High-flow-speed and low-gas-consumption multi-objective optimization design method for auxiliary nozzle

Abstract

The invention relates to a high-flow-speed and low-air-consumption multi-objective optimization design method for an auxiliary nozzle, and the method comprises the steps: collecting performance data of the auxiliary nozzle of an air jet loom, and carrying out the preprocessing of the performance data; inputting the preprocessed performance data into the trained variational auto-encoder, and extracting potential features; the potential features are input into a multi-layer perceptron based on Transform, and enhanced potential features are obtained; enabling the enhanced potential features to pass through a trained XGBoost regression model, and outputting an optimization target variable; the auxiliary jet outlet airflow velocity serves as a first target value, the air consumption serves as a second target value, maximization of the first target value serves as a first target, minimization of the second target value serves as a second target, optimization design is conducted through a multi-target optimization algorithm, and a Pareto frontier solution between the auxiliary jet outlet airflow velocity and the air consumption is obtained; the air jet loom auxiliary nozzle with high-speed weft insertion and low energy consumption is designed based on a Pareto frontier solution.

Description

Technical Field

[0001] The present application relates to the technical field of multi-objective optimization design of auxiliary nozzles, and in particular to a multi-objective optimization design method for auxiliary nozzles with high flow rate and low gas consumption. Background Art

[0002] Air-jet looms, with their high speed, high efficiency, and high degree of automation, have made them the most promising shuttleless looms. However, their high energy consumption increases production costs for textile companies. Therefore, energy conservation and emission reduction have become key goals for textile companies both domestically and internationally. The weft insertion system of an air-jet loom consists of a main nozzle, auxiliary nozzles, and a special reed. There are dozens of auxiliary nozzles, accounting for over 80% of the air consumption and over 70% of the electrical energy consumed by the entire air-jet loom. The airflow velocity of the auxiliary nozzles determines the weft insertion speed and efficiency of the entire air-jet loom. Therefore, air-jet loom manufacturers both domestically and internationally are committed to optimizing the auxiliary nozzle structure, including increasing the auxiliary jet velocity and reducing nozzle air consumption, to improve the weft insertion efficiency and reduce energy consumption. Summary of the Invention

[0003] Based on this, it is necessary to provide a multi-objective optimization design method for auxiliary nozzles with high flow rate and low gas consumption, which includes: S1: collecting performance data of the auxiliary nozzle of the air jet loom and preprocessing the data; the performance data includes shape parameters and control variables; S2: Inputting the preprocessed performance data into a trained variational autoencoder to extract latent features; inputting the latent features into a Transformer-based multi-layer perceptron to obtain enhanced latent features; S3: The enhanced potential features are subjected to the trained XGBoost regression model to output the optimization target variables, wherein the optimization target variables include the auxiliary nozzle outlet airflow velocity and air consumption; S4: taking the auxiliary nozzle outlet airflow velocity as a first target value and the air consumption as a second target value, and taking maximizing the first target value as a first target and minimizing the second target value as a second target, performing optimization design using a multi-objective optimization algorithm combining a whale migration algorithm, a fast non-dominated sorting optimization, an elite reverse learning strategy, an individual adaptive mutation, and a reverse perturbation strategy, to obtain a Pareto front solution between the auxiliary nozzle outlet airflow velocity and the air consumption; S5: Design auxiliary nozzles for air jet looms based on the Pareto front solution.

[0004] Beneficial effects: This method effectively captures the potential characteristics of the auxiliary nozzle, and obtains rich Pareto front solutions through a multi-objective optimization algorithm that combines the whale migration algorithm, fast non-dominated sorting optimization, elite reverse learning strategy, individual adaptive mutation and reverse perturbation strategy. It successfully achieves efficient optimization of the auxiliary nozzle outlet air flow velocity and air consumption of the auxiliary nozzle of the air jet loom, providing a comprehensive optimization solution and scientific basis for improving the performance of industrial equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0006] Figure 1 This is a flow chart of the multi-objective optimization design method for high flow rate and low gas consumption of the auxiliary nozzle in the embodiment of the present application.

[0007] Figure 2 2 is a cross-sectional view of the auxiliary nozzle in an embodiment of the present application.

[0008] Figure 3 This is a flowchart of the multi-objective optimization algorithm in the embodiment of this application.

