Mechanical design system based on multi-performance index cooperative control

Abstract

The invention belongs to the technical field of mechanical design, and discloses a mechanical design system based on multi-performance index cooperative control. The system comprises a self-adaptive control instruction generation module, a parameter adjustment instruction generation module, a machine learning model training module and a collaborative optimization feedback decision module, and is characterized in that a fusion data report is calculated by using a self-adaptive control algorithm to obtain an initial control instruction set; converting the initial control instruction set into an executable final parameter adjustment instruction set, generating an instruction report, training a machine learning model based on a historical fusion data report and the instruction report to obtain a final machine learning model, running a multi-objective optimization algorithm based on the final machine learning model to obtain an optimization feedback instruction report, and performing optimization on the machine learning model. Generally speaking, the method has the remarkable advantages of being high in data fusion cooperation capability, good in control strategy self-adaption effect and large in system continuous learning effect.

Description

Technical Field

[0001] This invention relates to the field of mechanical design technology, and more specifically, to a mechanical design system based on the coordinated control of multiple performance indicators. Background Technology

[0002] Mechanical design system refers to a comprehensive system that uses computer-aided technology to design product structure, performance, and process. In modern industry, with the increasing complexity of products, the requirements for mechanical design have changed from optimizing a single performance index (such as strength) to the coordinated control and optimization of multiple performance indicators (such as stress, temperature, vibration, cost, weight, etc.).

[0003] However, traditional mechanical design systems suffer from several shortcomings in practice. First, the evaluation process in traditional systems is often independent, with design data, simulation data, and operational data remaining isolated. This makes it difficult for staff to comprehensively and in real-time acquire the interconnections and changes between all indicators, leading to a situation where mechanical design requires multiple independent single-step optimizations, resulting in low design efficiency. Second, the control and optimization methods of traditional systems are mostly based on preset and fixed rules or models. When design parameters or external conditions change, traditional systems struggle to adapt, often sacrificing other indicators to meet a single metric. Third, traditional systems are largely limited to completing design tasks, relying on preset models or human experience for decision-making. Consequently, the system lacks learning capabilities, often requiring it to start from scratch when faced with new design tasks. This makes it difficult for the system to support continuous improvement and innovative design throughout the product lifecycle. In summary, effectively addressing the problems of weak data collaboration, insufficient adaptive control strategies, and a lack of adaptive decision-making capabilities in traditional systems has become a challenge that current mechanical design systems need to address.

