Model-based collaborative spinning machine design assistance method and system

Abstract

The application belongs to the technical field of computer-aided design, and provides a spinning equipment design auxiliary method and system based on model cooperation. The method comprises the following steps: a cooperative design system is constructed, a large language model receives user design requirements, task decomposition is performed and a corresponding professional agent model is called to execute a design subtask, and each professional agent model feeds back simulation calculation results to a closed-loop feedback engine; the closed-loop feedback engine analyzes the feedback results of each professional agent model, and generates an attention prompt signal pointing to potential design conflicts or optimization bottlenecks based on cross-disciplinary parameter coupling relationships; the large language model performs multidisciplinary coupling analysis by using the attention prompt signal, focuses on key parameters and key disciplines pointed by the attention prompt signal, dynamically adjusts task planning and design parameters, and drives related professional agent models to perform directional iterative optimization. The application can effectively avoid blind iteration problems, and greatly improves the convergence speed and scheme quality of design.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer-aided design, in particular to a spinning equipment design assistance method and system based on model collaboration. BACKGROUND

[0002] Spinning equipment design is a typical multi-disciplinary coupled complex system problem, and its design process involves deep collaboration of multiple professional fields such as process, control, transmission, rack and modeling. With the breakthrough of artificial intelligence technology, especially the strong ability of large language models in natural language understanding and task planning, a new paradigm for design automation is provided. In the prior art, researchers have tried to introduce large models into the field of mechanical design, and have formed the following several typical schemes: the first scheme adopts a multi-agent large model to perform collaborative design through language communication, although it stimulates design diversity, but it is difficult to handle the accuracy requirements of numerical calculation and physical simulation in mechanical design.

[0003] The second scheme integrates an external function library or professional software interface for the large model to perform specific calculation or simulation tasks. However, when multiple disciplinary models (such as process, transmission, control) are executed in parallel, this method will produce a large amount of heterogeneous simulation and parameter data for complex systems such as spinning equipment. At this time, the large language model as the overall designer faces the dilemma of information overload when performing multi-disciplinary coupled analysis. It lacks an effective mechanism to quickly and accurately identify the current most critical design conflicts, the most urgent optimization bottlenecks, or the most efficient iteration paths from the complex cross-influences. This leads to an inefficient decision-making process for the large model, and it may even ignore the deep nonlinear interactions between different disciplinary parameters, thus making the optimization process of the entire collaborative system fall into blind iteration or slow convergence, and unable to truly realize efficient and accurate automated design closed loop.

[0004] Therefore, how to construct an intelligent feedback mechanism in the design framework based on the collaboration of large models and multi-disciplinary models to significantly improve the efficiency of multi-disciplinary coupled analysis and the convergence speed of design iteration is a technical problem to be solved at present. SUMMARY

[0005] To solve the above technical problems, the present application provides a spinning equipment design assistance method and system based on model collaboration.

[0006] The application provides a spinning equipment design auxiliary method based on model cooperation, comprising the following steps: constructing a cooperative design system, wherein the cooperative design system comprises: a large language model fine-tuned by domain knowledge, serving as a total planning and decision unit; a plurality of field-specific generative professional agent models, used for performing professional calculation and simulation of each discipline; and a closed-loop feedback engine with signal generation capability; the large language model receives user design requirements, performs task decomposition, and calls corresponding professional agent models to perform design subtasks, each professional agent model feeds back simulation calculation results to the closed-loop feedback engine; the closed-loop feedback engine analyzes the feedback results of each professional agent model, and generates attention prompt signals pointing to potential design conflicts or optimization bottlenecks based on cross-disciplinary parameter coupling relationships; and the large language model performs multi-disciplinary coupling analysis by using the attention prompt signals, focuses on key parameters and key disciplines pointed by the attention prompt signals, dynamically adjusts task planning and design parameters, and drives related professional agent models to perform directional iterative optimization.

[0007] The application also provides a spinning equipment design auxiliary system based on model cooperation, comprising a processing unit and a storage unit, wherein the processing unit calls and executes a pre-stored computer program in the storage unit to perform the following steps: constructing a cooperative design system, wherein the cooperative design system comprises: a large language model fine-tuned by domain knowledge, serving as a total planning and decision unit; a plurality of field-specific generative professional agent models, used for performing professional calculation and simulation of each discipline; and a closed-loop feedback engine with signal generation capability; the large language model receives user design requirements, performs task decomposition, and calls corresponding professional agent models to perform design subtasks, each professional agent model feeds back simulation calculation results to the closed-loop feedback engine; the closed-loop feedback engine analyzes the feedback results of each professional agent model, and generates attention prompt signals pointing to potential design conflicts or optimization bottlenecks based on cross-disciplinary parameter coupling relationships; and the large language model performs multi-disciplinary coupling analysis by using the attention prompt signals, focuses on key parameters and key disciplines pointed by the attention prompt signals, dynamically adjusts task planning and design parameters, and drives related professional agent models to perform directional iterative optimization.

[0008] The application has at least the following beneficial technical effects: the application introduces an attention prompt signal generation mechanism in the closed-loop feedback engine, so that the large language model can quickly identify key design conflicts from a large amount of multi-disciplinary data, the mechanism can significantly improve the efficiency of multi-disciplinary parameter coupling analysis through active diagnosis and intelligent guidance, concentrates optimization resources on the most critical design bottlenecks, effectively avoids the problem of blind iteration in traditional methods, and finally greatly improves the convergence speed and scheme quality of the spinning equipment cooperative design. BRIEF DESCRIPTION OF DRAWINGS

[0009] Figure 1 This is a flowchart of a model-based collaborative spinning equipment design assistance method disclosed in an embodiment of the present invention.

[0010] Figure 2 This is a schematic diagram of the architecture of the collaborative design system disclosed in an embodiment of the present invention.