[0009] Figure 4 This is a comparison chart of the predicted value and the actual value of the air flow velocity of the auxiliary nozzle in the embodiment of the present application.

[0010] Figure 5 This is a comparison chart of the predicted value and the actual value of the gas consumption in the embodiment of this application.

[0011] Figure 6 Schematic diagram of the Pareto front solution obtained in the embodiment of the present application. DETAILED DESCRIPTION

[0012] To make the above-mentioned objects, features, and advantages of the present application more clearly understood, the specific embodiments of the present application are described in detail below with reference to the accompanying drawings. The following description sets forth many specific details to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar improvements without violating the scope of the present application. Therefore, the present application is not limited to the specific embodiments disclosed below.

[0013] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of such features. Throughout the description of this application, "plurality" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0014] like Figure 1 As shown, this embodiment provides a multi-objective optimization design method for an auxiliary nozzle with high flow rate and low gas consumption, the method comprising: S1: collecting performance data of the auxiliary nozzle of the air jet loom and preprocessing the data; the performance data includes shape parameters and control variables.

[0015] In this embodiment, the auxiliary nozzle is structured as follows Figure 2 As shown, the shape parameters include the inlet size D1 of the auxiliary nozzle, the pipe size D2, the outlet large hole diameter d, the number of small holes n, the diameter G between the centers of the two small holes and the outlet taper α The control variables are the auxiliary nozzle airflow velocity and air consumption. The auxiliary nozzle airflow velocity determines the auxiliary nozzle's weft insertion efficiency, while air consumption determines the energy consumption of the entire air jet loom. In this embodiment, the number of small holes n = [2, 3, 4, 5, 6]. In practical applications, this design method can also be applied to the multi-objective optimization design of multiple auxiliary nozzles with a larger number of small holes.

[0016] In this embodiment, the preprocessing includes: Performing outlier detection on the performance data and eliminating abnormal performance data; Normalize normal performance data.

[0017] S2: Input the preprocessed performance data into a trained variational autoencoder to extract potential features; input the potential features into a Transformer-based multi-layer perceptron to obtain enhanced potential features.

[0018] Specifically, the variational autoencoder includes a first input layer (a fully connected layer in this embodiment), an encoder, and a decoder. The encoder includes a first fully connected layer and a second fully connected layer. The decoder is used to reconstruct the latent features output by the encoder and output reconstructed values ​​of the performance data. In the trained variational autoencoder, the first input layer maps the performance data to a latent space, the first fully connected layer outputs the mean of the latent space, and the second fully connected layer outputs the logarithmic variance of the latent space; reparameterization is performed based on the mean and logarithmic variance of the latent space to obtain the latent features of the performance data.

[0019] Furthermore, the training process of the variational autoencoder is: with the goal of minimizing the reconstruction error and KL divergence, and using the Bayesian optimization algorithm to tune the hyperparameters of the variational autoencoder, and select the optimal first hyperparameter combination of the variational autoencoder, the first hyperparameter including the dimension of the latent space, the batch size, and the first learning rate.

[0020] By introducing the idea of ​​generative model, the variational autoencoder (VAE) can effectively handle the nonlinear relationship of data and has good feature learning capabilities, making it particularly suitable for complex and high-dimensional data.

[0021] In this embodiment, the Transformer-based multilayer perceptron includes: a second input layer (a fully connected layer in this embodiment), a Transformer module, and an output layer; the potential features are input into the multilayer perceptron via the second input layer, the Transformer module enhances the input potential features based on the self-attention mechanism, and the output layer maps the output of the Transformer module to the dimensions of each optimization target variable to obtain the enhanced potential features.

[0022] As a deep neural network, the multilayer perceptron can automatically learn the complex nonlinear relationships between input features, giving the model stronger expressive power. At the same time, through the self-attention mechanism, the Transformer can help the model capture more important information and complex features hidden in the data, enhance feature learning capabilities, improve the model's ability to capture long-range dependencies and global features, and enhance the ability to model complex data patterns, thereby further improving the model's fitting ability and prediction accuracy.

[0023] S3: The enhanced potential features are subjected to the trained XGBoost regression model to output the optimization target variables, where the optimization target variables include the auxiliary nozzle outlet air flow velocity and air consumption of the auxiliary nozzle.