[0004] In view of this, the present invention proposes a mechanical design system based on the coordinated control of multiple performance indicators to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution, including: The multi-source data acquisition module is used to collect multi-performance indicator data in real time based on external data sources, and to perform integrity verification to obtain the raw dataset; Furthermore, the steps for real-time collection of multiple performance indicator data from external data sources and subsequent integrity verification include: S1.1: Based on an external data source, continuously receive raw data items to obtain an external dataset. Each data item includes a collection timestamp, a data source identifier, and a metric value. S1.2: Perform data validation on external data items. When the validation result is invalid, remove the data items with invalid validation results and record them to obtain the original validation dataset and error log. At the same time, store the error log in the database. When the validation result is valid, obtain the original validation dataset and proceed to step S1.3. S1.3: Encapsulate the data items in the original validation dataset into raw data instructions, package all raw data instructions to obtain the original dataset. The structure of the raw data instructions includes the collection timestamp, data source identifier, and indicator value item. S1.4: Output the raw dataset to the data preprocessing module; The data preprocessing module is used to preprocess the original dataset to obtain a preprocessed dataset; Furthermore, the preprocessing steps based on the original dataset include: S2.1: Based on the original dataset, a moving average filter is used to smooth each index value in the original dataset to obtain a denoised dataset. The specific calculation formula for the smoothing process is as follows: ; Get the time point Value of the filtered index item ,in, To adjust the sliding window size, For time points The index value item; S2.2: The 3σ criterion method is used to remove outliers from the index values ​​in the denoised dataset, and linear interpolation is used to supplement the removed outliers to obtain the denoised dataset. S2.3: Normalize all index values ​​in the removed dataset using a normalization formula to obtain a preprocessed dataset. The normalization process is used to normalize each index value to the range of [0,1]. S2.4: Output the preprocessed dataset to the real-time data fusion and storage module; The real-time data fusion and storage module is used to fuse preprocessed datasets from multiple sources and perform time-series alignment to obtain a fused data report. Furthermore, the steps of fusing multi-source data and performing time-series alignment on the preprocessed dataset include: S3.1: Based on the preprocessed dataset, perform multi-source numerical fusion on each indicator value item in the preprocessed dataset to obtain the fused dataset; S3.2: Based on the collection timestamp, detect all fusion index values ​​in the fusion dataset. When there is missing data in the Cth timestamp, use linear interpolation to fill it in and obtain the fusion data report. S3.3: Store the merged data report in the database and provide an API interface to support querying historical data; S3.4: Output the fused data report to the adaptive control command generation module; The adaptive control instruction generation module is used to calculate the initial control instruction set by using the adaptive control algorithm on the fused data report. Furthermore, the steps for calculating the fused data report using the adaptive control algorithm include: S4.1: Based on the fusion data report, all fusion index values ​​in the fusion data report are used as the system state vector, and the size adjustment and material parameter adjustment values ​​retrieved from the database are used as the control input vector; S4.2: Based on the system state vector and control input vector, and using a model control algorithm, the optimal control parameters are obtained. The optimization problem expression of the model control algorithm is as follows: ; in, Let be the objective function. To predict the time domain, To predict the first in the time domain The system state vector of the step, For transpose, Here is the state penalty weight matrix. To predict the first in the time domain The control input vector of the step, To control the penalty weight matrix, For terminal penalty items, The terminal state penalty weight matrix; The constraints of the model control algorithm are: ; in, For the system matrix, The state matrix; The model control algorithm is solved to obtain the optimal control parameter sequence, and the first control parameter in the optimal control parameter sequence is selected as the optimal control parameter. S4.3: Convert the optimal control parameters into parameter adjustment values ​​and integrate the calculation timestamps to obtain the initial control instruction set; S4.4: Output the initial control instruction set to the parameter adjustment instruction generation module; The parameter adjustment instruction generation module is used to convert the initial control instruction set into an executable final parameter adjustment instruction set and generate an instruction report; Furthermore, the steps of converting the initial control instruction set into an executable final parameter tuning instruction set and generating an instruction report include: S5.1: Based on the initial control instruction set, a conversion is performed to obtain the final parameter adjustment instruction set; S5.2: Verify the final parameter adjustment instruction set. If the verification result is an allowed result, proceed to step S5.3. If the verification result is an unallowed result, adjust the adjustment instructions that exceed the preset adjustment threshold range to the nearest boundary value, and proceed to step S5.3. S5.3: Send the final parameter adjustment instruction set output in step S5.2 to the control system for execution. At the same time, generate an instruction report, which includes the execution timestamp, the final parameter adjustment instruction set in step S5.1, and the final parameter adjustment instruction set in step S5.2. S5.4: Store the instruction report in the database; The machine learning model training module is used to train the machine learning model based on historical fused data reports and instruction reports to obtain the final machine learning model; Furthermore, the steps for training the machine learning model based on historical fused data reports and instruction reports include: S6.1: Retrieve historical fusion data reports and instruction reports from the database, and extract features to obtain feature vectors and target vectors; S6.2: Perform feature extraction based on the fused data report and instruction report to obtain the current feature vector. The current feature vector includes the fused index value and the final parameter adjustment instruction set in step S5.2. S6.3: Based on the data, retrieve the machine learning model and train it according to the feature vector, target vector, and current feature vector to obtain the first machine learning model. The structural expression of the machine learning model is as follows: ; in, For activation function, and This is the weight matrix. and For bias vectors, The input vector; The training objective of a machine learning model is to minimize the mean squared error loss function, specifically calculated using the following formula: ; Obtain the loss function value ,in, The number of training samples, For the first The target vector of each training sample For the first The true target vector of each training sample; The gradient descent method is used to iteratively optimize the weight matrix of the machine learning model; S6.4: Evaluate the first machine learning model based on the feature vector, target vector, and current feature vector. If the evaluation result indicates that the performance meets the standard, proceed to step S6.5. If the evaluation result indicates that the performance does not meet the standard, return to step S6.3 to adjust the model parameters and retrain the machine learning model. S6.5: Save the first machine learning model output in step S6.4 as the final machine learning model and output it to the collaborative optimization feedback decision module; The collaborative optimization feedback decision module is used to run a multi-objective optimization algorithm based on the final machine learning model and obtain an optimization feedback instruction report. Furthermore, the steps for running a multi-objective optimization algorithm based on the final machine learning model include: S7.1: Load the final machine learning model; S7.2: Define the multi-objective optimization problem and obtain the optimization objective function, which includes minimizing weight, minimizing cost, and maximizing strength; S7.3: The NSGA-II fast non-dominated sorting genetic algorithm II is used to solve the optimization objective function to obtain the non-dominated solution set. Each solution in the non-dominated solution set is evaluated for the objective value through the loaded final machine learning model to obtain the objective evaluation value set. S7.4: Based on step S7.3, select the solution with the largest objective evaluation value in the non-dominated solution set as the optimal solution; S7.5: Convert the optimal solution into an optimization feedback instruction report and output it to step S4.2. The optimization feedback instruction report includes suggested values ​​for size adjustment and material parameter adjustment. Furthermore, S1: Collect multiple performance index data in real time based on external data sources and perform integrity verification to obtain the original dataset; S2: Preprocess the original dataset to obtain the preprocessed dataset; S3: Perform multi-source data fusion on the preprocessed dataset and time-series alignment to obtain a fused data report; S4: Calculate the fused data report using an adaptive control algorithm to obtain the initial control instruction set; S5: Convert the initial control instruction set into an executable final parameter tuning instruction set and generate an instruction report; S6: Train the machine learning model based on historical fusion data reports and instruction reports to obtain the final machine learning model; S7: Run a multi-objective optimization algorithm based on the final machine learning model to obtain an optimization feedback instruction report.