[0011] Figure 3 This is a structural diagram of a model-based collaborative spinning equipment design auxiliary system disclosed in an embodiment of the present invention. Detailed Implementation

[0012] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0013] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of the present invention.

[0015] like Figure 1 As shown, this embodiment of the invention discloses a model-based collaborative spinning equipment design assistance method 100, which includes the following steps: Step S1: Construct a collaborative design system, which includes: a large language model fine-tuned by domain knowledge as the overall planning and decision-making unit; multiple domain-specific generative professional agent models for performing professional calculations and simulations of various disciplines; and a closed-loop feedback engine with signal generation capabilities.

[0016] In this step, the solution of the present invention operates on an auxiliary platform specifically designed for the intelligent design of spinning equipment. The core of this auxiliary platform is the construction of a collaborative design system, referring to... Figure 2As shown, the collaborative design system specifically includes the following components: a large language model fine-tuned with domain knowledge: the large language model serves as the overall planning and decision unit of the design assistance system, and its core role is similar to that of the overall designer in the traditional design process. By fine-tuning the large language model with professional knowledge in the spinning equipment field and integrating the Retrieval-Augmentation-Generation (RAG) technology, it deeply masters multi-disciplinary knowledge such as textile technology, mechanical design principles, and material mechanics, forming a professional knowledge system covering the entire design process of spinning equipment. The large language model is responsible for understanding the user's macro or specific design requirements and making global task planning and decisions accordingly.

[0017] Multiple field-specific generative professional agent models: These professional agent models serve as the professional execution unit of the design assistance system, focusing on the process, control, transmission, frame, and styling of spinning equipment. These professional agent models are built using advanced modeling methods such as physical information neural networks, Transformer-Encoder, and graph neural networks, and have high-fidelity numerical calculation and physical simulation capabilities. For example, the process small model can dynamically calculate core process parameters such as draft ratio and distribution ratio; the transmission small model can replace traditional finite element analysis based on PINN to predict the dynamic performance of the transmission system.

[0018] A closed-loop feedback engine with signal generation capability: The closed-loop feedback engine is the core hub connecting the large language model and various professional agent models, responsible for task scheduling and data transmission between them, and has advanced analysis and signal generation capabilities.

[0019] Step S2: The large language model receives user design requirements, performs task decomposition, and calls the corresponding professional agent models to execute design sub-tasks, and each professional agent model feeds back simulation calculation results to the closed-loop feedback engine.

[0020] In this step, the large language model enhanced with domain knowledge receives user input design requirements, such as designing a high-performance, low-energy spinning frame. Based on its semantic understanding and logical reasoning capabilities, the large language model analyzes and decomposes complex design requirements, generating a global design task plan.

[0021] The large language model serves as a dispatch center, calling the corresponding professional agent models to execute specific design sub-tasks according to the task plan. For example, the large language model instructs the process small model to calculate the optimal process parameter combination, while instructing the transmission small model to perform strength simulation on key transmission components. Each called professional agent model runs its high-fidelity professional calculation or physical simulation and feeds back the simulation calculation results to the closed-loop feedback engine in real time. Simulation calculation results include, for example, draft force data output by the process small model and gear contact stress data output by the transmission small model.

[0022] Step S3: The closed-loop feedback engine analyzes the feedback results of each professional agent model, and generates attention prompt signals pointing to potential design conflicts or optimization bottlenecks based on the cross-disciplinary parameter coupling relationship.

[0023] In this step, the closed-loop feedback engine receives and integrates heterogeneous simulation result data from each professional agent model. At the same time, the closed-loop feedback engine internally predefines a rule library of cross-disciplinary parameter coupling relationships based on the multi-disciplinary design knowledge graph of the spinning equipment. This rule library defines the constraints, conflicts, and causal relationships between different disciplinary parameters. For example, an increase in draft ratio in the process parameters may lead to an increase in output torque of the transmission system, which in turn may cause the gear contact stress to exceed the standard.

[0024] The closed-loop feedback engine matches and analyzes the integrated heterogeneous simulation result data with the rule library, thereby actively diagnosing potential design conflicts or optimization bottlenecks under the current design state. Based on the diagnosis results, the closed-loop feedback engine generates structured attention prompt signals pointing to these key contradictions.

[0025] It can be understood that the above attention prompt signals at least include: focus parameters (e.g., gear contact stress), associated disciplines (e.g., process and transmission), conflict levels (e.g., high risk), and suggested optimization directions (e.g., please prioritize coordination of draft ratio and gear modulus).

[0026] Step S4: The large language model uses the attention prompt signals for multi-disciplinary coupling analysis, focusing on the key parameters and key disciplines pointed to by the attention prompt signals, dynamically adjusting task planning and design parameters, and driving related professional agent models for directional iterative optimization.

[0027] In this step, the large language model receives and uses the attention prompt signals from the closed-loop feedback engine in subsequent multi-disciplinary coupling analysis. The attention prompt signals can guide the large language model to quickly focus on the key parameters and key disciplines that need to be solved most urgently from the vast amount of multi-disciplinary data.

[0028] Based on the above focus information, the large language model dynamically adjusts its task planning and design parameters, and accurately drives related professional agent models for directional iterative optimization. For example, according to the above attention signal, the large language model prioritizes adjusting the draft ratio parameter and re-calls the process and transmission model for collaborative verification, rather than blindly adjusting all parameters.

[0029] Through the intelligent closed loop of perception-diagnosis-focusing-optimization described above, the design assistance system can effectively avoid invalid trial and error on non-key parameters, significantly improve the solving efficiency and convergence speed of complex multi-disciplinary coupling problems, and finally output an optimization design scheme that meets the multi-objective constraints.

