced and high-quality data foundation for model training.

[0061] A multi-level loss function is constructed, designed hierarchically according to the adaptation evaluation dimensions, to achieve synergistic optimization of adaptation accuracy across multiple dimensions. The first-level functional adaptation loss uses the mean squared error loss function, which quantifies performance adaptation deviation by calculating the mean squared difference between material performance parameters and requirement parameters, and penalizes substandard samples (such as materials whose tensile strength does not reach the threshold). The second-level scene fit loss uses the cross-entropy loss function, which assesses the degree of scene constraint satisfaction by measuring the difference in probability distribution between the material scene fit label and the actual scene constraints, assigning high loss values ​​to scene conflict samples (such as indoor materials used in outdoor scenes). The third-level cost controllability loss uses the Huber loss function, which manifests as mean squared error when the error is small and as absolute error when the error is large, effectively reducing the interference of extreme cost samples (such as special materials whose cost far exceeds the threshold) on training. The three-level loss functions are merged into a total loss function through weight coefficients (functional adaptation loss weight 0.5, scene fit loss weight 0.3, cost controllability loss weight 0.2), enabling model training to simultaneously consider the three core dimensions of performance, scene, and cost.

[0062] The optimized material-demand mapping dataset was imported into the improved XGBoost evaluation network. This network is an architectural optimization based on the native XGBoost, with the addition of a feature adaptation layer and a regularization layer. The feature adaptation layer dynamically adjusts feature weights through an attention mechanism, assigning higher weights to core features (such as functional parameters) and lower weights to secondary features, thereby improving the model's ability to capture key information. The regularization layer uses L2 regularization, which suppresses model overfitting (avoiding the model from excessively fitting the training data and thus reducing its generalization ability) by adding a parameter squared term to the loss function. Considering the upper limit of computing power, the iteration number and learning rate adjustment rhythm were determined through a training efficiency simulation algorithm (simulating training time and accuracy under different iteration numbers and learning rates): for complex feature scenarios, the iteration number was set to 500-800 times, with an initial learning rate of 0.01, decaying by 10% every 100 rounds to ensure sufficient training for complex features; for simple feature scenarios, the iteration number was set to 200-300 times, with an initial learning rate of 0.05, decaying by 15% every 50 rounds to improve training efficiency. Finally, a fully configured model training scheme is generated, which clarifies the core elements such as network architecture, training parameters, and iteration strategies.

[0063] Iterative training is performed according to the model training scheme. The batch gradient descent algorithm is used to update network parameters during training. The dataset is divided into training and validation sets in a 7:3 ratio. In each training round, the parameters are updated using the training set, while the loss value is calculated on the validation set. To avoid overfitting, an early stopping mechanism is implemented. Training automatically stops when the loss value on the validation set does not decrease for 10 consecutive rounds, retaining the current optimal model parameters. After training, the model outputs the feasibility probability values ​​of the material in three dimensions: functional suitability, scene fit, and cost controllability. The values ​​range from [0,1]. The closer the value is to 1, the better the suitability in that dimension. For example, a material with a functional suitability probability of 0.92, a scene fit probability of 0.85, and a cost controllability probability of 0.78 indicates high performance compliance, good scene suitability, and an acceptable cost.

[0064] A weighted fusion algorithm is used to calculate the overall score, which is calculated by weighting the scores according to the formula: "functional compatibility probability × 0.5 + scenario suitability probability × 0.3 + cost controllability probability × 0.2". The overall score ranges from [0,1]. All materials are sorted from highest to lowest based on their overall scores, generating a ranking of material overall compatibility. Simultaneously, multi-dimensional feasibility probability values ​​for each material are output, clearly presenting the compatibility advantages and disadvantages of each material in different dimensions. This result provides accurate data support for subsequent fusion strategies and allows designers to intuitively view material compatibility, facilitating further decision-making.

[0065] S104 generates the final material matching recommendation result through a combination of early functional weight allocation, mid-term scenario characteristic weighted fusion, or late-term dynamic weight adjustment voting / gradient enhancement fusion strategy based on material performance.

