Multi-material connection process intelligent selection system based on rule engine and gradient screening

Abstract

The present application relates to a multi-material connection process intelligent selection system based on a rule engine and gradient screening, comprising a process database module, a requirement input module, a screening module and a selection output module, the screening module is built-in with a rule engine gradient screening mechanism and a multi-objective optimization model; the rule engine gradient screening mechanism is used for gradually excluding unsuitable processes; the multi-objective optimization model quantifies performance, cost and manufacturability into a weighted scoring system based on actual engineering constraints and multi-objective optimization strategies, and performs three-dimensional weighted scoring on the screened processes; the paradigm transition from traditional experience-driven to data-driven is realized; a process gradient scheme can be automatically generated, and in the case of new energy vehicle body connection, the scheme generation efficiency is improved by 5.8 times, the development cycle is shortened by 60%, the lightweight rate is broken through by 18% while reducing the whole life cycle cost by 22%-40%, the key performance reaches 98%, and the first set of theory-practice closed-loop solution for multi-material hybrid connection is provided.

Description

Technical Field

[0001] This invention relates to the field of material joining technology, and more specifically to an intelligent selection system for multi-material joining processes based on rule engines and gradient filtering. Background Technology

[0002] The selection of lightweight automotive materials and their joining technologies currently faces challenges in three areas: technical difficulty, evaluation difficulty, and application difficulty. Technical difficulty: The gap in international-level lightweight automotive materials, joining processes, and structural design technologies is difficult to bridge. Evaluation difficulty: The construction of data, standards, and evaluation systems for lightweight automotive materials and joining processes urgently needs improvement; the collection, organization, and application of relevant material performance data are inadequate. Application difficulty: New materials and processes are difficult to apply across the automotive industry chain. Currently, joining process selection mainly relies on individual engineer experience or process guidance documents (such as welding parameter manuals and riveting technical specifications), and a systematic mapping logic between material properties, process parameter schemes, and product structural design requirements has not yet been established. In industrial design practice, current methods for determining process parameters still rely primarily on traditional iterative correction models and experience-driven design paths. These methods require multiple design cycle adjustments in the early stages of selection and heavily depend on large-scale physical testing verification, significantly increasing time and economic costs, and making it difficult to quickly lock in the globally optimal solution in engineering practice. Therefore, a multi-material joining process intelligent selection system based on rule engines and gradient filtering is urgently needed to solve these problems. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent selection system for multi-material connection processes based on rule engines and gradient screening. Through the rule engine gradient screening mechanism and multi-objective optimization model, multi-objective screening is performed to upgrade from providing feasible solutions through discretization to dynamically outputting the globally optimal solution within the design domain. This shortens the development cycle of the vehicle body connection system, reduces the total life cycle cost, and improves the key performance compliance rate.

[0004] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: a multi-material bonding process intelligent selection system based on rule engine and gradient screening, comprising:

[0005] The process database module stores various basic connection processes and their parameters.

[0006] The requirement input module receives user requirement input, structures the requirement input, and decomposes it into basic content and data content.

[0007] The filtering module includes a built-in rule engine gradient filtering mechanism and a multi-objective optimization model;

[0008] The rule engine gradient filtering mechanism is based on matrix mutual exclusion rules, as well as basic content and data content, to gradually eliminate inapplicable processes;

[0009] The multi-objective optimization model, based on practical engineering constraints and multi-objective optimization strategies, quantifies performance, cost, and manufacturability into a weighted scoring system, and applies a three-dimensional weighted scoring to the process after gradient screening by the rule engine:

[0010] ;

[0011] in, The overall score for process scheme i is... For performance scores, Score for cost. For manufacturing scores, α is the performance score weight, β is the cost score weight, and γ is the manufacturability score weight.

[0012] The selection output module outputs the process with the best score and provides a complete process selection support report.

[0013] Preferably, the basic connection process includes fusion welding, solid-state welding, mechanical connection, structural adhesive bonding, and composite connection; the matrix mutual exclusion rules include:

[0014] When the material meets the following requirements: carbon equivalent or Electrochemical potential difference Melting point difference Zn / Mg content >2% and the difference in thermal expansion coefficients between materials When any one or more of the following conditions are met, the fusion welding process is disabled;

[0015] When the material wall thickness Solid-state welding processes are prohibited during this period.