[0030] The application can quickly identify key design conflicts from massive multi-disciplinary data by introducing an attention prompt signal generation mechanism in the closed-loop feedback engine. This mechanism can significantly improve the efficiency of multi-disciplinary parameter coupling analysis, concentrate optimization resources on the most critical design bottlenecks, effectively avoid the problem of blind iteration in traditional methods, and ultimately improve the convergence speed and scheme quality of spinning equipment collaborative design.

[0031] As an example, the large language model is fine-tuned using domain knowledge, including: step S11: obtaining a professional text dataset in the field of spinning equipment design, including design manuals, academic literature, and historical design schemes; constructing training samples based on the professional text dataset, and using the instruction fine-tuning method to train the pre-trained large language model in the field.

[0032] In this step, the professional text dataset in the field of spinning equipment design is obtained from enterprise private databases and public resources. This professional text dataset mainly includes design manuals, including component standards, design specifications, tolerance matching, material selection, and other structured knowledge of spinning equipment (such as spinning machines and winding machines); academic literature, including frontier research papers and technical reports on new spinning processes, mechanism dynamics analysis, and control system optimization; historical design schemes, including successful design cases, simulation reports, and fault improvement records, etc.

[0033] Based on the above-mentioned professional text dataset, training samples are constructed. These training samples are organized in the format of instruction-input-output, for example, the instruction is "calculate the output torque under the given spindle speed and draft ratio", the input is specific parameters, and the expected output is the correct calculation steps and results. Using the instruction fine-tuning method, the pre-trained large language model is trained in the field. It can be understood that this process does not change the basic architecture of the model, but adjusts its internal parameters, so that it learns and internalizes the professional terminology, logical relationships and problem solving patterns of spinning equipment design, so as to accurately understand and respond to field-specific design instructions.

[0034] Step S12: integrate the trained large language model with the retrieval enhancement generation framework, so that it can retrieve and integrate relevant knowledge from the professional knowledge base based on the design context, to enhance the professionalism and accuracy of its design decisions; wherein the professional knowledge base is constructed by vectorizing the professional text dataset.

[0035] In this step, the large language model fine-tuned in step S11 is integrated with the retrieval enhancement generation framework. When performing a specific design task (for example, the user proposes the requirement of "optimizing the roller spacing to reduce the breakage rate"), the retrieval enhancement generation framework first automatically retrieves the most relevant knowledge fragments (such as the recommended value range in the design manual, the optimization model in the literature) related to "roller spacing", "breakage rate", and "process parameter coupling" from the constructed professional knowledge base based on the current design context.

[0036] Then, instead of generating an answer based solely on its parameters, the large language model fuses these retrieved relevant knowledge with the user's original question as a context for generating a reply. This mechanism can enhance the professionalism and accuracy of its design decisions, so that the final output of scheme suggestions, parameter calculations, or task planning is deeply rooted in verified domain knowledge rather than model speculation, thereby helping to significantly improve the reliability of the design assistance system in engineering design scenarios.

[0037] As an example, a closed-loop feedback engine with signal generation capability is constructed in the following way: Step S13: Construct a cross-disciplinary parameter coupling relationship rule base based on the spinning equipment multidisciplinary design knowledge graph, which defines the constraint relationships, conflict patterns, and causal relationships between parameters in the process, control, transmission, rack, and modeling fields.

[0038] In this step, a spinning equipment multidisciplinary design knowledge graph is constructed based on GraphRAG technology, which structurally organizes the core parameters, attributes, and their relationships in the process, control, transmission, rack, and modeling fields in the form of entity-relationship. On this basis, a cross-disciplinary parameter coupling relationship rule base is constructed.

[0039] This rule base is a logical and computable expression of complex relationships in the knowledge graph, which specifically defines the following three types of core relationships: Constraint relationship: defines the hard conditions that parameters must satisfy, for example, the torque of the transmission shaft must be less than the allowable torque of its material.

[0040] Conflict pattern: defines the inherent contradictions between different disciplinary parameter objectives, for example, increasing the spindle speed (process objective) to improve production will usually increase vibration (rack dynamics objective), constituting a pair of conflicts.

[0041] Causal relationship: defines the chain effect caused by parameter adjustment, for example, increasing the draft ratio (process parameter) will cause the draft force to increase (process result), which in turn will cause the roller transmission torque demand to rise (transmission parameter), and may cause the motor current to exceed the standard (control parameter).

[0042] Step S14: A signal generation logic module is configured to receive and integrate the simulation calculation results of each professional agent model, perform real-time matching analysis with the rule base, identify parameter deviations exceeding the preset threshold or conflict combinations violating design constraints, and generate a structured attention prompt signal including at least the focus parameter, associated disciplines, conflict level, and recommended optimization direction.

[0043] In this step, a signal generation logic module is configured with the following two core functions: real-time diagnosis: the generation logic module continuously receives and integrates simulation calculation results from each professional agent model (for example, the current draft force calculated by the process model, the gear contact stress calculated by the transmission model). Real-time matching analysis is performed on these real-time data and the rule base constructed in step S13. Through this analysis, the generation logic module can identify two types of key problems: (1) parameter deviations exceeding the preset threshold: for example, the calculated gear contact stress value exceeds the material fatigue strength safety threshold; (2) conflict combinations violating design constraints: for example, the current combination of "spindle speed" and "roller gauge" triggers the "high speed-small gauge" vibration conflict mode defined in the rule base.

[0044] Generation of guidance: based on the above identification results, the generation logic module generates a structured attention prompt signal. It should be noted that the attention prompt signal is not a simple alarm, but a data structure containing specific action guidance, at least including the focus parameter: clearly indicating the core of the problem, such as "gear contact stress"; associated disciplines: indicating the fields that need to be coordinated, such as "process" and "transmission"; conflict level: assessing the severity of the problem, such as "high risk"; recommended optimization direction: providing preliminary solution ideas, such as "suggest reducing draft multiplier or checking gear strength".