[0066] In one implementation, a comprehensive verification of the initial functional weight allocation strategy is conducted, focusing on three key elements: functional indicator priority, weight quantification standards, and adaptation association rules. A requirement matching degree verification algorithm is used to analyze the correlation between functional indicators and core product requirements, determining the priority sequence by calculating the contribution of indicators to functional implementation. A weight rationality evaluation algorithm (analytic hierarchy process) is employed, combined with industrial design industry selection standards, to calibrate the initial weights, ensuring that the weight allocation conforms to the actual application logic.

[0067] Through verification and evaluation, three types of core outputs are generated: functional weight allocation coefficients clarify the weight ratio of each functional indicator (such as tensile strength 30%, weather resistance 20%, processing feasibility 15%), core indicator adaptation thresholds define the minimum compliance standards for each functional parameter (such as tensile strength ≥300MPa, weather resistance ≥3), and weight adjustment constraints define the boundary range of weight modification (such as the adjustment range of core indicator weights not exceeding ±5%). These three together constitute the core mechanism information for the initial integration, providing a functional dimension basis for subsequent integration strategies.

[0068] For the mid-term scenario-based weighted fusion strategy, scenario adaptability verification and fusion efficiency calculation were carried out. Scenario dimensions were decomposed into a three-layer structure of usage environment, operating conditions, and constraint type, breaking down complex scenarios into quantifiable subdivisions. The feature weight allocation adopted the entropy weight method, calculating the objective weight of each feature based on the dispersion of scenario data. The adaptability of the fusion logic was verified through a scenario-feature matching matrix.

[0069] Scene adaptability verification uses real-world scenario data for testing to evaluate the adaptability of the fusion strategy to different scenarios. Fusion efficiency is calculated by statistically analyzing the time loss and resource consumption during the fusion process to optimize the fusion logic. Finally, scene characteristic adaptation parameters are generated (e.g., 25% weighting for UV resistance in outdoor scenarios and 30% weighting for heat resistance in high-temperature scenarios), dynamic weighting adjustment coefficients (the coefficient range of 0.8-1.2 is dynamically adjusted based on scene complexity), and scene-material matching association schemes (clearly defining the preferred material characteristic types for matching in different scenarios). This forms the foundational information for mid-term fusion, strengthening the scene adaptability of the fusion strategy.

[0070] The system incorporates a dynamic weighting adjustment strategy, focusing on three core aspects: performance weight calibration, result coordination rules, and probability integration process. Performance weight calibration employs a feedback calibration algorithm to dynamically adjust the weighting of each function and scenario characteristic based on the initial material adaptation performance. Result coordination rules clarify the conflict resolution logic for multiple strategy outputs (e.g., when function adaptation conflicts with scenario adaptation, the core requirement takes precedence). The probability integration process uses a weighted probability fusion algorithm to integrate the adaptation probabilities output by different strategies.

[0071] Based on the above process, strategic synergy optimization is performed on the core mechanism information of the early stage fusion and the basic information of the mid-stage fusion. Redundant information removal algorithms are used to delete duplicate and conflicting parameters and rules, while retaining key and effective information. Multi-strategy linkage execution parameters (such as the start conditions and execution order of each fusion stage) and result calibration correction coefficients (used to correct the deviation of single strategy output, with a coefficient range of 0.9-1.1) are generated to form fusion optimization information, realizing the organic synergy of the three fusion strategies.

[0072] By integrating information from the core mechanisms of early-stage integration, basic information from mid-stage integration, and information from integration optimization, a comprehensive integration evaluation system is constructed. With functional adaptability as the core, scenario suitability as a constraint, and performance stability as a guarantee, parameters, thresholds, rules, and coefficients from these three types of information are systematically integrated to form a unified integration decision-making model.

[0073] The model first calculates the material functionality fit score based on the initial functional weight allocation coefficients. Then, it adjusts the score by incorporating mid-term scenario characteristic fit parameters. Finally, it performs precise calibration by integrating the results from the optimization information to calibrate and correct the coefficients. Based on the calibrated comprehensive score and the material's multi-dimensional feasibility probability values, the model selects the materials with the best overall performance, generating the final material fit recommendation results. These recommendations ensure that the core functional requirements of the product are met, that the model aligns with scenario usage constraints, and that it exhibits good performance stability, providing a direct decision-making basis for material selection in industrial design.