[0006] The technical effects and advantages of the mechanical design system based on multi-performance index collaborative control of the present invention are as follows: This invention obtains a raw dataset by real-time acquisition of multiple performance index data from external data sources and performing integrity verification. The raw dataset is then preprocessed to obtain a preprocessed dataset. This preprocessed dataset is then fused from multiple sources and time-series aligned to generate a fused data report. An adaptive control algorithm is used to calculate the fused data report to obtain an initial control instruction set. This initial control instruction set is then converted into an executable final parameter adjustment instruction set, generating an instruction report. A machine learning model is trained based on historical fused data reports and instruction reports to obtain a final machine learning model. Finally, a multi-objective optimization algorithm is run on the final machine learning model to obtain an optimization feedback instruction report. This allows the system to effectively solve the data silos present in traditional systems through the collaborative operation of the multi-source data acquisition module, data preprocessing module, and real-time data fusion and storage module. The island phenomenon enables staff and subsequent modules to make decisions based on a comprehensive, consistent, and high-quality dataset, thus providing a solid data foundation for collaborative control. Furthermore, the invention, through the establishment of an adaptive control instruction generation module and a parameter adjustment instruction generation module, allows the system to calculate the optimal design parameter adjustment amount in real time and automatically generate the final parameter adjustment instruction set to apply to the design model. This ensures the system maximizes the optimization of multiple performance indicators. Finally, through the collaborative operation of the machine learning model training module and the collaborative optimization feedback decision module, the system not only solves current problems but also continuously learns and accumulates design knowledge, achieving intelligent design strategies. Overall, this invention has significant advantages such as strong data fusion and collaboration capabilities, good adaptive control strategy effects, and a large system continuous learning capacity. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of a mechanical design system based on multi-performance index collaborative control according to the present invention; Figure 2 This is a schematic diagram of a mechanical design method based on the coordinated control of multiple performance indicators according to the present invention. Detailed Implementation

[0008] 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, and 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.

[0009] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” used in the embodiments of this invention are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0010] Depending on the context, the words “if” or “suppose” as used here can be interpreted as “when” or “in response to determination” or “in response to detection.” Similarly, depending on the context, the phrases “if determination” or “if detection (of the stated condition or event)” can be interpreted as “when determination” or “in response to determination” or “when detection (of the stated condition or event)” or “in response to detection (of the stated condition or event).”