[0045] This embodiment enables the closed-loop feedback engine to have the ability of real-time diagnosis and intelligent guidance based on domain knowledge by constructing a cross-disciplinary parameter coupling relationship rule base and configuring a signal generation logic module, which can convert massive simulation data into attention prompt signals that precisely locate design conflicts and optimization paths, to solve the core problems of information overload and decision blindness in multidisciplinary collaborative design, and thus significantly improve the accuracy and efficiency of design iteration.

[0046] As an example, the large language model performs multidisciplinary coupling analysis using the attention prompt signal, focusing on the key parameters and key disciplines indicated by the attention prompt signal, and dynamically adjusting task planning and design parameters, including: step S41: the large language model analyzes the attention prompt signal to obtain the conflict level, focus parameter, associated disciplines, and recommended optimization direction.

[0047] In this step, the large language model deeply analyzes the received attention prompt signal and extracts all key information elements contained therein, including: conflict level, used to quantify the severity of the current design conflict, such as "high risk", "medium risk", or "warning"; focus parameter, used to indicate the core parameter that triggers the conflict, such as "gear contact stress"; associated disciplines, used to indicate the professional fields related to the conflict, such as "process" and "transmission"; and optimization direction, used to provide preliminary solution ideas, such as "suggest reducing draft factor or checking gear strength".

[0048] Step S42: determining parameter optimization priority based on the conflict level, determining the core parameter set to be optimized and the related professional agent model based on the focus parameter, the associated disciplines, and generating a specific parameter adjustment instruction set based on the optimization direction, which specifies the parameter identifier, adjustment direction, and adjustment amplitude range that need to be adjusted.

[0049] In this step, the large language model makes the following comprehensive decisions based on the information analyzed in step S41: (1) determining parameter optimization priority: according to the conflict level, the large language model decides the execution order of the optimization task. For example, the optimization of "high risk" conflict is placed at the highest priority to ensure that the system resources are prioritized to solve the most urgent problems.

[0050] (2) Locking optimization target: combining the focus parameter and the associated disciplines, accurately determine the core parameter set to be optimized and the related professional agent model. For example, the focus parameter is "gear contact stress", and the associated disciplines are "process" and "transmission", then the core parameter set includes "draft factor" in the process field and "gear modulus" in the transmission field, and the related professional agent model is "process small model" and "transmission small model".

[0051] (3) Generating specific instructions: referring to the optimization direction, generate a specific parameter adjustment instruction set. It should be noted that this parameter adjustment instruction set is an operation list that can be directly executed by the professional agent model, including: parameter identifier, such as process, draft factor; adjustment direction, such as reduction; adjustment amplitude range, such as 5%-10%.

[0052] Step S43: According to the parameter optimization priority, the professional agent model corresponding to the core parameter set is called in sequence to execute the parameter adjustment instruction set.

[0053] In this step, the large language model acts as a dispatch center and calls the professional agent model corresponding to the core parameter set determined in step S42 in sequence according to the parameter optimization priority determined in the previous steps, and makes it execute the parameter adjustment instruction set.

[0054] For example, for the conflict with the highest priority, the large language model first sends an instruction to the process small model, asking it to reduce the "drafting multiple" by 5%. After the process small model completes the calculation and returns the new "drafting force" data, the large language model calls the transmission small model again, uses the new process data to simulate the transmission performance, and verifies whether the "gear contact stress" has been reduced below the safety threshold.

[0055] This embodiment can convert macro conflict diagnosis into precise, quantifiable and ordered optimization operations by parsing the attention prompt signal into specific parameter optimization priorities, core parameter sets and executable parameter adjustment instruction sets, thereby realizing from problem identification to precise execution, and further significantly improving the pertinence, automation and iteration convergence efficiency of multidisciplinary coupling optimization.

[0056] As an example, the suggested optimization direction output by the closed-loop feedback engine is associated with a confidence level, and the generation of the specific parameter adjustment instruction set based on the suggested optimization direction includes: step S421: combining and mapping the conflict level and the confidence level to determine the global optimization strategy of this iteration, including an aggressive optimization strategy, a robust optimization strategy and an exploration verification strategy.

[0057] In this step, due to the significant differences in urgency and solution reliability of different design conflicts, using a unified optimization strategy will lead to low efficiency and even cause new design risks. Therefore, in this embodiment, the conflict level and the confidence level are combined and mapped to establish a global optimization strategy based on risk assessment: (1) when the conflict level is high-risk and the confidence level is high, since a serious design problem is faced and the solution is reliable, the risk needs to be quickly eliminated, so the aggressive optimization strategy is adopted.

[0058] (2) When the conflict level is high-risk but the confidence level is low, since the problem is serious but the solution is uncertain, the problem needs to be solved while controlling the potential risk, so the robust optimization strategy is adopted.

[0059] (3) When the conflict level is medium / low-risk and the confidence level is low, since it is in a relatively safe state, it is suitable for accumulating design knowledge through multi-directional exploration, so the exploration verification strategy is adopted.

[0060] Step S422: Select a generation template for parameter adjustment instructions according to the global optimization strategy. Specifically, if it is an aggressive optimization strategy, generate a single instruction with a large adjustment amplitude; if it is a robust optimization strategy, generate an instruction sequence containing a main adjustment instruction and a backup adjustment instruction; if it is an exploration verification strategy, generate multiple small-amplitude, different-direction parallel test instructions.

[0061] This step aims to transform the aforementioned determined, abstract global optimization strategy into specific, executable code generation logic. The specific implementation is as follows: the large language model selects a generation template for parameter adjustment instructions from a pre-defined instruction template library according to the determined global optimization strategy.

[0062] If the strategy is an aggressive optimization strategy, the first template is selected, and the logic of this template is to generate a single instruction with a large adjustment amplitude. For example, the instruction form is {parameter identifier: draft ratio; adjustment direction: decrease; adjustment amplitude: 15%}, and then a single large adjustment can be used to quickly approach the target.

