Method, device, equipment and medium for predicting auxiliary jet air consumption of air jet loom

Abstract

The present application discloses a method, device, equipment and medium for predicting the auxiliary jet air consumption of an air jet loom, and belongs to the field of industrial automation technology. The method comprises: obtaining the structural parameters to be measured of the nozzle to be measured in the air jet loom, the structural parameters to be measured include the inlet diameter, the straight tube diameter, the nozzle taper, the elliptical outlet major axis and the elliptical outlet minor axis; inputting the structural parameters to be measured into the air flow prediction model, and obtaining the nozzle air consumption of the nozzle to be measured output by the air flow prediction model, and the air flow prediction model is obtained based on the sample structural parameters and air consumption label training of the sample nozzle. The method can accurately predict the air consumption of the nozzle in a complex industrial environment.

Description

Technical Field

[0001] The present application belongs to the field of industrial automation technology, and in particular relates to a method, device, equipment and medium for predicting auxiliary jet air consumption of an air jet loom. Background Art

[0002] In the modern textile industry, air jet looms are widely used due to their efficient weaving capabilities. However, their energy consumption has always been a focus of attention in the industry. The nozzle, as a key component in the air jet loom, is used to effectively guide the weft yarn through the shed formed by the warp yarn. The air consumption of the nozzle is an important parameter to measure the performance and efficiency of the air jet loom, and its size directly affects the energy consumption of the air jet loom.

[0003] Currently, most existing gas consumption prediction methods are based on empirical formulas or traditional numerical simulation techniques.

[0004] However, the above methods are difficult to accurately predict in complex industrial environments. Summary of the invention

[0005] The present application aims to solve at least one of the technical problems existing in the prior art. To this end, the present application proposes a method, device, equipment and medium for predicting the air consumption of the auxiliary jet of an air jet loom, which can accurately predict the air consumption of the nozzle in a complex industrial environment.

[0006] In a first aspect, the present application provides a method for predicting auxiliary jet air consumption of an air jet loom, the method comprising:

[0007] Acquire the structural parameters to be measured of the nozzle to be measured in the air jet loom, wherein the structural parameters to be measured include the inlet diameter, the straight tube diameter, the nozzle taper, the elliptical outlet major axis and the elliptical outlet minor axis;

[0008] The structural parameters to be tested are input into an airflow prediction model to obtain the nozzle air consumption of the nozzle to be tested output by the airflow prediction model, wherein the airflow prediction model is trained based on sample structural parameters and air consumption labels of sample nozzles.

[0009] According to one embodiment of the present application, the airflow prediction model includes an input layer, a hidden layer and an output layer, and the hidden layer includes a convolution layer, a pooling layer and a fully connected layer;

[0010] The step of inputting the structural parameters to be measured into an air flow prediction model to obtain the nozzle air consumption of the nozzle to be measured output by the air flow prediction model comprises:

[0011] Receiving the structural parameters to be measured through the input layer;

[0012] Extracting features of the structural parameters to be measured through the convolution layer to obtain a first morphological feature of the nozzle to be measured;

[0013] Performing data compression on the first morphological feature through the pooling layer, and performing feature extraction on the compressed first morphological feature through the convolution layer to obtain a second morphological feature of the nozzle to be tested;

[0014] The first morphological feature and the second morphological feature are fused and mapped to the nozzle flow feature through the fully connected layer to generate the nozzle air consumption;

[0015] The nozzle gas consumption is output via the output layer.

[0016] According to one embodiment of the present application, the airflow prediction model is obtained based on the following steps:

[0017] Constructing a convolutional neural network model based on the input layer, the hidden layer and the output layer;

[0018] Optimizing the learning rate of the convolutional neural network model, as well as the convolution kernel size and the number of convolutions of the convolutional layer;

[0019] The optimized convolutional neural network model is trained to obtain the airflow prediction model.