[0016] When the material meets the following conditions: the wall thickness of the connector is less than 0.8 times the nominal diameter of the thread, and the elongation is... Rockwell hardness Base material thickness And the thickness ratio of the two plates When any one or more of the following conditions are met, mechanical connection processes are disabled;

[0017] Structural adhesive bonding processes are prohibited when the service temperature T is greater than the glass transition temperature of the adhesive +30℃.

[0018] Composite bonding processes are prohibited when the electrochemical potential difference ΔE between dissimilar materials is greater than 0.5V.

[0019] Preferably, the rule engine gradient filtering mechanism introduces a dynamic interruptor, which is triggered based on the basic content and data content. The dynamic interruptor includes:

[0020] Level 1 interruption: The number of repeated disassembly and assembly operations exceeds the corresponding threshold;

[0021] Level 2 interruption: Surface flatness tolerance is less than the corresponding threshold;

[0022] Level 3 interrupt: Connection point strength margin is greater than the corresponding threshold;

[0023] A progressive priority association is established between interrupt levels: Level 1 interrupts forcibly terminate the current process evaluation and activate the alternative process library; Level 2 interrupts trigger the tolerance compensation algorithm or roll back to a historically feasible solution when correction fails; Level 3 interrupts initiate load path reallocation or redundant design; and

[0024] After an interrupt, the system outputs structured response instructions: Level 1 interrupt returns an irreversible process exclusion report; Level 2 interrupt outputs a surface morphology correction scheme; and Level 3 interrupt generates intensity enhancement simulation data.

[0025] Preferably, after filtering by the rule engine gradient filtering mechanism, the remaining processes are further filtered based on their connection performance according to the failure risk level, including:

[0026] shear strength and peel strength These are Level I critical indicators; if any one of them is not met, the device will be immediately discontinued. The maximum permissible shear strength of the material / joint corresponding to the process. Tensile load applied externally;

[0027] Fatigue strength and durability These are Level II and Level III indicators. Failure to meet either one triggers a warning process. The weights are dynamically calculated using the entropy weight method, and the process closeness is generated based on the ideal solution approximation method. ,reserve The process.

[0028] Preferably, the weights generated based on the entropy weight method are used for weighted scoring of the process. When process data is added or removed, the ideal solution and weights are recalculated, and a new ranking result is output. The entropy weight method calculates the information entropy of each indicator by acquiring the current indicator dataset of the process to be evaluated in real time; including:

[0029] Input the current set of process schemes to be evaluated and their index data; construct a dynamic evaluation matrix:

[0030] ;

[0031] in, Let be the value of process i on index j. The matrix dimension changes dynamically with the number of processes m, and n represents the total number of evaluation indicators.

[0032] Eliminate the dimensions of the dynamic evaluation matrix and calculate the information entropy:

[0033] ;

[0034] Information entropy; The weighting of indicator characteristics reflects the proportion of data for a specific process scheme i on indicator j to the total amount of that indicator; among which... The information entropy value is recalculated in real time as the number of processes (m) changes. These are the normalized values ​​of the process parameters after standardization.

[0035] Generate entropy weights:

[0036] ;

[0037] The entropy weight of the j-th indicator; specifically, in the system, it refers to the number of performance dimensions that affect process selection.

[0038] Calculation of process approximation degree based on ideal solution approximation method ,include:

[0039] The weights W output by the entropy weight method are used to construct the weighting matrix. ;

[0040] Calculate the ideal solution With negative ideal solution :

[0041] ;

[0042] ;

[0043] ;

[0044] in, This represents the distance between the current technology and the ideal solution; This represents the distance between the current process and the negative ideal solution.

[0045] Preferably, the performance score :

[0046] ;

[0047] For actual measurement of shear strength during the process; Minimum allowable peel strength; This is the importance coefficient for shear strength; This is the importance coefficient for peel strength;

[0048] The cost score :

[0049] ;

[0050] Let i be the total unit cost of process i; is the industry benchmark cost, ℎ is the process loss rate, and ℎ∈[0,1];

[0051] The manufacturability score :

[0052] ;

[0053] For spatial adaptability coefficient, ; To achieve the production cycle time target, ; Spatial weights; Efficiency weight.