[0063] If the strategy is a robust optimization strategy, the second template is selected, and the logic of this template is to generate an instruction sequence containing a main adjustment instruction and a backup adjustment instruction. For example, the sequence is [{main instruction: draft ratio decrease 10%}, {backup instruction: if the stress is still over the limit, increase the gear modulus by one gear}].

[0064] If the strategy is an exploration and verification strategy, the third template is selected, and the logic of this template is to generate multiple small-amplitude, different-direction parallel test instructions. For example, the instruction set is {test 1: draft ratio decrease 5%; test 2: draft ratio increase 3%; test 3: roller spacing increase 2%}. This logic is not to immediately solve the conflict, but to quickly obtain data on the impact of different parameter changes on the system through parallel experiments.

[0065] Step S423: Fill in the specific content of the proposed optimization direction into the selected generation template to form the final parameter adjustment instruction set.

[0066] This step is the instantiation process of the template, which fills in the specific engineering parameters into the template to form the final instructions that can drive the professional agent model to execute. Specifically, the large language model fills in the specific content of the proposed optimization direction (such as the proposal "decrease draft ratio" or "check gear strength") as the key parameter into the generation template selected in step S422.

[0067] For the aggressive optimization strategy, fill in the specific proposal "decrease draft ratio" into the single instruction template and automatically match a larger amplitude (such as 15%) to form the final instruction {parameter identifier: draft ratio; adjustment direction: decrease; adjustment amplitude: 15%}.

[0068] For the robust optimization strategy, "decrease draft ratio" is filled in as the main instruction content, and "check gear strength" is converted into backup instruction content to form an ordered instruction sequence.

[0069] For the exploration and verification strategy, a set of parallel test instructions for exploration is generated with "draft ratio" and "roller spacing" as core parameters.

[0070] The embodiment realizes the leap from single instruction generation to multi-strategy adaptive decision-making by establishing a combined mapping mechanism of conflict level and confidence, enabling the auxiliary design system to dynamically select aggressive optimization, robust optimization or exploration verification strategy according to the urgency of design risk and the reliability of the solution, and generate a matching parameter adjustment instruction set, thereby significantly improving the intelligent level and execution efficiency of the optimization process while ensuring design safety.

[0071] As an example, the conflict level and the confidence are combined and mapped to determine the global optimization strategy of this iteration, including: step S4211: constructing an adversarial evaluation network containing noise injection, taking the conflict level and the confidence as the input feature vector, and adding random Gaussian noise to generate a noisy feature vector.

[0072] Due to the complex nonlinear coupling relationship between spinning process parameters (such as draft ratio and yarn tension, spindle speed and energy consumption, etc.), and the dynamic change of the sensitivity of each parameter with the equipment working condition, simple fixed preset rules cannot achieve accurate strategy decision. In addition, there are interference factors such as equipment vibration and sensor error in the actual engineering environment, resulting in uncertainty in conflict level and confidence evaluation. Therefore, the embodiment introduces an intelligent decision-making mechanism based on deep learning, which realizes intelligent strategy decision-making with adversarial interference and adaptive weighting by constructing an adversarial evaluation network containing noise injection, feature weighting based on attention mechanism, and reinforcement learning strategy network, thereby improving the decision-making quality and robustness of the system in real industrial environment.

[0073] In this step, the conflict level is quantized as a continuous value in the interval [0, 1], where 0.8-1.0 represents high risk (such as excessive breakage rate), 0.5-0.8 represents medium risk (such as high energy consumption), and 0-0.5 represents low risk. At the same time, the confidence is also quantized as a value in the interval [0, 1], reflecting the reliability of the recommended optimization direction. These two features are combined into a two-dimensional feature vector [Xc, Xs], where Xc represents the conflict level and Xs represents the confidence.

[0074] Next, Gaussian noise with mean 0 and standard deviation 0.1 is added to the feature vector [Xc, Xs] to generate a noisy feature vector [Xc+εc, Xs+εs], i.e. interference signal. The noise level is based on statistical analysis of spinning equipment parameter measurement error, which can effectively simulate the measurement fluctuations in actual production.

[0075] Through this adversarial training method, the network can learn to maintain stable feature representation under noise interference, which is particularly suitable for handling parameter noise problems specific to spinning equipment such as draft force fluctuations and spindle speed measurement errors.

[0076] Step S4212: The noisy feature vector is nonlinearly transformed by a multilayer perceptron to extract an anti-interference deep feature representation. Based on the anti-interference deep feature representation, the dynamic weights of the conflict level feature and the confidence feature are calculated using an attention mechanism. The anti-interference deep feature representation is then weighted and fused to generate a strategy preference score.

[0077] In this step, the noisy feature vector is input into a three-layer MLP network, which includes: an input layer with two neurons that receive the noisy feature vector [Xc+εc, Xs+εs].

[0078] Hidden layer: Nonlinear transformation is performed through 16 neurons, using the ReLU activation function.

[0079] Output layer: 8 neurons, outputting an 8-dimensional robust deep feature representation H={h1,h2,...,h8}. This deep feature representation H has been trained for noise robustness and can effectively resist interference signals in the input data.

[0080] Subsequently, the deep feature representation is dynamically weighted based on an attention mechanism. Since the components in the deep feature H contribute differently to the final policy decision, a learnable attention parameter vector is used. Calculate the weight coefficients for each feature: ;in, It is a learnable attention parameter vector. Let j be the depth feature representation.

[0081] This weighting mechanism can automatically adjust the importance of each feature based on the current design state. For example, in high-speed spinning, vibration-related features are given higher weights; in precision spinning scenarios, tension control features are given higher weights.