[0020] According to one embodiment of the present application, the optimizing the learning rate of the convolutional neural network model, and the convolution kernel size and the number of convolutions of the convolution layer includes:

[0021] Determine and optimize the fitness function of the convolutional neural network model, and obtain a first fitness value corresponding to a first search agent position of each individual in the population, wherein the individual is characterized as a hyperparameter group to be optimized of the convolutional neural network model;

[0022] Training each of the individuals, calculating a second search agent position of each of the individuals, and grouping the population based on the first fitness value;

[0023] Performing group training on each of the grouped individuals, and updating the population, so as to obtain a second fitness value of each of the individuals based on the second search agent position and the fitness function;

[0024] Based on the second fitness value, determining a first target individual in the population;

[0025] Training the first target individual to update the population and obtain an optimal search agent position of the first target individual;

[0026] Repeat to obtain multiple optimal search agent positions until the maximum number of iterations is reached to obtain a second target individual, and optimize the learning rate, the convolution kernel size and the number of convolutions of the convolutional neural network model through the target hyperparameter group corresponding to the second target individual.

[0027] According to an embodiment of the present application, the calculating the second search agent position of each individual includes:

[0028]

[0029] in, k is the number of iterations of the current training, For the i The individual in dimension j On k The second search agent position after iterations, For the i The individual in dimension j On k The first search agent position before the iteration, For the k Iterations in dimension j Best search proxy locations on; for A random number in the range, for A random number in the range, All are golden ratio coefficients.

[0030] According to one embodiment of the present application, performing group training on each of the grouped individuals and updating the population includes:

[0031] determining a learning probability, a communication probability, and an error probability for the group training;

[0032] Based on the learning probability, the communication probability and the error probability, group training is performed on each individual, and the training state of each individual is determined, so as to update the population according to the training state.

[0033] According to an embodiment of the present application, the training of the first target individual to update the population and obtain the best search agent position of the first target individual includes:

[0034]

[0035]

[0036] in, is an adaptive parameter used to adjust tThe scale of the random numbers generated by the distribution is used to control the search step size of each individual in each iteration; m is a positive integer, k is the number of iterations of the current training, is the size of the population, t () is generated t Function for distributing random numbers, is the search agent position of the first target individual in the initialization phase, The first target individual is k The best search agent position before iterations; The first target individual is k The best search agent position after iterations.

[0037] In a second aspect, the present application provides a device for predicting auxiliary jet air consumption of an air jet loom, the device comprising:

[0038] An acquisition module is used to acquire the structural parameters to be measured of the nozzle to be measured in the air jet loom, wherein the structural parameters to be measured include the inlet diameter, the straight tube diameter, the nozzle taper, the elliptical outlet major axis and the elliptical outlet minor axis;

[0039] A processing module is used to input the structural parameters to be tested into an air flow prediction model to obtain the nozzle air consumption of the nozzle to be tested output by the air flow prediction model, wherein the air flow prediction model is trained based on sample structural parameters and air consumption labels of sample nozzles.

[0040] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for predicting the auxiliary jet air consumption of a jet loom as described in the first aspect above is implemented.

[0041] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for predicting the auxiliary jet air consumption of an air jet loom as described in the first aspect above.

[0042] In a fifth aspect, the present application provides a chip, comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run a program or instruction to implement the method for predicting the auxiliary jet air consumption of a jet loom as described in the first aspect.

[0043] In a sixth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the method for predicting the auxiliary jet air consumption of an air jet loom as described in the first aspect above.

[0044] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application.

[0045] The method, device, equipment and medium for predicting the auxiliary jet air consumption of an air jet loom provided by the present invention have the following beneficial effects compared with the prior art:

[0046] (1) The air flow prediction model is used to predict the structural parameters that affect the air consumption, and the nozzle air consumption of the nozzle to be tested is obtained, which improves the prediction accuracy and efficiency. It is more suitable for complex industrial scenarios such as textiles. It can accurately predict the nozzle air consumption in complex industrial environments and provide data reference for the energy consumption management of air jet looms.