[0054] Preferably, a dynamic adjustment mechanism is constructed to dynamically adjust α, β, and γ, including:

[0055] Prioritize and generate requirement tags based on basic and data content;

[0056] Based on the demand tags, initial weights are assigned using a preset matrix from the rule base; the preset matrix includes:

[0057] Cost-sensitive matrix: ;

[0058] High-performance guided matrix: ;

[0059] Mass production oriented matrix: .

[0060] Preferably, the process selection support report includes recommendation results, recommendation basis, performance prediction, economic analysis and potential risk warning, and the potential risk warning includes marking the potential risks in the recommended process.

[0061] Preferably, the selection system further includes a feedback module, which performs the following operations:

[0062] Collect data on process intensity, cost, and yield in actual production;

[0063] Compare the predicted and actual values. If the performance deviation exceeds 20% or the cost deviation exceeds 15%, it is marked as a failure case.

[0064] Automatically reduce the corresponding weight based on the type of failure case;

[0065] After a preset number of failure cases have been accumulated, the preset matrix of the rule base is updated.

[0066] Beneficial effects: This invention utilizes a rule-based gradient filtering mechanism to perform gradient filtering operations, combined with a multi-objective optimization model, achieving a paradigm shift from traditional experience-driven (selection deviation rate >30%) to data-driven (matching accuracy ≥93%) approaches. The system embeds quantitative indicators such as dynamic strength margin and carbon footprint coefficient, and can automatically and quickly generate process gradient solutions based on user input (such as prioritizing advanced technologies like auxiliary bonding). In new energy vehicle body connection cases, it has been verified that the solution generation efficiency is improved by 5.8 times, the development cycle is shortened by 60%, the lightweighting rate exceeds 18% while reducing the total life cycle cost by 22%-40%, and the key performance compliance rate is 98%, providing the first theoretical-practice closed-loop solution for multi-material hybrid connection. Attached Figure Description

[0067] Figure 1 This is a flowchart of the intelligent selection system of the present invention;

[0068] Figure 2 This is a diagram of the process database module of the present invention;

[0069] Figure 3 Interface 1 for the input module of the present invention;

[0070] Figure 4 Interface 2 for the input module of the present invention;

[0071] Figure 5 This is a flowchart of the weighted scoring process for a specific process screening case of the present invention. Detailed Implementation

[0072] To make the objectives and advantages of this invention clearer, the invention will be specifically described below with reference to embodiments. It should be understood that the following text is merely used to describe one or more specific embodiments of the invention and does not strictly limit the scope of protection specifically claimed by the invention.

[0073] Example: Reference Figure 1 As shown, the intelligent selection system for multi-material bonding processes based on rule engines and gradient filtering includes:

[0074] Process database module, see reference Figure 2 As shown, it stores various basic connection processes and their parameters, including fusion welding processes (resistance spot welding, laser welding, gas shielded welding), solid-state welding processes (friction welding, ultrasonic welding), mechanical connection processes (self-piercing riveting, rivetless connection, bolted connection), structural adhesive bonding processes, and other composite connection processes (adhesive riveting composite, adhesive welding composite). These connection processes will serve as a basic process type library, which can be filtered according to user needs and finally generate selection schemes.

[0075] Requirement input module, see reference Figures 3-4 As shown, the input connection requirements are structured and coded, decomposing user requirements into basic content (BUR00X series coding) and data content (DUR00X series coding). The basic content includes material properties and surface treatment methods, while the data content includes thickness combinations, strength requirements, surface flatness thresholds, etc. Taking a user as an example, the organized requirement data is shown in Table 1:

[0076] Table 1 Demand Data:

[0077]

[0078] The filtering module includes a built-in rule engine gradient filtering mechanism and a multi-objective optimization model;

[0079] The rule engine's gradient filtering mechanism mainly involves: for filtering: sheet material combination, overlap form, process constraints, and connection performance; for scoring: cost constraints and environmental constraints. This mechanism operates based on matrix mutual exclusion rules, progressively eliminating inapplicable processes.