[0074] S105 processes adaptation indicators, performance parameters, and recommendation results based on a client-side visual interactive platform, supports designers in interactively modifying parameters and scene constraints, and generates exportable material selection reports and parameter configuration files.

[0075] In one implementation, the client-side visual interaction platform constructs a multi-dimensional data display module, presenting data categorized into three main types: compatibility indicators, performance parameters, and recommendation results. The compatibility indicator module intuitively displays quantitative scores and weightings for functional matching, scenario suitability, and cost controllability, presenting core data in the form of progress bars and numerical labels. The performance parameter module lists comparisons between key material parameters and demand thresholds according to dimensions such as mechanical properties, weather resistance, and processing feasibility, clearly highlighting compatibility strengths and differences. The recommendation results module displays candidate materials sorted by overall compatibility, simultaneously presenting multi-dimensional feasibility probability values.

[0076] The platform features an interactive user interface with three main functional areas: parameter modification, scene constraint adjustment, and filter condition configuration. The parameter modification area allows designers to adjust core parameters such as functional indicator weights, cost threshold ranges, and performance parameter precision. The scene constraint adjustment area provides functions for switching scene types and adding / deleting constraints. The filter condition configuration area supports setting filter rules based on dimensions such as compatibility rating, performance level, and cost range. The interface's operation logic aligns with the industrial design workflow, ensuring convenient interaction.

[0077] When designers initiate parameter modification operations through the interactive interface, the platform employs a real-time response recalculation algorithm. Based on pre-compiled calculation logic and a caching mechanism, it quickly recalculates the suitability score and candidate material list. For example, when modifying the weight of functional indicators, the system automatically updates the weight coefficients of the multi-dimensional weighted algorithm and recalculates the comprehensive score; when adjusting the cost threshold range, it triggers the range comparison method to re-verify the compliance of material costs and simultaneously updates the priority ranking.

[0078] The scene constraint modification feature supports adding, deleting, and modifying constraints. When adding a constraint, the system automatically associates it with the scene-material matching scheme and supplements the corresponding material property verification. When deleting or modifying a constraint, the scene constraint matching algorithm is re-executed, and the material-scene fit score is adjusted. The results of all modification operations are fed back to the interface in real time, allowing designers to intuitively view the impact of parameter changes on the recommendation results, achieving a closed-loop operation of "modification-preview-confirmation".

[0079] The platform automatically generates material selection reports using standardized templates. The reports include five core modules: basic project information (product name, design scenario, requirement date), detailed requirement parameters (quantitative standards for functional indicators, scenario constraints, and cost thresholds), material compatibility analysis (individual scores, overall scores, compatibility highlights, and potential risks for each candidate material), recommendation results (a list of prioritized materials and selection criteria), and attachments (performance parameter comparison table and scenario compatibility verification records). The reports use a combination of text and graphics, with key data presented visually in charts to enhance readability and decision-making value.

[0080] The parameter configuration file is generated according to industry-standard formats and includes three core components: requirement parameter configuration, adaptation rule configuration, and model parameter configuration. The requirement parameter configuration records the final confirmed quantization standards and accuracy ranges; the adaptation rule configuration specifies core rules such as weight allocation coefficients, similarity thresholds, and validation standards; and the model parameter configuration records model-related parameters such as training strategies, iteration counts, and loss function weights. The configuration file supports exporting to common formats such as XML and JSON, and can be directly imported into subsequent production management systems or model training platforms, enabling data reuse.

[0081] The platform offers diverse file export functions, allowing designers to choose report formats (PDF, Word, Excel) and configuration file formats (XML, JSON, TXT) as needed. It supports single-file export or batch export. Data encryption algorithms ensure file security during the export process, while also retaining export logs that record crucial information such as export time, file type, and operator.