[0082] Weighted fusion of deep feature representations: Finally, the weighted comprehensive feature representation that reflects the key information of the current design status is... Feature representation Input three parallel scoring functions: ;in, These are scoring functions for aggressive, robust, and exploratory strategies, respectively. Through training with a large amount of spinning equipment design data, they can accurately assess the applicability of each strategy in the current design state.

[0083] Step S4213: Input the policy preference score into the reinforcement learning policy network, combine it with the feedback of historical optimization results, and output the global optimization policy for this iteration.

[0084] In this step, a reinforcement learning policy network based on DQN is constructed, whose state space includes:

[0085] Current policy tendency score .

[0086] Historical optimization effect index: for example, the target function improvement rate of the last 5 iterations.

[0087] Device running state features: for example, cumulative running time, key component wear coefficient.

[0088] The action space is the selection of three strategies. The reward function is designed as follows: ; wherein, is the target function improvement amount, is the constraint violation improvement amount, is the negative reward of computing resource consumption; the weight coefficient Adjust according to the specific spinning process requirements.

[0089] It can be understood that, is used to quantify the improvement degree of the optimization process to the main design target. In the design of spinning equipment, the target function is usually a comprehensive embodiment of multiple targets, including: production efficiency improvement rate, by comparing the theoretical output before and after iteration, ∆output=(new scheme output-original scheme output) / original scheme output; quality index improvement degree, including yarn CV value improvement, broken end rate reduction, etc., calculated according to the weighted improvement amount of each quality index; energy consumption reduction rate, the reduction amplitude of equipment running energy consumption is calculated through the power model.

[0090] The specific calculation formula is: . Wherein is the weight coefficient of each target, which is determined according to the specific design requirements.

[0091] It is used to evaluate the improvement of the satisfaction degree of the design scheme to various constraint conditions. The constraints in the design of spinning equipment include: process constraints, such as draft ratio range, twist range, etc.; structure constraints, such as maximum allowable stress, deformation, etc.; performance constraints, such as vibration amplitude limit, noise level, etc.

[0092] The constraint violation degree is calculated by the constraint violation amount function: , wherein, is the inequality constraint function, is the equality constraint function.

[0093] . Wherein, the positive value indicates that the constraint violation degree is reduced, and the feasibility of the design scheme is improved.

[0094] The calculation cost of the control optimization process is used to avoid excessive consumption of resources. It includes: simulation calculation time cost, the total time of each professional agent model executing simulation; data transmission cost, the resource consumption of data transmission and processing between system modules; storage resource cost, the storage demand of intermediate results and iteration history. The specific calculation method is: ; wherein, is the normalized value of simulation time, is the normalized value of data transmission volume, is the normalized value of storage occupancy, is the corresponding cost coefficient.

[0095] The reinforcement learning strategy network based on DQN is trained through the Q-learning algorithm: ; wherein, state includes the current policy tendency and device state, indicating the environment state perceived by the reinforcement learning strategy network in the current iteration period; action corresponds to policy selection, indicating the global optimization strategy selected by the reinforcement learning strategy network in this iteration from the action space, i.e. selecting one from {aggressive optimization strategy, robust optimization strategy, exploration and verification strategy}; reward is calculated based on the optimization effect, representing the feedback income obtained immediately after executing action a (i.e. adopting a certain optimization strategy) and completing this iteration; indicates the new environment state perceived by the policy network after executing action a and entering the next iteration; indicates all possible actions in the next state . In this way, the large language model can learn the optimization rules in the long-term operation of the spinning equipment, and tend to choose aggressive strategy for fast optimization in the new equipment stage, and switch to robust strategy to ensure stability in the equipment wear period, realizing adaptive optimization in the whole life cycle.

[0096] This embodiment simulates the parameter fluctuation in the actual working condition of the spinning equipment by constructing an adversarial evaluation network containing noise injection, extracts anti-interference deep features using a multi-layer perception machine, and realizes dynamic weight allocation of conflict level and confidence features based on an attention mechanism. Finally, the reinforcement learning strategy network combines the historical optimization effect to realize adaptive decision of the global optimization strategy. In this way, the strategy decision problem of traditional methods in complex working conditions such as multi-parameter coupling of spinning equipment, measurement noise interference and long-term running performance degradation can be effectively solved.

[0097] As Figure 3As shown, the embodiment of the application discloses a spinning equipment design auxiliary system 200 based on model cooperation, which comprises a processing unit 2001 and a storage unit 2002. The processing unit 2001 calls and executes a computer program pre-stored in the storage unit 2002 to perform the following steps: a collaborative design system is constructed, which comprises: a large language model fine-tuned by domain knowledge as a total planning and decision unit; a plurality of field-specific generative professional agent models for performing professional calculation and simulation of each discipline; and a closed-loop feedback engine with signal generation capability.

[0098] The large language model receives user design requirements, performs task decomposition, and calls the corresponding professional agent model to execute design subtasks. Each professional agent model feeds back simulation calculation results to the closed-loop feedback engine.

[0099] The closed-loop feedback engine analyzes the feedback results of each professional agent model and generates attention prompt signals pointing to potential design conflicts or optimization bottlenecks based on cross-disciplinary parameter coupling relationships.

[0100] The large language model performs multi-disciplinary coupling analysis using the attention prompt signals, focuses on key parameters and key disciplines pointed by the attention prompt signals, dynamically adjusts task planning and design parameters, and drives related professional agent models to perform directional iterative optimization.

[0101] It should be noted that the spinning equipment design auxiliary system 200 based on model cooperation of the present application should also include at least an I / O interface 2003, a network port 2004, a bus 2005, etc., which will not be described in detail.