[0080] Sheet metal combination: Primarily used for initial comparison of the user's BUR requirements, including material grade, thickness, coating, and surface treatment (e.g., quenching, tempering, normalizing, induction hardening, isothermal hardening, carburizing, etc.). "Grade" is used to identify the characteristics of the joining materials and whether they are the same metal. For example (case study): if the user's required material is low-carbon alloy steel with a carbon content below 0.2%, or cast iron, its weldability is poor, and the welding process will be excluded. If the material is steel-aluminum, resistance spot welding will also be excluded due to the difficulty in achieving such dissimilar metal connections. "Thickness" refers to the sheet thickness of the required material to help screen joining processes. For example: if the user's sheet thickness is relatively thick, then processes like press riveting, which have high requirements for riveting processes and limitations on sheet thickness combinations, will be excluded.

[0081] Overlap type: To determine the matching degree between the connection method in the process type library and the user's requirements, it mainly includes planar butt joint, right angle butt joint, single overlap joint, double overlap joint, T-joint, etc. For example, it receives the overlap type code input by the user (T01 planar butt joint / T02 single overlap joint, etc.) and deletes the process in the process library that cannot implement this type.

[0082] Process constraints: These mainly include "puncture permission" and "single / double-sided connection," which correspond to the user's contact requirements for the connecting plates. For example, in the connection process type library, self-riveting and press-fit riveting technologies both require that both ends of the rivet gun can contact both sides of the plates to be connected. If the user's parts to be connected can only be connected from one side, the above two processes will be excluded. "Removable requirement" refers to further screening the connection process based on whether the user's parts need to be disassembled. For example (case study): the connection process library includes threaded connections as removable connections, while riveting and welding are not removable. "Surface flatness" refers to the user's requirements for the surface of the material after connection. For example, structural adhesive bonding technology does not damage the base layer / coating and has high structural surface quality. If the user has high requirements for surface flatness, this technology will be retained and enter the next round of screening. "Out-of-tolerance requirement" means that the external dimensions of the parts to be connected exceed the tolerance range specified in the product standard.

[0083] Matrix mutual exclusion rules include:

[0084] When the carbon equivalent of the material or When this is the time, fusion welding is prohibited;

[0085] When the material required by the user is one with poor weldability, such as cast iron, copper, or aluminum alloy, welding processes are prohibited.

[0086] When the electrochemical potential difference of dissimilar materials Resistance spot welding is prohibited during this period.

[0087] When the melting point of the material is different Arc welding is prohibited during this period.

[0088] Arc welding is prohibited for materials containing volatile elements, such as galvanized sheets, with a Zn / Mg content >2%.

[0089] When the difference in thermal expansion coefficients between materials Laser welding is prohibited during this period.

[0090] When the wall thickness of thin-walled parts Friction stir welding is prohibited.

[0091] When the wall thickness of the connector is less than 0.8 times the nominal diameter of the thread, such as the wall thickness of a thin-walled pipe fitting... Standard thread fastening process is prohibited;

[0092] When the material is too brittle, the elongation... For materials such as gray cast iron and hardened steel, riveting is prohibited.

[0093] When the material is an ultrahard material, Rockwell hardness For example, for cemented carbide, riveting is prohibited;

[0094] When the base material thickness is an ultra-thin sheet, the thickness Riveting is prohibited.

[0095] When the thickness ratio of the two plates is too large, the thickness ratio of the two plates is... Riveting is prohibited.

[0096] When the electrochemical potential difference of dissimilar materials In cases such as aluminum-copper or magnesium-steel combinations, conductive adhesive bonding processes are prohibited.

[0097] When the service temperature T is greater than the glass transition temperature (Tg) of the adhesive, the adhesive process is prohibited.

[0098] In the rule engine's gradient filtering mechanism, the current process evaluation is immediately terminated when the following user requirements are detected:

[0099] If "Permission to puncture" is "No", all puncture procedures will be eliminated.

[0100] When "non-destructive disassembly permission" is "yes", all thermosetting adhesive processes are eliminated;

[0101] When the "operating space" is "single-sided", the double-sided contact process must be eliminated.