[0011] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0012] In practice, the server-side equipment deployed in a mechanical design system based on multi-performance index collaborative control may consist of one or more devices. This system can be implemented as a business instance, a virtual machine, or hardware devices. For example, it can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, it can be understood as software deployed on a cloud node, providing a mechanical design system based on multi-performance index collaborative control to various user terminals. Alternatively, it can be implemented as a virtual machine deployed on one or more devices in a cloud node, with application software installed to manage various user terminals. Or, it can also be implemented as a server composed of numerous identical or different types of hardware devices, with one or more devices configured to provide the mechanical design system based on multi-performance index collaborative control to various user terminals.

[0013] In terms of implementation, the mechanical design system based on multi-performance index collaborative control and the user terminal are mutually compatible. That is, if the mechanical design system based on multi-performance index collaborative control is implemented as an application installed on a cloud service platform, then the user terminal is a client that establishes a communication connection with the application; or if the mechanical design system based on multi-performance index collaborative control is implemented as a website, then the user terminal is implemented as a webpage; or if the mechanical design system based on multi-performance index collaborative control is implemented as a cloud service platform, then the user terminal is implemented as a mini-program in an instant messaging application.

[0014] like Figure 1 The figure shown is a system architecture diagram of a mechanical design system based on multi-performance index collaborative control provided by an embodiment of the present invention.

[0015] The mechanical design system based on multi-performance index collaborative control described in this invention can be hosted on a cloud server. In terms of implementation, it can function as one or more service devices, or as an application installed in the cloud (e.g., a mobile service operator's server or server cluster), or it can be developed into a website. Depending on the functions implemented, the mechanical design system based on multi-performance index collaborative control may include a multi-source data acquisition module, a data preprocessing module, a real-time data fusion and storage module, an adaptive control instruction generation module, a parameter adjustment instruction generation module, a machine learning model training module, and a collaborative optimization feedback decision-making module. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.

[0016] In this embodiment of the invention, in the mechanical design system based on multi-performance index collaborative control, each of the above modules can be implemented independently and can call other modules. Here, "calling" can be understood as a module connecting to multiple modules of another type and providing corresponding services to those connected modules. For example, the collaborative optimization feedback decision module can call the same information acquisition module to obtain information collected by that module. Based on the above characteristics, in the mechanical design system based on multi-performance index collaborative control provided in this embodiment of the invention, the applicable scope of the mechanical design system architecture based on multi-performance index collaborative control can be adjusted by adding modules and directly calling them without modifying the program code, achieving cluster-based horizontal expansion, so as to achieve the purpose of quickly and flexibly expanding the mechanical design system based on multi-performance index collaborative control. In practical applications, the above modules can be set in the same device or different devices, or they can be set in a virtual device, such as a service instance in a cloud server.