[0024] Specifically, the process of obtaining the optimized target variable includes: Merging the enhanced latent features with the original target variable to obtain a feature set; The feature set is input into the trained XGBoost regression model, and the optimized target variable is output.

[0025] In this embodiment, the training process of the XGBoost regression model includes: Dividing the feature set into a training set and a test set; The XGBoost regression model is trained based on the training set to output a predicted value; with the goal of minimizing the root mean square error between the predicted value and the true value, the XGBoost regression model is hyperparameter tuned using a Bayesian optimization algorithm to select the optimal second hyperparameter combination of the XGBoost regression model, where the second hyperparameter includes the depth of the tree, the second learning rate, and the regularization strength; The trained XGBoost regression model is evaluated based on the test set.

[0026] The XGBoost regression model improves the generalization ability of the model through integrated learning methods, can handle high-dimensional features, and is more robust to outliers and noisy data.

[0027] S4: Taking the airflow velocity of the auxiliary nozzle outlet as the first target value, and the air consumption as the second target value, and taking maximizing the first target value as the first target, and minimizing the second target value as the second target, a multi-objective optimization algorithm combining whale migration algorithm, fast non-dominated sorting optimization, elite reverse learning strategy, individual adaptive mutation and reverse perturbation strategy is used for optimization design to obtain the Pareto front solution between the airflow velocity of the auxiliary nozzle outlet and the air consumption.

[0028] Specifically, such as Figure 3 As shown in Figure 2, the process of optimizing design using a multi-objective optimization algorithm includes: Step 1: Generate a population using an elite reverse learning strategy. Each individual in the population consists of different enhanced latent features, first target values, and second target values. Each individual is assigned a randomly selected index, and the first target and second target corresponding to each individual are calculated using a proxy model to obtain a solution for each individual. The solution includes a solution for the first target and a solution for the second target. In this embodiment, the position information of each individual in the whale migration algorithm is initialized randomly. Since random initialization will increase the range of feasible solutions, prolong the search time, and fail to obtain the optimal solution position, an elite reverse learning strategy is adopted to generate the population, which specifically includes: Generate a population randomly, generate reverse individuals for each individual in the random population, and form a reverse group. The calculation formula is: ; in, represents the i-th reverse individual in the random population; Indicates a random number in the range (0,1); L indicates the maximum range; U indicates the minimum range; represents the i-th individual in a random population; Calculate the fitness value of each individual in the random population, and form the elite group with the first N / 2 individuals with the smallest fitness value; compare the fitness values ​​of the first elite individual in the elite group and its corresponding reverse individual, and retain the better one as the first elite individual of the new generation. The calculation formula is: ; in, represents the first elite individual of the e-th new generation; represents the e-th first elite individual; represents the reverse individual of the e-th first elite individual; Traverse all the first elite individuals in the elite group to obtain a new elite group; Merge the first N / 2 reverse individuals with the smallest fitness value in the reverse group with the N / 2 first elite individuals in the new elite group to obtain a population P (with N individuals).

[0029] Step 2: Perform the first non-dominated sorting based on the solutions corresponding to each individual, and assign a rank value to each individual, which represents the level of the frontier to which the individual belongs; Step 3: For any frontier layer, calculate the crowding degree of each individual; include: For all solutions corresponding to individuals in any layer of the frontier, sort them by the first target value and the second target value respectively; For the first goal / second goal, the crowding degree of the first and last individuals sorted under this goal is set to infinity; For each non-boundary individual in the frontier, the calculation formula for its crowding component on the target m is: ; in, represents the crowding component of the i-th individual on the target m; Represents the target value of the adjacent individual i+1 of the i-th individual on the target m after sorting; Represents the target value of the adjacent individual i-1 of the i-th individual on the target m after sorting; Represents the maximum target value of all individuals on target m; Represents the minimum target value of all individuals on the target m; The normalized crowding components calculated for each target are accumulated to obtain the individual crowding, which is calculated as follows: ; in, represents the crowding degree of the i-th individual; M is the number of objectives; in multi-objective optimization, the larger the crowding distance, the greater the sparseness of the individuals in the frontier, and the better solutions are retained in a limited manner; Step 4: Select the individual with the highest crowding degree in the first layer as the leader, introduce a gradually decreasing random perturbation to the leader's position; guide other individuals to move towards the position of the randomly perturbed leader to update the positions of other individuals and obtain the second population; Step 5: Merge the population with the second population to obtain a new population; Step 6: Perform a second non-dominated sort based on the solutions corresponding to each individual in the new population, and assign a second rank value to each individual, where the second rank value indicates the level of the second frontier to which the individual belongs. Step 7: For any layer of the second frontier, calculate the second crowding degree of each individual in it; select the individual with the largest second crowding degree in the first layer of the second frontier as the new leader; Step 8: Perform adaptive mutation operations on individuals in the new population to generate new individuals and enhance the algorithm's ability to jump out of the local space; Specifically, each individual in the population has a certain probability of undergoing mutation operation: ; ; in, represents the i-th new individual; Represents the average position of the current NL leaders in the whale group; Indicates the location of the leader of the whale group found; Represents the adaptive probability variation factor, and its value increases as the number of iteration steps increases; Indicates the current iteration step; Indicates the maximum number of iterations; represents the Hadamard product; It means generating a matrix with 1 row and D columns, where the value of each element is a random number uniformly distributed in the interval [0,1), and D represents the dimension of the population; represents the i-th individual; represents the i-1th individual.