[0102] In one embodiment: A dynamic interruptor is introduced into the gradient filtering mechanism of the rule engine. The dynamic interruptor is triggered based on the basic content and data content. The dynamic interruptor includes:

[0103] Level 1 interruption: The number of repeated disassembly and assembly operations exceeds the corresponding threshold;

[0104] Level 2 interruption: Surface flatness tolerance is less than the corresponding threshold;

[0105] Level 3 interrupt: Connection point strength margin is greater than the corresponding threshold;

[0106] The triggering conditions of the dynamic interruptor are determined by the multimodal input data input by the user, including: connection process type (adhesive / welding / riveting), material combination, and the constraint threshold of the preset industry rule base industry standard (ISO / ASTM), such as the disassembly requirement: repeated disassembly and assembly ≥ 5 times (level 1 interrupt triggering condition).

[0107] Surface flatness tolerance: (Level 2 interrupt triggering conditions);

[0108] Strength margin factor: Safety factor ≥ 1.8 (Level 3 interrupt trigger condition);

[0109] The three-level interrupt is a progressive collaborative triggering mechanism, establishing a progressive priority association between interrupt levels: Level 1 interrupt forcibly terminates the current process evaluation and activates the alternative process library; Level 2 interrupt triggers the tolerance compensation algorithm or rolls back to a historically feasible solution when correction fails; Level 3 interrupt initiates load path reallocation or redundant design; after the interrupt, structured response instructions are output: Level 1 interrupt returns an irreversible process exclusion report (including error code ERR-901); Level 2 interrupt outputs a surface morphology correction scheme (including warning code WARN-302); Level 3 interrupt generates intensity enhancement simulation data (including risk code RISK-503).

[0110] After filtering by the rule engine's gradient filtering mechanism, the remaining processes are further filtered based on their connection performance according to their failure risk level, including:

[0111] The connection structure is required to have shear strength under actual working conditions. and peel strength These are Level I critical indicators; if any one of them is not met, the device will be immediately discontinued. The maximum permissible shear strength of the material / joint corresponding to the process. Tensile load applied externally;

[0112] Fatigue strength and durability These are classified as Level II and Level III indicators. Failure to meet any of them triggers a warning process and introduces an intelligent evaluation operation. This involves acquiring the current indicator dataset (e.g., performance / cost / yield) of the process under evaluation in real time using the entropy weight method, calculating the information entropy of each indicator, and then weighting the processes based on the weights generated by the entropy weight method. When process data is added or removed, the ideal solution and weights are recalculated, and a new ranking result is output. Finally, the TOPSIS method is used to calculate the process closeness of each process to the ideal solution. ,satisfy The process then proceeds to the next scoring step. The system analyzes the dispersion of user demand data using the entropy weighting method, automatically increasing the weight of important demand indicators in the comprehensive evaluation. If a certain indicator (such as shear strength) shows significant differences in value across all processes (high dispersion), its entropy value is low, and its weight is high, indicating that this indicator is more important in distinguishing the quality of processes. If the indicator data converges (e.g., fatigue life is similar across all processes), the system will proceed more smoothly. If the entropy value is high, the weight is low. Dynamically allocating weights avoids human bias and improves the objectivity of the sorting.

[0113] Entropy weight-TOPSIS method calculation process:

[0114] Input the current set of process schemes to be evaluated and their index data; construct a dynamic evaluation matrix:

[0115] ;

[0116] in, Let i be the value of process i on index j, and let the matrix dimension change dynamically with the number of processes m.

[0117] Standardized matrix (dimension-free):

[0118] ;

[0119] These are the normalized values ​​of the process parameters after standardization. Performance data for all processes on the same index j

[0120] Calculate information entropy (reflecting dispersion):

[0121] ;

[0122] Information entropy; The weighting of indicator features reflects the proportion of data for a specific process scheme i on indicator j relative to the total amount of that indicator; dynamic logic: The information entropy value is recalculated in real time as the number of processes m changes.