[0017] Please refer to Example 1 Figure 1 As shown in this embodiment, a mechanical design system based on multi-performance index collaborative control is described. The system includes: The multi-source data acquisition module is used to collect multi-performance indicator data in real time based on external data sources, and to perform integrity verification to obtain the original dataset. Furthermore, the steps for real-time collection of multiple performance indicator data from external data sources and verification of its integrity include: S1.1: Based on an external data source, continuously receive raw data items to obtain an external dataset. Each data item includes a collection timestamp, a data source identifier, and a metric value. It should be explained that the index values ​​include, but are not limited to, stress, temperature, vibration amplitude, and cost; S1.2: Perform data validation on external data items. When the validation result is invalid, remove the data items with invalid validation results and record them to obtain the original validation dataset and error log. At the same time, store the error log in the database. When the validation result is valid, obtain the original validation dataset and proceed to step S1.3. It should be explained that validation includes checking the completeness of fields and the reasonableness of data items. Completeness refers to, for example, the existence of required fields, while reasonableness of data items refers to, for example, whether the temperature index value is within a preset range. S1.3: Encapsulate the data items in the original validation dataset into raw data instructions, package all raw data instructions to obtain the original dataset. The structure of the raw data instructions includes the collection timestamp, data source identifier, and indicator value item. S1.4: Output the raw dataset to the data preprocessing module; The data preprocessing module is used to preprocess the original dataset to obtain a preprocessed dataset; Further, the preprocessing steps based on the original dataset include: S2.1: Based on the original dataset, a moving average filter is used to smooth each index value in the original dataset to obtain a denoised dataset. The specific calculation formula for the smoothing process is as follows: ; Get the time point Value of the filtered index item ,in, To adjust the sliding window size, For time points The index value item; S2.2: The 3σ criterion method is used to remove outliers from the index values ​​in the denoised dataset, and linear interpolation is used to supplement the removed outliers to obtain the denoised dataset. S2.3: Normalize all index values ​​in the removed dataset using a normalization formula to obtain a preprocessed dataset. The normalization process is used to normalize each index value to the range of [0,1]. S2.4: Output the preprocessed dataset to the real-time data fusion and storage module; The real-time data fusion and storage module is used to fuse preprocessed datasets from multiple sources and perform time-series alignment to obtain a fused data report. Furthermore, the steps of fusing multi-source data and performing time-series alignment on the preprocessed dataset include: S3.1: Based on the preprocessed dataset, perform multi-source numerical fusion on each indicator value item in the preprocessed dataset to obtain the fused dataset; It needs to be explained that, taking the stress index value term as an example, it is assumed that the stress index value term exists. If there are multiple data sources, the specific formula for fusion is: The stress fusion index value was obtained. ,in, For the first Weight coefficients of each data source, For the first Stress index values ​​from each data source; S3.2: Based on the collection timestamp, detect all fusion index values ​​in the fusion dataset. When there is missing data in the Cth timestamp, use linear interpolation to fill it in and obtain the fusion data report. S3.3: Store the merged data report in the database and provide an API interface to support querying historical data; S3.4: Output the fused data report to the adaptive control command generation module; The adaptive control instruction generation module is used to calculate the fused data report using an adaptive control algorithm to obtain an initial control instruction set; Furthermore, the steps for calculating the fused data report using the adaptive control algorithm include: S4.1: Based on the fusion data report, all fusion index values ​​in the fusion data report are used as the system state vector, and the size adjustment and material parameter adjustment values ​​retrieved from the database are used as the control input vector; S4.2: Based on the system state vector and control input vector, and using a model control algorithm, the optimal control parameters are obtained. The optimization problem expression of the model control algorithm is as follows: ; in, Let be the objective function. To predict the time domain, To predict the first in the time domain The system state vector of the step, For transpose, Here is the state penalty weight matrix. To predict the first in the time domain The control input vector of the step, To control the penalty weight matrix, For terminal penalty items, The terminal state penalty weight matrix; It should be explained that the terminal penalty term is used to penalize the prediction time-domain terminal. The state at any given moment ensures that the system remains in a stable state at the end of the prediction time domain; The constraints of the model control algorithm are: ; in, For the system matrix, The state matrix; It should be explained that the system matrix refers to the vector describing the current system state. How to evolve to the next system state vector The state matrix is ​​similar; The model control algorithm is solved to obtain the optimal control parameter sequence, and the first control parameter in the optimal control parameter sequence is selected as the optimal control parameter. S4.3: Convert the optimal control parameters into parameter adjustment values ​​and integrate the calculation timestamps to obtain the initial control instruction set; S4.4: Output the initial control instruction set to the parameter adjustment instruction generation module; The parameter adjustment instruction generation module is used to convert the initial control instruction set into an executable final parameter adjustment instruction set and generate an instruction report; Furthermore, the steps of converting the initial control instruction set into an executable final parameter tuning instruction set and generating an instruction report include: S5.1: Based on the initial control instruction set, a conversion is performed to obtain the final parameter adjustment instruction set; It should be explained that conversion refers to converting the initial control instruction set into the system instruction format; S5.2: Verify the final parameter adjustment instruction set. If the