[0082] To ensure data traceability, the platform has established a complete data operation log system, recording parameter modification history, scene constraint adjustment records, and recommendation result change trajectories. Each operation is associated with a timestamp and operator information. Designers can use the logs to trace the configuration status and recommendation results at any point in time, facilitating solution iteration and optimization, problem troubleshooting, and providing data support for the compliance of the industrial design process.

Claims

1. An integrated application method for an intelligent material adaptation and selection system for industrial design and a multi-scenario structural device, characterized in that, include: The industrial design requirement parameters are preprocessed, uniformly converted into a standardized data format, and key indicators are quantified. The parameter matching and verification are completed with product functional requirements as the core reference and scenario usage conditions as constraints. The parameter consistency is evaluated through the deviation coefficient, and a standardized requirement parameter dataset is generated. Based on a deep learning matching model, the standardized requirement parameter dataset is processed to support fully automated intelligent matching or matching assisted by designers with a small number of parameter adjustments. A candidate material list and adaptability score are generated, and then the core performance indicators and key points of scenario adaptation are extracted and synchronously associated with material characteristic parameters to generate a material-requirement corresponding dataset. The improved XGBoost evaluation network based on the TensorFlow framework processes the material-demand correspondence dataset, and improves the fitting accuracy by using feature engineering optimization, sample balancing and multi-level loss function to generate material comprehensive fitting ranking results and multi-dimensional feasibility probability values. The final material matching recommendation result is generated by using a combination of early functional weight allocation, mid-term scenario characteristic weighted fusion, or late-term dynamic weight adjustment voting / gradient enhancement fusion strategy based on material performance. Based on a client-side visual interactive platform, the system processes adaptation indicators, performance parameters, and recommendation results, supports designers in interactively modifying parameters and scene constraints, and generates exportable material selection reports and parameter configuration files.

2. The method as described in claim 1, characterized in that, The industrial design requirement parameters are preprocessed, uniformly converted into a standardized data format, and key indicators are quantified. Parameter matching and verification are completed using product functional requirements as the core reference and scenario usage conditions as constraints. Parameter consistency is evaluated through deviation coefficients, generating a standardized requirement parameter dataset, including: To address the multi-dimensional and heterogeneous nature of industrial design requirements parameters, which include functional indicators, scenario constraints, and cost thresholds, we first standardize the data format through format conversion, and then quantify the mechanical properties, weather resistance, processing feasibility, cost range, and environmental protection level to clarify the quantification standards and accuracy range of each parameter. Using the core functional requirements of the product as the core reference and the usage conditions of the scenario as the rigid constraint, we carried out layered verification of basic parameter screening and core indicator adaptation, and checked the compliance of the parameters with functional requirements and scenario constraints one by one. A deviation coefficient evaluation mechanism is introduced to verify parameter consistency, and a parameter integrity threshold is set for double verification. A graded handling mechanism is established for automatic correction of minor abnormalities and manual intervention for serious abnormalities. Through a complete process of format unification, indicator quantification, hierarchical verification, consistency assessment, and anomaly correction, a standardized requirement parameter dataset is generated that combines data standardization, parameter completeness, and material selection adaptability.

3. The method as described in claim 2, characterized in that, Based on a deep learning matching model, a standardized requirement parameter dataset is processed, supporting fully automated intelligent matching or designer-assisted matching with minor parameter adjustments. This generates a candidate material list and suitability score, then extracts core performance indicators and key points for scenario adaptation, and synchronously associates them with material characteristic parameters to generate a material-requirement mapping dataset, including: Based on the indicator dimension features of the standardized demand parameter dataset and the characteristics of the material library, the intelligent matching strategy is determined by the deep learning matching model and the adaptation rule base through negotiation. The matching retrieval range and similarity threshold are determined by demand parameter priority parsing and material characteristic correlation statistics. A parameter classification retrieval mechanism is used to split the standardized demand parameter dataset, extract core parameters according to functional indicators, scenario constraints, and cost threshold types, associate them with corresponding material property matching tags, and generate a preliminary candidate material set. The initial candidate material set is tested for compatibility. Supplementary search service is initiated for materials whose performance parameters do not meet the standards. Rule adjustment mechanism is triggered for materials with conflicting scenario compatibility. Priority reordering process is executed for samples whose cost exceeds the threshold. An optimized candidate material list containing material screening strategy and compatibility correction scheme is generated. The optimized candidate material list is evaluated for its suitability. A multi-dimensional weighted algorithm is used to quantify the degree of functional matching, scenario fit, and cost controllability. A material suitability score report and characteristic association details are generated simultaneously. It supports fully automated intelligent matching or matching assisted by designers adjusting a few parameters, extracts core performance indicators and key points for scene adaptation, and synchronously associates material characteristic parameters to form a final material-demand mapping dataset containing material information, adaptability scores, and characteristic correlation data.