[0030] The optimal individual in the current iteration guides the position updates of other individuals, and all other whale individuals follow the leader in position updates. The mutation operation randomly selects position features of the whale leader and changes their values. After the mutation, the position changes of the leader and followers in the whale population will generate new solutions. The adaptive mutation operation randomly selects certain characteristic position features of the whale leader and changes their values. After the mutation, the position changes of the leader and followers in the whale population will generate new solutions.

[0031] Step 9: After the perturbation strategy mentioned above is used to enhance the algorithm’s ability to jump out of the local space and perform perturbation mutation to update the position, a greedy strategy is introduced to compare the fitness values ​​of individuals before and after the mutation and update the leader. The update formula is: ; in, Indicates the updated leader's position; represents the individual before mutation at the tth iteration; represents the individual after mutation at the tth iteration; represents the fitness value; Step 10: Before each generation update, the top k individuals with the highest level values ​​are taken as elite individuals. Each elite individual is subjected to a reverse perturbation strategy, and the corresponding solution or perturbation solution is output. The solution or perturbation solution is directly passed to the next generation to improve the algorithm accuracy and convergence speed. Step 11: Iterate steps 2-10 until a preset maximum number of iterations is reached to obtain a Pareto front solution between the air flow velocity at the auxiliary nozzle outlet and the air consumption.

[0032] Furthermore, non-dominated sorting includes: Determine the dominance relationship between individual A and individual B: If the solutions of the two objective functions corresponding to individual A are both better than those of individual B, then individual A dominates individual B; Otherwise, individual B dominates individual A; When the solutions of the two objective functions corresponding to individual A are better than the solutions of the two objective functions corresponding to all individuals, individual A is a non-dominated individual; Set up multiple frontiers, the first frontier contains all non-dominated individuals, and the subsequent frontiers contain individuals dominated by different numbers of individuals; According to the number of individuals dominated in each frontier, multiple frontiers are divided into different levels; The individuals in each front are assigned the level of the front as the rank value of the corresponding individual; the lower the rank value of the individual, the higher the quality of the corresponding solution.

[0033] Furthermore, the elite individuals are subjected to reverse perturbation strategies including: Step 1: Subtract the solution corresponding to each elite individual from the sum of the upper and lower bounds of the solution dimension to obtain the preliminary reverse solution corresponding to each elite individual; Step 2: Calculate and compare the fitness of the preliminary reverse solution and the solution. When the fitness of the preliminary reverse solution is less than the fitness of the solution, the solution is still adopted; otherwise, the preliminary reverse solution is used as the reverse solution of the corresponding solution. Step 3: Calculate the perturbation solution based on the solution and its inverse solution. The calculation formula is: ; ; in, represents the perturbation solution corresponding to the jth solution of the i-th elite individual; represents the jth solution of the i-th elite individual; represents the reverse solution corresponding to the j-th solution of the i-th elite individual; represents a random disturbance term, which generally conforms to the normal distribution; represents the standard deviation; represents the disturbance intensity factor, , controls the ratio of approaching the reverse solution. When it is equal to 1, it jumps completely to the reverse solution. When it is equal to 0, it keeps the original solution. In this embodiment, it is necessary to ensure that the perturbation solution is still within the upper and lower bounds of the solution, which can be expressed as: ; in, represents the perturbation solution corresponding to the jth solution of the i-th elite individual after constraint; represents the upper bound of the solution dimension; represents the lower bound of the solution dimension; Step 4: Determine whether the perturbation solution is better than the corresponding solution. If the perturbation solution is better than the corresponding solution, replace the corresponding solution with the perturbation solution; otherwise, the solution is still used. Step 5: Iterate steps 1-4 until the preset maximum number of iterations is reached.