verification result is an allowed result, proceed to step S5.3. If the verification result is an unallowed result, adjust the adjustment instructions that exceed the preset adjustment threshold range to the nearest boundary value, and proceed to step S5.3. It should be explained that the preset adjustment threshold range is manually set and entered into the system; verification means, for example, when the size adjustment amount is greater than E2 in the preset adjustment threshold range (E1, E2), the size adjustment amount is adjusted to E2. S5.3: Send the final parameter adjustment instruction set output in step S5.2 to the control system for execution. At the same time, generate an instruction report, which includes the execution timestamp, the final parameter adjustment instruction set in step S5.1, and the final parameter adjustment instruction set in step S5.2. S5.4: Store the instruction report in the database; The machine learning model training module is used to train the machine learning model based on historical fusion data reports and instruction reports to obtain the final machine learning model. Furthermore, the steps for training the machine learning model based on historical fusion data reports and instruction reports include: S6.1: Retrieve historical fusion data reports and instruction reports from the database, and extract features to obtain feature vectors and target vectors; It should be explained that the feature vector refers to, for example, historical fusion index values ​​and the final parameter adjustment instruction set in step S5.2; the target vector refers to, for example, future fusion index values. S6.2: Perform feature extraction based on the fused data report and instruction report to obtain the current feature vector. The current feature vector includes the fused index value and the final parameter adjustment instruction set in step S5.2. S6.3: Based on the data, retrieve the machine learning model and train it according to the feature vector, target vector, and current feature vector to obtain the first machine learning model. The structural expression of the machine learning model is as follows: ; in, For activation function, and This is the weight matrix. and For bias vectors, The input vector; The training objective of a machine learning model is to minimize the mean squared error loss function, specifically calculated using the following formula: ; Obtain the loss function value ,in, The number of training samples, For the first The target vector of each training sample For the first The true target vector of each training sample; The gradient descent method is used to iteratively optimize the weight matrix of the machine learning model; S6.4: Evaluate the first machine learning model based on the feature vector, target vector, and current feature vector. If the evaluation result indicates that the performance meets the standard, proceed to step S6.5. If the evaluation result indicates that the performance does not meet the standard, return to step S6.3 to adjust the model parameters and retrain the machine learning model. S6.5: Save the first machine learning model output in step S6.4 as the final machine learning model and output it to the collaborative optimization feedback decision module; The collaborative optimization feedback decision module is used to run a multi-objective optimization algorithm based on the final machine learning model and obtain an optimization feedback instruction report. Furthermore, the steps for running a multi-objective optimization algorithm based on the final machine learning model include: S7.1: Load the final machine learning model; S7.2: Define the multi-objective optimization problem and obtain the optimization objective function, which includes minimizing weight, minimizing cost, and maximizing strength; S7.3: The NSGA-II fast non-dominated sorting genetic algorithm II is used to solve the optimization objective function to obtain the non-dominated solution set. Each solution in the non-dominated solution set is evaluated for the objective value through the loaded final machine learning model to obtain the objective evaluation value set. S7.4: Based on step S7.3, select the solution with the largest objective evaluation value in the non-dominated solution set as the optimal solution; S7.5: Convert the optimal solution into an optimization feedback instruction report and output it to step S4.2. The optimization feedback instruction report includes suggested values ​​for size adjustment and material parameter adjustment. It should be explained that outputting the optimization feedback instruction report to step S4.2 is used to update the penalty weight matrix in the model control algorithm; In this embodiment, the beneficial effects are achieved by acquiring multi-performance indicator data in real time from external data sources and verifying its integrity to obtain the original dataset. Preprocessing of the original dataset yields a preprocessed dataset. This preprocessed dataset is then fused from multiple sources and time-series aligned to generate a fused data report. An adaptive control algorithm is used to calculate the fused data report, resulting in an initial control instruction set. This initial control instruction set is then converted into an executable final parameter adjustment instruction set, generating an instruction report. A machine learning model is trained based on historical fused data reports and instruction reports to obtain the final machine learning model. Finally, a multi-objective optimization algorithm is run on the final machine learning model to obtain an optimization feedback instruction report. This allows the system to effectively solve the problems existing in traditional systems through the collaborative operation of the multi-source data acquisition module, data preprocessing module, and real-time data fusion and storage module. The data silo phenomenon enables staff and subsequent modules to make decisions based on a comprehensive, consistent, and high-quality dataset, thus providing a solid data foundation for collaborative control. Furthermore, the invention, through the establishment of adaptive control instruction generation and parameter adjustment instruction generation modules, allows the system to calculate the optimal design parameter adjustment amount in real time and automatically generate the final parameter adjustment instruction set to apply to the design model. This ensures the system maximizes the optimization of multiple performance indicators. Finally, through the collaborative operation of the machine learning model training module and the collaborative optimization feedback decision module, the system not only solves current problems but also continuously learns and accumulates design knowledge, achieving intelligent design strategies. Overall, this invention has significant advantages such as strong data fusion and collaboration capabilities, good adaptive control strategy effects, and a large system continuous learning capacity.