[0102] As an example, the large language model is fine-tuned using domain knowledge, including: obtaining a professional text dataset in the field of spinning equipment design, including design manuals, academic literature and historical design schemes; constructing training samples based on the professional text dataset, and performing domain adaptation training on the pre-trained large language model using the instruction fine-tuning method; integrating the trained large language model with a retrieval and enhancement generation framework, so that it can retrieve and integrate related knowledge from a professional knowledge base based on design context, to enhance the professionalism and accuracy of its design decisions.

[0103] As an example, the closed-loop feedback engine with signal generation capability is constructed in the following way: a cross-disciplinary parameter coupling relationship rule library is constructed, which is constructed based on a spinning equipment multi-disciplinary design knowledge graph, and defines the constraint relationship, conflict mode and causal relationship among process, control, transmission, rack and modeling domain parameters.

[0104] The configuration signal generation logic module is configured to receive and integrate simulation calculation results of each professional agent model, perform real-time matching analysis on the simulation calculation results and the rule base, identify parameter deviations exceeding a preset threshold or conflict combinations violating design constraints, and generate a structured attention prompt signal based on the identification results, the attention prompt signal including at least a focus parameter, an associated discipline, a conflict level, and a recommended optimization direction.

[0105] As an example, the large language model performs multi-disciplinary coupling analysis using the attention prompt signal, focuses on key parameters and key disciplines indicated by the attention prompt signal, and dynamically adjusts task planning and design parameters, including: the large language model analyzes the attention prompt signal to obtain a conflict level, a focus parameter, an associated discipline, and a recommended optimization direction.

[0106] Based on the conflict level, a parameter optimization priority is determined, based on the focus parameter and the associated discipline, a core parameter set to be optimized and related professional agent models are determined, and based on the recommended optimization direction, a specific parameter adjustment instruction set is generated, the parameter adjustment instruction set specifying a parameter identifier to be adjusted, an adjustment direction, and an adjustment amplitude range.

[0107] According to the parameter optimization priority, the professional agent models corresponding to the core parameter set are sequentially called to execute the parameter adjustment instruction set.

[0108] As an example, the recommended optimization direction output by the closed-loop feedback engine is associated with a confidence level, and the specific parameter adjustment instruction set is generated based on the recommended optimization direction, including: the conflict level and the confidence level are combined and mapped to determine a global optimization strategy for this iteration, including an aggressive optimization strategy, a robust optimization strategy, and an exploration and verification strategy.

[0109] According to the global optimization strategy, a generation template for parameter adjustment instructions is selected, specifically: if the aggressive optimization strategy is selected, a single instruction with a large adjustment amplitude is generated; if the robust optimization strategy is selected, a sequence of instructions including a main adjustment instruction and a backup adjustment instruction is generated; if the exploration and verification strategy is selected, multiple small-amplitude, different-direction parallel test instructions are generated.

[0110] The specific content of the recommended optimization direction is filled into the selected generation template to form the final parameter adjustment instruction set.

[0111] As an example, the conflict level and the confidence level are combined and mapped to determine a global optimization strategy for this iteration, including: an adversarial evaluation network containing noise injection is constructed, the conflict level and the confidence level are taken as input feature vectors, and random Gaussian noise is added to generate a noisy feature vector.

[0112] The noise-containing feature vector is nonlinearly transformed by a multilayer perceptron to extract an anti-interference deep feature representation; and a dynamic weight distribution between a conflict level feature and a confidence feature is calculated based on an attention mechanism to generate a strategy inclination score.

[0113] The strategy inclination score is input into a reinforcement learning strategy network, combined with historical optimization effect feedback, to output a global optimization strategy for this iteration.

[0114] Although the present application has been particularly shown and described with reference to preferred embodiments, it will be understood by those skilled in the art that various changes in form and details can be made therein without departing from the spirit and scope of the application. Accordingly, the disclosed application is to be considered as illustrative only and is limited by the scope of the appended claims.

Claims

1. A model-based collaborative design assistance method for spinning equipment, characterized in that, The process includes the following steps: constructing a collaborative design system, comprising: a large language model fine-tuned with domain knowledge as the overall planning and decision-making unit; multiple domain-specific generative professional agent models for performing professional calculations and simulations in each discipline; and a closed-loop feedback engine with signal generation capabilities. The large language model receives user design requirements, decomposes the tasks, and calls the corresponding professional agent models to execute design sub-tasks. Each professional agent model feeds back simulation calculation results to the closed-loop feedback engine. The closed-loop feedback engine analyzes the feedback results of each professional agent model and, based on cross-disciplinary parameter coupling relationships, generates attention prompts pointing to potential design conflicts or optimization bottlenecks. The language model utilizes the attention cues for multidisciplinary coupling analysis, focusing on the key parameters and disciplines indicated by the attention cues, dynamically adjusting task planning and design parameters, and driving related professional agent models for targeted iterative optimization. The large language model also utilizes the attention cues for multidisciplinary coupling analysis, focusing on the key parameters and disciplines indicated by the attention cues, and dynamically adjusting task planning and design parameters. This includes: the large language model parsing the attention cues to obtain conflict levels, focus parameters, related disciplines, and suggested optimization directions; determining parameter optimization priorities based on the conflict levels; and determining the core parameters to be optimized based on the focus parameters and related disciplines. The core parameter set and related professional agent models are used to generate a specific parameter adjustment instruction set based on the suggested optimization direction. This instruction set specifies the parameter identifier, adjustment direction, and adjustment range to be adjusted. According to the parameter optimization priority, the professional agent models corresponding to the core parameter set are sequentially invoked to execute the parameter adjustment instruction set. Since the suggested optimization direction output by the closed-loop feedback engine is associated with a confidence level, generating the specific parameter adjustment instruction set based on the suggested optimization direction includes: combining and mapping the conflict level with the confidence level to determine the global optimization strategy for this iteration, including an aggressive optimization strategy, a robust optimization strategy, and an exploratory verification strategy; and selecting the appropriate strategy based on the global optimization strategy. The parameter adjustment instruction generation template is as follows: if it is an aggressive optimization strategy, a single instruction with a large adjustment range is generated; if it is a robust optimization strategy, an instruction sequence containing a main adjustment instruction and a backup adjustment instruction is generated; if it is an exploratory verification strategy, multiple parallel test instructions with small amplitudes and different directions are generated. The specific content of the suggested optimization direction is filled into the selected generation template to form the final parameter adjustment instruction set. The closed-loop feedback engine with signal generation capability is constructed in the following way: a cross-disciplinary parameter coupling relationship rule base is constructed. This rule base is based on the multi-disciplinary design knowledge graph of spinning equipment and defines the constraint relationships, conflict modes and causal relationships between parameters in the fields of process, control, transmission, frame and shape.The configuration signal generation logic module is set to: receive and integrate the simulation calculation results of various professional agent models, perform real-time matching analysis with the rule base, identify parameter deviations exceeding preset thresholds or conflict combinations violating design constraints; and generate structured attention cue signals based on the identification results, including at least the focus parameter, related disciplines, conflict level, and suggested optimization direction.