[0034] S5: Based on the Pareto front solution, an auxiliary nozzle for a high-speed weft insertion and low energy consumption air jet loom is designed.

[0035] In order to reflect the performance advantages of the multi-objective optimization design method for the auxiliary nozzle with high flow rate and low gas consumption provided in this embodiment, the mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE) and R 2 The value was used as the first evaluation index, and a comparative experiment was conducted. The experimental results are shown in Table 1; Table 1 is a comparison table of the results of the first evaluation index; Model MAE MSE RMSE <![CDATA[R 2 ]]> VAE-XGB 3.1619 21.2751 4.6125 0.8389 VAE-XGB-OPT 0.7281 1.3031 1.1415 0.9791 VAE-MLP-XGB-OPT 0.3901 0.5472 0.7397 0.9932 VAE-Transformer-XGB-OPT 0.3321 0.4504 0.6711 0.9939 In Table 1, the VAE-XGB model includes a variational autoencoder and an XGBoost regression model; the VAE-XGB-OPT model includes a Bayesian-optimized variational autoencoder and an XGBoost regression model; the VAE-MLP-XGB-OPT model includes a conventional multi-layer perceptron, a Bayesian-optimized variational autoencoder, and an XGBoost regression model; the VAE-Transformer-XGB-OPT model is the model provided in this embodiment, which includes a Transformer-based multi-layer perceptron (TR-MLP), a Bayesian-optimized (OPT) variational autoencoder (VAE), and an XGBoost regression model (XGB).

[0036] The performance of the four models in Table 1 shows a significant optimization trend. The MAE, MSE, and RMSE indicators gradually decrease, and R 2 The value gradually increases, which shows that the improvement from the VAE-XGB model to the model provided in this embodiment has continuously improved the prediction ability of the model; by comparison, the error of the VAE-Transformer-XGB-OPT model is the lowest, indicating that the model after multi-layer optimization performs best among the four models with extremely high accuracy. 2 The value is 0.9939, indicating that the model performs very well in data fitting and can explain 99.39% of the variability in the data. The combination of VAE, TR-MLP, and XGB enables the model to achieve a good balance in feature learning, nonlinear relationship modeling, and efficient tree model training, thereby improving overall performance.

[0037] VAE can effectively encode input data and extract key features, while TR-MLP and XGB further improve the accuracy of the model through hierarchical learning and ensemble learning, enabling the model to better cope with the complexity and noise of data, have strong generalization capabilities, and avoid the risk of overfitting.

[0038] As a tree-based ensemble method, XGB is highly robust to outliers and noise in row data and can maintain high prediction accuracy. This combination of deep learning and ensemble learning can effectively avoid the shortcomings of a single model and enhance the model's adaptability to different data sets.

[0039] The VAE-Transformer-XGB-OPT model structure is highly flexible and can adjust the parameters of VAE, TR-MLP, and XGB according to different data characteristics and task requirements to optimize model performance. It is particularly suitable for processing various types of data (such as numerical, categorical, and high-dimensional data), enabling it to achieve good performance in various scenarios.

[0040] The VAE-Transformer-XGB-OPT model combines the advantages of VAE, MLP, and XGB to perform very well in accuracy, robustness, and generalization. It can handle complex and high-dimensional data, has strong adaptability, and is particularly suitable for tasks that require high prediction accuracy. In addition, the VAE-Transformer-XGB-OPT model is the best performing of the four models, with the lowest prediction error and the highest R² value, making it suitable for tasks that require high precision. Since industrial machines often require higher machine precision, the model's prediction requirements are also higher. The comparison chart of the auxiliary nozzle outlet airflow velocity and air consumption predicted by the VAE-Transformer-XGB-OPT model and the actual values ​​is shown below. Figure 4 、 Figure 5 As shown in the figure, it can be seen that the predicted value of the model is close to the true value. Therefore, the model performs extremely well in terms of fitting degree and accuracy. It can be used for prediction in industrial production and can provide good prediction data for nozzles, saving costs.