[0047] (2) Through the convolution layer, feature extraction is performed to automatically learn features and reduce the number of parameters. Through the pooling layer, data compression is performed to reduce the amount of calculation and provide translation invariance. Through the feature fusion of the fully connected layer, features can be integrated to perform the final regression prediction. The airflow prediction model effectively extracts features from the input structural parameters to be tested and performs accurate and fast predictions.

[0048] (3) By optimizing the learning rate, convolution kernel size, and number of convolutions of the convolutional neural network model, the difficulty of parameter adjustment is reduced, the training process is simpler, and the prediction accuracy, training efficiency, and generalization ability of the air flow prediction model are improved, so that the best prediction effect can be achieved under different working conditions;

[0049] (4) In the collective training stage, the population is updated through the golden ratio coefficient, so that the search agent can adjust its search path through the golden ratio and sinusoidal fluctuations. In the individual additional training stage, the adaptive t distribution is introduced to adjust the position, so as to better balance the global and local search capabilities in complex optimization problems and accelerate the optimization process. The CNN model parameters are automatically optimized through IFTTA, which improves the prediction accuracy and adaptability of the CNN model. The problems of low prediction accuracy, difficult tuning of model parameters, and long optimization iteration time in the existing technology are overcome, thereby achieving high-precision prediction of nozzle air consumption. The air flow prediction model can be integrated into the overall control system of the jet loom to accurately control the output flow of the air compressor and reduce air and power consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0051] Figure 1 It is a flow chart of a method for predicting auxiliary jet air consumption of an air jet loom provided in an embodiment of the present application;

[0052] Figure 2is a schematic structural diagram of a longitudinal section of an auxiliary nozzle of an air jet loom provided in an embodiment of the present application;

[0053] Figure 3 It is a structural schematic diagram of the airflow prediction model provided in the embodiment of the present application;

[0054] Figure 4 This is one of the flow charts of the improved football team training optimization algorithm provided in the embodiment of the present application;

[0055] Figure 5 This is the second flow chart of the improved football team training optimization algorithm provided in the embodiment of the present application;

[0056] Figure 6 is a schematic diagram of the training results of the training set provided in the embodiment of the present application;

[0057] Figure 7 is a schematic diagram of test results of a test set provided in an embodiment of the present application;

[0058] Figure 8 is a bar graph of the sensitivity index of the auxiliary spray structure characteristic parameters of the auxiliary nozzle provided in the embodiment of the present application;

[0059] Fig. 9 It is a schematic diagram of the relationship between the number of PSO iteration steps and the Fitness function provided in the embodiment of the present application;

[0060] Fig.10 It is a structural schematic diagram of a device for predicting auxiliary jet air consumption of an air jet loom provided in an embodiment of the present application;

[0061] Fig.11 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application belong to the scope of protection of this application.

[0063] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0064] In combination with the accompanying drawings, the method for predicting the auxiliary jet air consumption of a jet loom, the device for predicting the auxiliary jet air consumption of a jet loom, the electronic device and the readable storage medium provided in the embodiments of the present application are described in detail through specific embodiments and their application scenarios.

[0065] Among them, the method for predicting the auxiliary jet air consumption of an air jet loom can be applied to a terminal, and can be specifically executed by hardware or software in the terminal.

[0066] The terminal includes, but is not limited to, a portable communication device such as a mobile phone or a tablet computer having a touch-sensitive surface (e.g., a touch screen display and / or a touch pad). It should also be understood that in some embodiments, the terminal may not be a portable communication device, but a desktop computer having a touch-sensitive surface (e.g., a touch screen display and / or a touch pad).

[0067] In the following various embodiments, a terminal including a display and a touch-sensitive surface is described. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, a mouse and a joystick.