2. The model-based collaborative design assistance method for spinning equipment according to claim 1, characterized in that: Fine-tuning the large language model using domain knowledge includes: acquiring a professional text dataset in the field of spinning equipment design, including design manuals, academic literature, and historical design schemes; constructing training samples based on the professional text dataset; and performing domain adaptation training on the pre-trained large language model using an instruction fine-tuning method; integrating the trained large language model with a retrieval enhancement generation framework, enabling it to retrieve and integrate relevant knowledge from a professional knowledge base based on the design context, thereby enhancing the professionalism and accuracy of its design decisions; wherein, the professional knowledge base is constructed from the professional text dataset after vectorization processing.

3. The model-based collaborative design assistance method for spinning equipment according to claim 1, characterized in that: The process of combining and mapping the conflict level with the confidence level to determine the global optimization strategy for this iteration includes: constructing an adversarial evaluation network with noise injection, using the conflict level and confidence level as input feature vectors, and adding random Gaussian noise to generate noisy feature vectors; performing a nonlinear transformation on the noisy feature vectors using a multilayer perceptron to extract anti-interference deep feature representations; calculating the dynamic weights of the conflict level features and confidence level features based on the anti-interference deep feature representations using an attention mechanism, and performing weighted fusion on the anti-interference deep feature representations to generate a strategy preference score; inputting the strategy preference score into a reinforcement learning policy network, and combining feedback from historical optimization effects to output the global optimization strategy for this iteration.

4. A model-based collaborative design assistance system for spinning equipment, characterized in that, The system includes a processing unit and a storage unit. The processing unit calls and executes pre-stored computer programs in the storage unit to perform the following steps: constructing a collaborative design system, which includes: a large language model fine-tuned with domain knowledge as the overall planning and decision-making unit; multiple domain-specific generative professional agent models for performing professional calculations and simulations in various disciplines; and a closed-loop feedback engine with signal generation capabilities. The large language model receives user design requirements, decomposes tasks, and calls the corresponding professional agent models to execute design sub-tasks. Each professional agent model feeds back simulation calculation results to the closed-loop feedback engine. The feedback engine analyzes the feedback results of various professional agent models and, based on interdisciplinary parameter coupling relationships, generates attention cue signals pointing to potential design conflicts or optimization bottlenecks. The large language model utilizes these attention cue signals for multidisciplinary coupling analysis, focusing on the key parameters and disciplines indicated by the attention cue signals, dynamically adjusting task planning and design parameters, and driving relevant professional agent models to perform targeted iterative optimization. The large language model's multidisciplinary coupling analysis, focusing on the key parameters and disciplines indicated by the attention cue signals, and dynamically adjusting task planning and design parameters, includes: the large language model parsing the attention cue signals... The system receives a prompt signal, obtains the conflict level, focus parameters, related disciplines, and suggested optimization directions; determines the parameter optimization priority based on the conflict level, identifies the core parameter set to be optimized and related professional agent models based on the focus parameters and related disciplines, and generates a specific parameter adjustment instruction set based on the suggested optimization directions. This instruction set specifies the parameter identifier, adjustment direction, and adjustment range to be adjusted. According to the parameter optimization priority, the professional agent models corresponding to the core parameter set are sequentially invoked to execute the parameter adjustment instruction set. If the suggested optimization directions output by the closed-loop feedback engine have a confidence level, then the optimization based on the suggested optimization direction... The process of generating a specific set of parameter adjustment instructions includes: mapping the conflict level to the confidence level to determine the global optimization strategy for this iteration, including an aggressive optimization strategy, a robust optimization strategy, and an exploratory verification strategy; selecting a template for generating parameter adjustment instructions based on the global optimization strategy, specifically: if it is an aggressive optimization strategy, generating a single instruction with a large adjustment range; if it is a robust optimization strategy, generating an instruction sequence containing a main adjustment instruction and backup adjustment instructions; if it is an exploratory verification strategy, generating multiple parallel test instructions with small amplitudes in different directions; and filling the specific content of the suggested optimization direction into the selected template to form the final set of parameter adjustment instructions.A closed-loop feedback engine with signal generation capabilities is constructed as follows: A cross-disciplinary parameter coupling relationship rule base is built, based on a multi-disciplinary design knowledge graph of spinning equipment, defining constraint relationships, conflict modes, and causal relationships between parameters in the fields of process, control, transmission, frame, and styling; a signal generation logic module is configured to: receive and integrate simulation calculation results from various professional proxy models, perform real-time matching analysis with the rule base, identify parameter deviations exceeding preset thresholds or conflict combinations violating design constraints; and generate structured attention prompt signals based on the identification results, including at least the focus parameter, related disciplines, conflict level, and suggested optimization direction.

Images