4. The method as described in claim 1, characterized in that, An improved XGBoost evaluation network based on the TensorFlow framework is used to process the material-demand mapping dataset. Feature engineering optimization, sample balancing, and multi-level loss functions are employed to improve the fitting accuracy, generating a comprehensive material fit ranking result and multi-dimensional feasibility probability values, including: The adaptation feature complexity of the material-demand correspondence dataset is processed, and the core evaluation indicators, including the functional matching accuracy threshold and the scenario constraint fit conditions, are identified and classified. At the same time, the computing power resources of the TensorFlow framework are quantitatively evaluated, and feature complexity classification results and computing power resource evaluation data are generated. Based on the feature complexity classification results and computing resource assessment data, a refined or efficient training strategy is determined through a supply and demand balancing algorithm. The range of model input data is defined with the principle of full coverage of core features, and the model training strategy and input data range definition results are generated. Based on the results of defining the range of input data, feature engineering optimization and sample balancing are performed on the dataset to generate an optimized material-demand correspondence dataset. Construct a multi-level loss function, import the optimized material-demand correspondence dataset into the improved XGBoost evaluation network, determine the number of iterations and the learning rate adjustment rhythm based on the upper limit of computing power, and generate a fully configured model training scheme. Iterative training is performed according to the model training scheme, and the multi-dimensional adaptation probability of materials is output. The comprehensive score is calculated by weighted fusion, and the comprehensive adaptation ranking result and multi-dimensional feasibility probability value of materials are generated.

5. The method as described in claim 4, characterized in that, The final material matching recommendation results are generated through a combination of early-stage functional weight allocation, mid-stage scenario-based weighted fusion, and late-stage dynamic weight adjustment voting / gradient enhancement fusion strategies based on material performance. These include: The functional indicator priority, weight quantification standard, and adaptation association rules in the early functional weight allocation strategy are verified for demand matching degree and weight rationality assessment. Functional weight allocation coefficients, core indicator adaptation thresholds, and weight adjustment constraints are generated to form the early integration core mechanism information. The scene dimension decomposition, feature weight allocation, and fusion logic adaptation in the mid-term scene feature weighted fusion are verified for scene adaptability and fusion efficiency are calculated. Scene feature adaptation parameters, dynamic weighted adjustment coefficients, and scene-material matching association schemes are generated to form the basic information for mid-term fusion. By combining the performance weight calibration, result coordination rules, and probability integration process in the later dynamic weight adjustment voting / gradient boosting fusion, the core mechanism information of the early fusion and the basic information of the mid-term fusion are optimized by strategy coordination and redundant information is removed to generate multi-strategy linkage execution parameters and result calibration correction coefficients, thus forming fusion optimization information. By integrating the core mechanism information from the early stage of fusion, the basic information from the mid-stage of fusion, and the optimization information from the fusion, a final material matching recommendation result is generated that takes into account functional adaptability, scenario fit, and performance stability.

6. An electronic device, characterized in that, include: First processor; and memory for storing executable instructions of the first processor; The first processor is configured to execute the integrated application method of any one of claims 1 to 5 in the intelligent adaptation and selection system for industrial design materials and the multi-scenario structural device by executing the executable instructions.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the second processor, it implements the integrated application method of any one of claims 1 to 5 in the intelligent adaptation and selection system for industrial design materials and multi-scenario structural devices.

Images