[0068] The embodiment of the present application provides a method for predicting the auxiliary jet air consumption of a jet loom. The execution subject of the method can be an electronic device or a functional module or functional entity in the electronic device that can implement the method for predicting the auxiliary jet air consumption of a jet loom. The electronic devices mentioned in the embodiment of the present application include but are not limited to mobile phones, tablet computers, computers, cameras, and wearable devices. The method for predicting the auxiliary jet air consumption of a jet loom provided in the embodiment of the present application is described below using an electronic device as an example of the execution subject.

[0069] like Figure 1 As shown, the method for predicting the auxiliary jet air consumption of an air jet loom includes: step 110 and step 120.

[0070] Step 110, obtaining the structural parameters to be measured of the nozzle to be measured in the air jet loom, wherein the structural parameters to be measured include the inlet diameter, the straight tube diameter, the nozzle taper, the elliptical outlet major axis and the elliptical outlet minor axis.

[0071] The nozzle to be tested may be an auxiliary nozzle in an air jet loom.

[0072] The gas consumption of a nozzle refers to the gas flow rate consumed by the nozzle per unit time, and its unit can be cubic meters per minute (m³ / min) or liters per minute (L / min).

[0073] Air consumption is an important parameter to measure the performance and efficiency of air jet looms. Air consumption directly affects the production speed of fabrics and the energy consumption of air jet looms. Optimizing the air consumption of the nozzle helps to improve production efficiency and reduce operating costs.

[0074] The technical solution of the present application is described below using the auxiliary nozzle as an example, which is not to be regarded as limiting the protection scope of the present application.

[0075] like Figure 2 As shown, the auxiliary nozzle is a specially designed nozzle with a total length of 48 cm. A is the air outlet end of the auxiliary nozzle, AA is the longitudinal section of the air outlet end A enlarged by 2:1, and B is the cross-section of the air outlet end A.

[0076] The A1 axis direction of the elliptical hole formed in the cross section of the gas outlet A defines the major axis d1 of the elliptical outlet. The rotation angle of the A1 axis of the elliptical hole is β , the short axis d2 of the elliptical outlet is defined in the A2-axis direction of the elliptical hole.

[0077] As a key component in the air jet loom, the auxiliary nozzle is used to provide additional compressed air to help the weft yarn maintain speed and stability throughout the entire weaving width and prevent the weft yarn from deviating or being disturbed during the passage. The auxiliary main nozzle effectively guides the weft yarn through the shed formed by the warp yarn, which can improve the stability and efficiency of the weft yarn guidance. The air consumption at its outlet directly affects the energy efficiency of the entire machine.

[0078] In this step, data of the nozzle to be tested is collected to obtain the inlet diameter D1, straight tube diameter D2, nozzle taper α , the major axis d1 and the minor axis d2 of the elliptical outlet, etc., which can characterize the jet performance of the nozzle to be tested, and are related to the size and shape characteristics of the nozzle and affect the gas consumption.

[0079] Step 120, input the structural parameters to be tested into an airflow prediction model to obtain the nozzle air consumption of the nozzle to be tested output by the airflow prediction model, wherein the airflow prediction model is trained based on sample structural parameters and air consumption labels of sample nozzles.

[0080] Among them, the air flow prediction model can be built based on a deep learning model to predict the air consumption of the nozzle to be tested.

[0081] It can be understood that the sample structure parameters include the inlet diameter, the straight tube diameter, the nozzle taper, the elliptical outlet major axis and the elliptical outlet minor axis of the sample nozzle.

[0082] In addition, if the sample structural parameters of the training air flow prediction model include other structural parameters of the sample nozzle, the accuracy of the prediction of nozzle air consumption of other types of nozzles can be improved.

[0083] The nozzle gas consumption can be used to characterize the amount of gas consumed by the nozzle to be tested per unit time, and is the predicted value of the nozzle to be tested by the gas flow prediction model.