[0041] At the same time, the multi-objective optimization algorithm that combines the whale migration algorithm, fast non-dominated sorting optimization, elite reverse learning strategy, individual adaptive mutation and reverse perturbation strategy has the following advantages over the simple fast elite multi-objective genetic algorithm (NSGA-II): 1. Comparative experiments with other optimization algorithms demonstrate that the optimization framework of this embodiment exhibits significant advantages in predicting and optimizing the performance of air jet loom nozzles, effectively balancing the multi-objective optimization problem of speed and air consumption. The experimental results also demonstrate that combining a trained surrogate model with the optimization algorithm significantly improves the accuracy of performance predictions and generates more diverse Pareto front solutions, providing reliable technical support for industrial process optimization.

[0042] 2. The multi-objective optimization algorithm can stably find 16 Pareto front solutions in each run (while the NSGA-II algorithm has approximately 8 Pareto front solutions). The consistency of the results shows that the search process of the multi-objective optimization algorithm combined with whale migration provided in this embodiment is more stable and has the ability to overcome the influence of random disturbances. In order to reflect the performance advantages of the multi-objective optimization algorithm used in the high flow rate and low gas consumption optimization design of the auxiliary nozzle provided in this embodiment, Inverse Generational Distance (IGD), Generational Distance (GD), Hypervolume (HV) and Spacing value are used as evaluation indicators. The comparison results of the multi-objective optimization algorithm and the NSGA-II algorithm are shown in Table 2. Table 2 is a comparison table of the results of the second evaluation index; Evaluation indicators MOWMA NSGA-II IGD(Obtained_Pareto,True_Pareto) 0.0478 0.0482 GD(Obtained_Pareto,True_Pareto) 0.0003 0.0005 HV(Obtained_Pareto,True_Pareto) 0.1831 0.1831 Spacing(Obtained_Pareto) 0.0682 0.0696 IGD is a metric that measures the distance between the set of non-dominated solutions generated by the algorithm and the true Pareto front. It calculates the sum of the distances from each point on the true Pareto front to the nearest non-dominated solution and takes the average. A smaller IGD value indicates that the solution set generated by the algorithm is closer to the true Pareto front.

[0043] GD is a variant of IGD that calculates the sum of the distances from each non-dominated solution to the nearest point on the true Pareto front and takes the average. The smaller the GD value, the better the solution set generated by the algorithm is distributed on the Pareto front.

[0044] The HV metric measures the area covered by the set of non-dominated solutions generated by the algorithm. Typically, this area is defined between the minimum and maximum values ​​of the objective function. A larger HV value indicates that the solution set generated by the algorithm covers a wider range of the objective function space.

[0045] Spacing is a metric that measures the average distance between solutions in the non-dominated solution set generated by the algorithm. A smaller Spacing value indicates a denser solution set and higher diversity.

[0046] The results are stable at a relatively high number of solutions, showing the robustness of the multi-objective optimization algorithm in the optimization problem.

[0047] 3. In the data test, the Pareto front solution found by the multi-objective optimization algorithm covers a wider range of auxiliary nozzle outlet air flow velocity and air consumption combinations, and the solutions are evenly distributed on different objectives, such as Figure 6 This diversity stems from the perturbations introduced by simulated annealing, which enables the algorithm to jump out of local areas during the optimization process, ensuring that the solution set is not only larger in number but also more evenly distributed.

[0048] 4. The multi-objective optimization algorithm uses a reverse perturbation strategy to jump out of the local optimum through reverse solution and perturbation solution when the objective function is embedded in the local optimum. Data experiments show that the multi-objective optimization algorithm can more effectively find a balance between maximizing speed and minimizing gas consumption.

[0049] 5. Across multiple runs, the multi-objective optimization algorithm demonstrated a significant advantage in convergence speed, despite incorporating a reverse perturbation strategy, which slightly increases computational overhead. Compared to NSGA-II, the multi-objective optimization algorithm found a richer Pareto front of solutions in fewer iterations, reducing overall computational time.

[0050] 6. Multi-objective optimization algorithms can adjust themselves more flexibly to explore different solution areas to cope with the diversity of data.