[0018] Please refer to Example 2 Figure 2 As shown, the parts not described in detail in this embodiment are described in Embodiment 1. A mechanical design method based on multi-performance index collaborative control is provided. The method includes: S1: Real-time collection of multi-performance index data based on external data sources and integrity verification to obtain the original dataset; S2: Preprocess the original dataset to obtain the preprocessed dataset; S3: Perform multi-source data fusion on the preprocessed dataset and time-series alignment to obtain a fused data report; S4: Calculate the fused data report using an adaptive control algorithm to obtain the initial control instruction set; S5: Convert the initial control instruction set into an executable final parameter tuning instruction set and generate an instruction report; S6: Train the machine learning model based on historical fusion data reports and instruction reports to obtain the final machine learning model; S7: Run a multi-objective optimization algorithm based on the final machine learning model to obtain an optimization feedback instruction report.

[0019] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the present invention.

Claims

1. A mechanical design system based on multi-performance index collaborative control, characterized in that, The system includes: an adaptive control command generation module, a parameter adjustment command generation module, a machine learning model training module, and a collaborative optimization feedback decision module, wherein: The adaptive control instruction generation module is used to calculate the fused data report using an adaptive control algorithm to obtain an initial control instruction set; The parameter adjustment instruction generation module is used to convert the initial control instruction set into an executable final parameter adjustment instruction set and generate an instruction report; The machine learning model training module is used to train the machine learning model based on historical fusion data reports and instruction reports to obtain the final machine learning model. The collaborative optimization feedback decision module is used to run a multi-objective optimization algorithm based on the final machine learning model and obtain an optimization feedback instruction report.

2. The mechanical design system based on multi-performance index collaborative control according to claim 1, characterized in that, The system also includes: a multi-source data acquisition module, a data preprocessing module, and a real-time data fusion and storage module, wherein: The multi-source data acquisition module is used to collect multi-performance indicator data in real time based on external data sources, and to perform integrity verification to obtain the original dataset. The data preprocessing module is used to preprocess the original dataset to obtain a preprocessed dataset; The real-time data fusion and storage module is used to fuse preprocessed datasets from multiple sources and perform time-series alignment to obtain a fused data report.

3. The mechanical design system based on multi-performance index collaborative control according to claim 2, characterized in that, The steps for collecting multiple performance metrics data in real time from external data sources and verifying their integrity include: S1.1: Based on an external data source, continuously receive raw data items to obtain an external dataset. Each data item includes a collection timestamp, a data source identifier, and a metric value. S1.2: Perform data validation on external data items. When the validation result is invalid, remove the data items with invalid validation results and record them to obtain the original validation dataset and error log. At the same time, store the error log in the database. When the validation result is valid, obtain the original validation dataset and proceed to step S1.

3. S1.3: Encapsulate the data items in the original validation dataset into raw data instructions, package all raw data instructions to obtain the original dataset. The structure of the raw data instructions includes the collection timestamp, data source identifier, and indicator value item. S1.4: Output the raw dataset to the data preprocessing module.

4. A mechanical design system based on multi-performance index collaborative control according to claim 3, characterized in that, The preprocessing steps based on the original dataset include: S2.1: Based on the original dataset, a moving average filter is used to smooth each index value in the original dataset to obtain a denoised dataset; S2.2: The 3σ criterion method is used to remove outliers from the index values ​​in the denoised dataset, and linear interpolation is used to supplement the removed outliers to obtain the denoised dataset. S2.3: Normalize all index values ​​in the removed dataset using a normalization formula to obtain a preprocessed dataset. The normalization process is used to normalize each index value to the range of [0,1]. S2.4: Output the preprocessed dataset to the real-time data fusion storage module.

5. A mechanical design system based on multi-performance index collaborative control according to claim 4, characterized in that, The steps for fusing preprocessed datasets from multiple sources and performing time-series alignment include: S3.1: Based on the preprocessed dataset, perform multi-source numerical fusion on each indicator value item in the preprocessed dataset to obtain the fused dataset; S3.2: Based on the collection timestamp, detect all fusion index values ​​in the fusion dataset. When there is missing data in the Cth timestamp, use linear interpolation to fill it in and obtain the fusion data report. S3.3: Store the merged data report in the database and provide an API interface to support querying historical data; S3.4: Output the fused data report to the adaptive control command generation module.