[0084] In this step, the inlet diameter, straight tube diameter, nozzle taper, elliptical outlet major axis and elliptical outlet minor axis and other test structural parameters are input into the air flow prediction model, and the air flow prediction model predicts the air consumption of the nozzle to be tested, and the nozzle air consumption of the nozzle to be tested is obtained and output.

[0085] According to the method for predicting the auxiliary jet air consumption of a jet loom provided in an embodiment of the present application, the air flow prediction model is used to predict the structural parameters to be measured that affect the air consumption, and the nozzle air consumption of the nozzle to be measured is obtained, thereby improving the prediction accuracy and efficiency. The method is more suitable for complex industrial scenarios such as textiles, and can accurately predict the nozzle air consumption in a complex industrial environment, providing data reference for the energy consumption management of the jet loom.

[0086] In some embodiments, the airflow prediction model includes an input layer, a hidden layer, and an output layer, and the hidden layer includes a convolutional layer, a pooling layer, and a fully connected layer;

[0087] The step of inputting the structural parameters to be measured into an air flow prediction model to obtain the nozzle air consumption of the nozzle to be measured output by the air flow prediction model comprises:

[0088] Receiving the structural parameters to be measured through the input layer;

[0089] Extracting features of the structural parameters to be measured through the convolution layer to obtain a first morphological feature of the nozzle to be measured;

[0090] Performing data compression on the first morphological feature through the pooling layer, and performing feature extraction on the compressed first morphological feature through the convolution layer to obtain a second morphological feature of the nozzle to be tested;

[0091] The first morphological feature and the second morphological feature are fused and mapped to the nozzle flow feature through the fully connected layer to generate the nozzle air consumption;

[0092] The nozzle gas consumption is output via the output layer.

[0093] like Figure 3As shown, the airflow prediction model includes an input layer (input), a hidden layer and an output layer (output).

[0094] The hidden layer can be constructed based on a one-dimensional convolutional neural network (1D-CNN) model to extract features. The hidden layer can be a convolutional layer and a pooling layer that are alternately connected and then connected to a fully connected layer.

[0095] The fully connected layer adopts the concept of weight sharing and uses numerical filters to extract features from the first morphological features and the second morphological features.

[0096] It is understandable that compared with traditional machine learning methods, convolutional neural network (CNN) not only has strong self-learning and self-debugging capabilities, but also has excellent classification and regression capabilities.

[0097] The core of the air flow prediction model is the convolution layer. Through local connection and pooling, the convolution layer can capture the first morphological features and second morphological features related to air consumption from the structural parameters to be measured.

[0098] The convolution kernel in the convolution layer is used to extract features, and its calculation formula is as follows:

[0099]

[0100] in, is the first morphological feature or the second morphological feature, A i is the input of the convolution layer; * is the convolution operation; λ For the activation function, choose the rectified linear unit function (Relu); and b i are the weight matrix and bias respectively.

[0101] It can be understood that the convolution kernel shares parameters on the entire input structure parameters to be measured, which can reduce the number of parameters of the airflow prediction model and improve training efficiency.

[0102] In addition, the pooling layer can be used to compress data and remove unnecessary information, thereby reducing the amount of computation and the number of parameters, while retaining important features and reducing the risk of overfitting. For example, the pooling layer can select the maximum pooling layer.

[0103] The fully connected layer fuses the first morphological features and the second morphological features extracted from each convolutional layer to achieve information aggregation and obtain the final predicted nozzle air consumption.

[0104] In actual execution, the measured structural parameters are received through the input layer of the airflow prediction model, the input data is processed through the convolution layer through a convolution operation, local features such as the first morphological features are extracted from the measured structural parameters, the first morphological features are compressed through the pooling layer, the first morphological features after data compression are further extracted through the convolution layer to obtain the second morphological features, the pooling layer further compresses the data of the extracted second morphological features, and so on, the first morphological features and the second morphological features extracted by the convolution layer are feature fused through the fully connected layer, the fused morphological features are mapped to nozzle flow features, the nozzle air consumption is generated, and the nozzle air consumption is output through the output layer.