[0123] Generate entropy weights:

[0124] ;

[0125] The entropy weight of the j-th indicator; n represents the total number of evaluation indicators, specifically the number of performance dimensions that affect process selection in the system;

[0126] Calculation of process approximation degree based on ideal solution approximation method ,include:

[0127] Constructing a matrix m is the number of processes;

[0128] Calculate the ideal solution With negative ideal solution :

[0129] ;

[0130] ;

[0131] ;

[0132] in, These represent the maximum values ​​of shear strength, peel strength, fatigue life, and durability in the process, respectively. This represents the distance between the current process and the ideal solution (the smaller the value, the closer it is to the optimal solution). This represents the distance between the current process and the negative ideal solution (the larger the value, the further away from the worst-case scenario).

[0133] The selection system, through an intelligent scoring model based on actual engineering constraints and multi-objective optimization strategies, quantifies performance, cost, and manufacturability into a weighted scoring system, and performs a three-dimensional weighted scoring on the processes after gradient screening by the rule engine:

[0134] ;

[0135] in, The overall score for process scheme i is... For performance scores, Score for cost. For manufacturing scores, α is the performance score weight, β is the cost score weight, and γ is the manufacturability score weight.

[0136] Performance Score :

[0137] ;

[0138] For actual measurement of shear strength during the process; Minimum allowable peel strength; This is the importance coefficient for shear strength; This is the importance coefficient for peel strength;

[0139] Cost Score :

[0140] ;

[0141] The total cost per unit of process i (including materials / processing / energy consumption); The industry benchmark cost is represented by ℎ, which represents the process loss rate. ℎ ∈ [0,1], such as the machining waste rate.

[0142] Manufacturability score :

[0143] ;

[0144] For spatial adaptability coefficient, ; To achieve the production cycle time target, ; Spatial weight (default 0.4, 0.6 for compact factory); Efficiency weight (default 0.6, a key factor in mass production).

[0145] A dynamic adjustment mechanism is established to dynamically adjust α, β, and γ, including:

[0146] Prioritize and generate requirement tags based on basic and data content;

[0147] Based on the demand tags, initial weights are assigned using a preset matrix from the rule base; the preset matrix includes:

[0148] Cost-sensitive matrix ;

[0149] High-performance guided matrix: ;

[0150] Mass production oriented matrix: ;

[0151] The selection output module, as the final selection presentation unit of the system, has the core function of presenting the process solutions after gradient screening and scoring optimization to the user in a structured, visual, and highly interpretable manner. This module not only provides recommendation results, but also simultaneously outputs the basis for recommendation, performance prediction, economic analysis, and potential risk warnings, forming a complete process selection support report.

[0152] The selected output module supports multiple output formats to meet the needs of different user groups, including:

[0153] Structured Process Selection Report: Automatically generates a PDF process selection report containing all the above content, suitable for internal enterprise review, technical exchange, and project archiving.

[0154] Process Comparison Radar Chart: Displays the comprehensive performance of each process in terms of performance, cost, and manufacturability in a multi-dimensional radar chart format, making it easy for users to quickly identify advantages and disadvantages.

[0155] Process knowledge cards: Each recommended process is accompanied by a process card, which includes information such as definition, applicable scenarios, typical applications, equipment requirements, advantages and disadvantages, and process window, thereby improving the system's knowledge dissemination capabilities.

[0156] Potential risks in the recommended process (such as material incompatibility, poor environmental adaptability, etc.) are marked, and improvement suggestions or alternative solutions are provided.

[0157] refer to Figure 5The diagram shows the scoring calculation process for a new energy vehicle battery box process selection case, which comprehensively evaluates performance, cost, and manufacturability based on three core indicators. New energy vehicle battery boxes must meet the core requirements of high shear strength, low-cost control, and production line adaptability. This case study targets three candidate processes—friction stir welding (FSW-A), laser-MIG composite welding (B), and semi-hollow self-piercing riveting (SPR)—and uses an intelligent scoring system to achieve multi-objective selection. Process A wins with the best overall performance adaptability (0.86) and cost control (0.98), meeting the requirements for high-reliability battery boxes. If the cost weight is increased to 0.5, process C (low cost + high-efficiency production) can surpass process A, demonstrating the model's flexibility. Calculations based on measured strength, equipment depreciation rate, and other parameters reduce empirical bias and adapt to the iterative needs of enterprise process parameter libraries.