6. A mechanical design system based on multi-performance index collaborative control according to claim 5, characterized in that, The steps for calculating the fused data report using an adaptive control algorithm include: S4.1: Based on the fusion data report, all fusion index values ​​in the fusion data report are used as the system state vector, and the size adjustment and material parameter adjustment values ​​retrieved from the database are used as the control input vector; S4.2: Based on the system state vector and control input vector, and using model control algorithms, the optimal control parameters are obtained. S4.3: Convert the optimal control parameters into parameter adjustment values ​​and integrate the calculation timestamps to obtain the initial control instruction set; S4.4: Output the initial control instruction set to the parameter adjustment instruction generation module.

7. A mechanical design system based on multi-performance index collaborative control according to claim 6, characterized in that, The steps of converting the initial control instruction set into an executable final parameter tuning instruction set and generating an instruction report include: S5.1: Based on the initial control instruction set, a conversion is performed to obtain the final parameter adjustment instruction set; S5.2: Verify the final parameter adjustment instruction set. If the verification result is an allowed result, proceed to step S5.

3. If the verification result is an unallowed result, adjust the adjustment instructions that exceed the preset adjustment threshold range to the nearest boundary value, and proceed to step S5.

3. S5.3: Send the final parameter adjustment instruction set output in step S5.2 to the control system for execution. At the same time, generate an instruction report, which includes the execution timestamp, the final parameter adjustment instruction set in step S5.1, and the final parameter adjustment instruction set in step S5.

2. S5.4: Store the instruction report in the database.

8. A mechanical design system based on multi-performance index collaborative control according to claim 3, characterized in that, The steps for training a machine learning model based on historical fusion data reports and instruction reports include: S6.1: Retrieve historical fusion data reports and instruction reports from the database, and extract features to obtain feature vectors and target vectors; S6.2: Perform feature extraction based on the fused data report and instruction report to obtain the current feature vector. The current feature vector includes the fused index value and the final parameter adjustment instruction set in step S5.

2. S6.3: Retrieve the machine learning model based on the data, and train it according to the feature vector, target vector and current feature vector to obtain the first machine learning model; S6.4: Evaluate the first machine learning model based on the feature vector, target vector, and current feature vector. If the evaluation result indicates that the performance meets the standard, proceed to step S6.

5. If the evaluation result indicates that the performance does not meet the standard, return to step S6.3 to adjust the model parameters and retrain the machine learning model. S6.5: Save the first machine learning model output in step S6.4 as the final machine learning model and output it to the collaborative optimization feedback decision module.

9. A mechanical design system based on multi-performance index collaborative control according to claim 8, characterized in that, The steps for running a multi-objective optimization algorithm based on the final machine learning model include: S7.1: Load the final machine learning model; S7.2: Define the multi-objective optimization problem and obtain the optimization objective function, which includes minimizing weight, minimizing cost, and maximizing strength; S7.3: The NSGA-II fast non-dominated sorting genetic algorithm II is used to solve the optimization objective function to obtain the non-dominated solution set. Each solution in the non-dominated solution set is evaluated for the objective value through the loaded final machine learning model to obtain the objective evaluation value set. S7.4: Based on step S7.3, select the solution with the largest objective evaluation value in the non-dominated solution set as the optimal solution; S7.5: Convert the optimal solution into an optimization feedback instruction report and output it to step S4.

2. The optimization feedback instruction report includes suggested values ​​for size adjustment and material parameter adjustment.

10. A mechanical design method based on multi-performance index collaborative control, implemented according to any one of claims 1-9, characterized in that, The work includes the following steps: S1: Collect multiple performance index data in real time based on external data sources, and verify their integrity to obtain the original dataset; S2: Preprocess the original dataset to obtain the preprocessed dataset; S3: Perform multi-source data fusion on the preprocessed dataset and time-series alignment to obtain a fused data report; S4: Calculate the fused data report using an adaptive control algorithm to obtain the initial control instruction set; S5: Convert the initial control instruction set into an executable final parameter tuning instruction set and generate an instruction report; S6: Train the machine learning model based on historical fusion data reports and instruction reports to obtain the final machine learning model; S7: Run a multi-objective optimization algorithm based on the final machine learning model to obtain an optimization feedback instruction report.

Images