[0051] The multi-objective optimization design method for high flow rate and low air consumption of auxiliary nozzles provided in this embodiment has been compared with other optimization algorithms. The results show that this method exhibits significant advantages in the performance prediction and optimization of air jet loom nozzles, effectively balancing the multi-objective optimization problem of speed and air consumption. The experimental results also show that combining a trained surrogate model with an optimization algorithm can significantly improve the accuracy of performance prediction and obtain more diverse Pareto front solutions, providing reliable technical support for industrial process optimization.

[0052] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0053] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A multi-objective optimization design method for auxiliary nozzles with high flow rate and low gas consumption, characterized in that: include: S1: collecting performance data of the auxiliary nozzle of the air jet loom and preprocessing the data; the performance data includes shape parameters and control variables; S2: Input the preprocessed performance data into the trained variational autoencoder to extract potential features; Inputting the latent features into a Transformer-based multi-layer perceptron to obtain enhanced latent features; S3: The enhanced potential features are subjected to the trained XGBoost regression model to output the optimization target variables, wherein the optimization target variables include the auxiliary nozzle outlet airflow velocity and air consumption; S4: taking the auxiliary nozzle outlet airflow velocity as a first target value and the air consumption as a second target value, and taking maximizing the first target value as a first target and minimizing the second target value as a second target, performing optimization design using a multi-objective optimization algorithm combining a whale migration algorithm, a fast non-dominated sorting optimization, an elite reverse learning strategy, an individual adaptive mutation, and a reverse perturbation strategy, to obtain a Pareto front solution between the auxiliary nozzle outlet airflow velocity and the air consumption; S5: Design auxiliary nozzles for air jet looms based on the Pareto front solution.

2. The multi-objective optimization design method for auxiliary nozzles with high flow rate and low gas consumption according to claim 1 is characterized in that: The pretreatment includes: Performing outlier detection on the performance data and eliminating abnormal performance data; Normalize normal performance data.

3. The multi-objective optimization design method for auxiliary nozzles with high flow rate and low gas consumption according to claim 1 is characterized in that: The variational autoencoder includes a first input layer and an encoder, and the encoder includes a first fully connected layer and a second fully connected layer; In the trained variational autoencoder, the first input layer maps the performance data to a latent space, the first fully connected layer outputs a mean of the latent space, and the second fully connected layer outputs a logarithmic variance of the latent space; Reparameterization is performed based on the mean and the log variance of the latent space to obtain latent features of the performance data.

4. The multi-objective optimization design method for auxiliary nozzles with high flow rate and low gas consumption according to claim 2, characterized in that: The variational autoencoder further includes a decoder, which is used to reconstruct the potential features output by the encoder and output a reconstructed value of performance data; The training process of the variational autoencoder is as follows: with the goal of minimizing the reconstruction error and the KL divergence, the Bayesian optimization algorithm is used to tune the hyperparameters of the variational autoencoder, and the optimal first hyperparameter combination of the variational autoencoder is selected. The first hyperparameter includes the dimension of the latent space, the batch size, and the first learning rate.

5. The multi-objective optimization design method for auxiliary nozzles with high flow rate and low gas consumption according to claim 1 is characterized in that: The Transformer-based multilayer perceptron includes: a second input layer, a Transformer module, and an output layer; the potential features are input into the multilayer perceptron via the second input layer, the Transformer module enhances the input potential features based on a self-attention mechanism, and the output layer maps the output of the Transformer module to the dimensions of each optimization target variable to obtain the enhanced potential features.

6. The multi-objective optimization design method for auxiliary nozzles with high flow rate and low gas consumption according to claim 1, characterized in that: The process of obtaining the optimized target variable includes: Merging the enhanced latent features with the original target variable to obtain a feature set; The feature set is input into the trained XGBoost regression model, and the optimized target variable is output.

7. The multi-objective optimization design method for auxiliary nozzles with high flow rate and low gas consumption according to claim 6, characterized in that: The training process of the XGBoost regression model includes: Dividing the feature set into a training set and a test set; The XGBoost regression model is trained based on the training set to output a predicted value; with the goal of minimizing the root mean square error between the predicted value and the true value, the XGBoost regression model is hyperparameter tuned using a Bayesian optimization algorithm to select the optimal second hyperparameter combination of the XGBoost regression model, where the second hyperparameter includes the depth of the tree, the second learning rate, and the regularization strength; The trained XGBoost regression model is evaluated based on the test set.