[0158] The selection system also includes a feedback module, which performs the following operations:

[0159] (1) Data collection: Collect data on process intensity, cost and yield in actual production;

[0160] (2) Deviation analysis: Compare the difference between the predicted value and the actual value. If the performance deviation exceeds 20% or the cost deviation exceeds 15%, it is marked as a failure case.

[0161] (3) Weight adjustment: Automatically reduce the corresponding weight according to the type of failure case (reduce α if the performance deviation is large, and increase β if the cost deviation is large).

[0162] (4) Rule update: After accumulating 50 valid feedbacks, retrain the weight allocation model and update the rule base version.

[0163] The embodiments of the present invention have been described in detail above with reference to the examples. However, the present invention is not limited to the above embodiments. For those skilled in the art, after learning the contents described in the present invention, several equivalent changes and substitutions can be made without departing from the principle of the present invention. These equivalent changes and substitutions should also be considered to fall within the protection scope of the present invention.

Claims

1. A multi-material bonding process intelligent selection system based on rule engine and gradient filtering, characterized in that: include: The process database module stores various basic connection processes and their parameters. The requirement input module receives user requirement input, structures the requirement input, and decomposes it into basic content and data content. The filtering module includes a built-in rule engine gradient filtering mechanism and a multi-objective optimization model; The rule engine gradient filtering mechanism is based on matrix mutual exclusion rules, as well as basic content and data content, to gradually eliminate inapplicable processes; The multi-objective optimization model, based on practical engineering constraints and multi-objective optimization strategies, quantifies performance, cost, and manufacturability into a weighted scoring system, and applies a three-dimensional weighted scoring to the process after gradient screening by the rule engine: S scorei =αP i +βV i +γM i ; Among them, S scorei P is the overall score for process scheme i. i For performance scoring, V i For cost score, M i For manufacturing scores, α is the performance score weight, β is the cost score weight, and γ is the manufacturability score weight. The selection output module outputs the process with the best score and provides a complete process selection support report. The basic connection processes include fusion welding, solid-state welding, mechanical connection, structural adhesive bonding, and composite connection; the matrix mutual exclusion rules include: Materials must meet the following conditions: carbon equivalent CE ≥ 0.6% or CE ≤ 0.2%, electrochemical potential difference ΔE > 0.5V, melting point difference ΔTm > 450℃, Zn / Mg content > 2%, and difference in thermal expansion coefficient between materials Δα > 18 × 10⁻⁶. -6 When any one or more of / K are selected, the fusion welding process is disabled; Solid-state welding is prohibited when the material wall thickness T < 1.0 mm. Mechanical connection processes are prohibited when the material meets any one or more of the following conditions: connector wall thickness < 0.8 times the nominal diameter of the thread, elongation δ < 5%, Rockwell hardness HRC > 50, base material thickness t < 0.6 mm, and the thickness ratio of the two plates t1:t2 > 4:

1. Structural adhesive bonding processes are prohibited when the service temperature T is greater than the glass transition temperature of the adhesive +30℃. Composite bonding processes are prohibited when the electrochemical potential difference ΔE between dissimilar materials is greater than 0.5V. The rule engine gradient filtering mechanism introduces a dynamic interruptor, which is triggered based on the basic content and data content. The dynamic interruptor includes: Level 1 interruption: The number of repeated disassembly and assembly operations exceeds the corresponding threshold; Level 2 interruption: Surface flatness tolerance is less than the corresponding threshold; Level 3 interrupt: Connection point strength margin is greater than the corresponding threshold; A progressive priority association is established between interrupt levels: Level 1 interrupts forcibly terminate the current process evaluation and activate the alternative process library; Level 2 interrupts trigger the tolerance compensation algorithm or roll back to a historically feasible solution when correction fails; Level 3 interrupts initiate load path reallocation or redundant design; and After an interrupt, the system outputs structured response instructions: Level 1 interrupt returns an irreversible process exclusion report; Level 2 interrupt outputs a surface morphology correction scheme; and Level 3 interrupt generates intensity enhancement simulation data. After filtering by the rule engine's gradient filtering mechanism, the remaining processes are further filtered based on their connection performance according to their failure risk level, including: Shear strength τ rep ≤0.7τ max and peel strength σ peel ≥2F t / t is a Level I critical indicator; failure to meet any of these indicators will result in immediate disuse. max F represents the maximum permissible shear strength of the material / joint corresponding to the process. t Tensile load applied externally; Fatigue strength N f ≥10 6 and durability t corr ≥15 years represents Level II and Level III indicators, respectively. Failure to meet either indicator triggers a warning process. The weight is dynamically calculated using the entropy weight method, and the process closeness C is generated based on the ideal solution approximation method. i Keep C i Processes with a density greater than 0.6; The weights generated by the entropy weight method are used for weighted scoring of the process. When process data is added or removed, the ideal solution and weights are recalculated, and a new ranking result is output. The entropy weight method calculates the information entropy of each indicator by acquiring the current indicator dataset of the process to be evaluated in real time; including: Input the current set of process schemes to be evaluated and their index data; construct a dynamic evaluation matrix: Where, x ij Let be the value of process i on index j. The matrix dimension changes dynamically with the number of processes m, and n represents the total number of evaluation indicators. Eliminate the dimensions of the dynamic evaluation matrix and calculate the information entropy: e j Information entropy; p ij The weighting of indicator features reflects the proportion of data for a specific process scheme i on indicator j to the total amount of that indicator; where e j The information entropy value is recalculated in real time as the number of processes m changes, r ij These are the normalized values ​​of the process parameters after standardization. Generate entropy weights: w j The entropy weight of the j-th indicator; specifically, in the system, it refers to the number of performance dimensions that affect process selection. The process approximation degree C is calculated based on the ideal solution approximation method. i ,include: The weights W output by the entropy weight method are used to construct a weighting matrix X = [X... ij ] m×4 ; Calculate the ideal solution X + With negative ideal solution X - : X + =t max ,s max ,N fmax ,t corrmax ; X - =t min ,s min ,N fmin ,t corrmin ; in, This represents the distance between the current technology and the ideal solution; This represents the distance between the current process and the negative ideal solution.

2. The intelligent selection system for multi-material joining processes based on rule engine and gradient filtering according to claim 1, characterized in that: The performance score P i : τ actual For the actual measured shear strength during the process; σ min k1 represents the minimum allowable peel strength; k2 represents the importance coefficient of shear strength; k3 represents the importance coefficient of peel strength. The cost score V i : v i V represents the total unit cost of process i; base Let h be the industry benchmark cost, and h be the process loss rate, where h∈[0,1]. The manufacturability score M i : M i =μ1·A s +m2·h cyc ; A s A is the spatial adaptability coefficient. s ∈[0,1]; η cyc To achieve the production cycle time rate, η cyc ∈[0.5,1.5]; μ1 is the spatial weight; μ2 is the efficiency weight.

3. The intelligent selection system for multi-material joining processes based on rule engine and gradient filtering according to claim 1, characterized in that: A dynamic adjustment mechanism is constructed to dynamically adjust α, β, and γ, including: Prioritize and generate requirement tags based on basic and data content; Based on the demand tags, initial weights are assigned using a preset matrix from the rule base; the preset matrix includes: Cost-sensitive matrix: α:β:0.2:0.6:0.2; High-performance guided matrix: α:β:0.7:0.2:0.1; Mass production oriented matrix: α:β:0.3:0.3:0.

4.

4. The intelligent selection system for multi-material joining processes based on rule engine and gradient filtering according to claim 1, characterized in that: The process selection support report includes recommendation results, recommendation basis, performance prediction, economic analysis, and potential risk warnings, and the potential risk warnings include marking the potential risks in the recommended processes.

5. The intelligent selection system for multi-material joining processes based on rule engine and gradient filtering according to claim 1, characterized in that: The selection system also includes a feedback module, which performs the following operations: Collect data on process intensity, cost, and yield in actual production; Compare the predicted and actual values. If the performance deviation exceeds 20% or the cost deviation exceeds 15%, it is marked as a failure case. Automatically reduce the corresponding weight based on the type of failure case; After a preset number of failure cases have been accumulated, the preset matrix of the rule base is updated.

Images