8. The multi-objective optimization design method for auxiliary nozzles with high flow rate and low gas consumption according to claim 1, characterized in that: The process of optimizing design using a multi-objective optimization algorithm includes: Step 1: Generate a population using an elite reverse learning strategy. Each individual in the population consists of different enhanced latent features, first target values, and second target values. Each individual is assigned a randomly selected index, and the first target and second target corresponding to each individual are calculated using a proxy model to obtain a solution for each individual. The solution includes a solution for the first target and a solution for the second target. Step 2: Perform non-dominated sorting based on the solutions corresponding to each individual, and assign a rank value to each individual, which represents the level of the frontier to which the individual belongs; Step 3: For any frontier layer, calculate the crowding degree of each individual; Step 4: Select the individual with the highest crowding degree in the first layer as the leader, introduce a gradually decreasing random perturbation to the leader's position; guide other individuals to move towards the position of the randomly perturbed leader to update the positions of other individuals; Step 5: Merge the population with the second population to obtain a new population; Step 6: Perform a second non-dominated sort based on the solutions corresponding to each individual in the new population, and assign a second rank value to each individual, where the second rank value indicates the level of the second frontier to which the individual belongs. Step 7: For any layer of the second frontier, calculate the second crowding degree of each individual in it; select the individual with the largest second crowding degree in the first layer of the second frontier as the new leader; Step 8: Perform adaptive mutation operations on individuals in the population to generate new individuals; Step 9: Compare the fitness of individuals before and after mutation and update the leader; Step 10: Before each generation update, the top k individuals with the highest level values ​​are taken as elite individuals. Each elite individual is subjected to a reverse perturbation strategy, and the corresponding solution or perturbation solution is output. The solution or perturbation solution is directly passed to the next generation to improve the algorithm accuracy and convergence speed. Step 11: Iterate steps 2-10 until a preset maximum number of iterations is reached to obtain a Pareto front solution between the air flow velocity at the auxiliary nozzle outlet and the air consumption.

9. The multi-objective optimization design method for auxiliary nozzles with high flow rate and low gas consumption according to claim 8, characterized in that: Non-dominated sorting includes: Determine the dominance relationship between individual A and individual B: If the solutions of the two objective functions corresponding to individual A are both better than those of individual B, then individual A dominates individual B; Otherwise, individual B dominates individual A; When the solutions of the two objective functions corresponding to individual A are better than the solutions of the two objective functions corresponding to all individuals, individual A is a non-dominated individual; Set up multiple frontiers, the first frontier contains all non-dominated individuals, and the subsequent frontiers contain individuals dominated by different numbers of individuals; According to the number of individuals dominated in each frontier, multiple frontiers are divided into different levels; The individuals in each front are assigned the level of the front as the rank value of the corresponding individual; the lower the rank value of the individual, the higher the quality of the corresponding solution.

10. The multi-objective optimization design method for auxiliary nozzles with high flow rate and low gas consumption according to claim 8, characterized in that: The reverse perturbation strategy for elite individuals includes: Step 1: Subtract the solution corresponding to each elite individual from the sum of the upper and lower bounds of the solution dimension to obtain the preliminary reverse solution corresponding to each elite individual; Step 2: Calculate and compare the fitness of the preliminary reverse solution and the solution. When the fitness of the preliminary reverse solution is less than the fitness of the solution, the solution is still adopted; otherwise, the preliminary reverse solution is used as the reverse solution of the corresponding solution. Step 3: Calculate the perturbation solution based on the solution and its inverse solution. The calculation formula is: ； in, represents the perturbation solution corresponding to the jth solution of the i-th elite individual; represents the jth solution of the i-th elite individual; represents the reverse solution corresponding to the j-th solution of the i-th elite individual; represents a random disturbance term, which generally conforms to the normal distribution; represents the disturbance intensity factor; Step 4: Determine whether the perturbation solution is better than the corresponding solution. If the perturbation solution is better than the corresponding solution, replace the corresponding solution with the perturbation solution; otherwise, the solution is still used. Step 5: Iterate steps 1-4 until the preset maximum number of iterations is reached.

Images