[0105] In this embodiment, feature extraction is performed through the convolution layer to automatically learn features and reduce the number of parameters. Data compression is performed through the pooling layer to reduce the amount of calculation and provide translation invariance. Feature fusion is performed through the fully connected layer to integrate features and perform final regression prediction. The airflow prediction model effectively extracts features from the input structural parameters to be measured and performs accurate and fast predictions.

[0106] It should be noted that direct prediction through convolutional neural networks (CNN) has defects such as difficulty in parameter adjustment, complex training process, and difficulty in achieving the best prediction effect under different working conditions.

[0107] In some embodiments, the airflow prediction model is obtained based on the following steps:

[0108] Constructing a convolutional neural network model based on the input layer, the hidden layer and the output layer;

[0109] Optimizing the learning rate of the convolutional neural network model, as well as the convolution kernel size and the number of convolutions of the convolutional layer;

[0110] The optimized convolutional neural network model is trained to obtain the airflow prediction model.

[0111] In the actual implementation, an initialized convolutional neural network model is constructed, which includes an input layer, a hidden layer, and an output layer. The hidden layer can be represented as a series of convolutional layers and pooling layers that are alternately connected and then connected to a fully connected layer.

[0112] The learning rate of the CNN model, as well as the convolution kernel size and the number of convolutions of the convolution layer are optimized by using an improved football team training algorithm (IFTTA) to obtain an optimized CNN model.

[0113] A data set is constructed, which includes multiple training samples, and each training sample includes a sample structure parameter and an air consumption label of a sample nozzle.

[0114] The optimized CNN model is trained using the data set, and the trained CNN model is used as the airflow prediction model.

[0115] In this embodiment, by optimizing the learning rate, convolution kernel size and number of convolutions of the convolutional neural network model, the difficulty of parameter adjustment is reduced, the training process is simpler, and the prediction accuracy, training efficiency and generalization ability of the air flow prediction model are improved, so that the best prediction effect can be achieved under different working conditions.

[0116] In some embodiments, the optimizing the learning rate of the convolutional neural network model, and the convolution kernel size and the number of convolutions of the convolution layer includes:

[0117] Determine and optimize the fitness function of the convolutional neural network model, and obtain a first fitness value corresponding to a first search agent position of each individual in the population, wherein the individual is characterized as a hyperparameter group to be optimized of the convolutional neural network model;

[0118] Training each of the individuals, calculating a second search agent position of each of the individuals, and grouping the population based on the first fitness value;

[0119] Performing group training on each of the grouped individuals, and updating the population, so as to obtain a second fitness value of each of the individuals based on the second search agent position and the fitness function;

[0120] Based on the second fitness value, determining a first target individual in the population;

[0121] Training the first target individual to update the population and obtain an optimal search agent position of the first target individual;

[0122] Repeat to obtain multiple optimal search agent positions until the maximum number of iterations is reached to obtain a second target individual, and optimize the learning rate, the convolution kernel size and the number of convolutions of the convolutional neural network model through the target hyperparameter group corresponding to the second target individual.

[0123] Among them, the IFTTA algorithm is implemented by simulating the activities of players in a football training class. The football training class is divided into three parts: collective training, group training and individual additional training. The individuals in the population act as players.

[0124] It can be understood that the hyperparameter group to be optimized is a candidate combination of hyperparameters configured for the CNN model to be optimized, which may include hyperparameters such as the learning rate, convolution kernel size, and number of convolutions of the CNN model.

[0125] In the IFTTA algorithm, the individuals in the population are closely related to the parameters of the CNN model. Each individual represents a set of hyperparameters to be optimized for the CNN model configuration. The IFTTA algorithm continuously optimizes these hyperparameters through iteration to minimize the training error of the CNN model.

[0126] Among them, the core of the fitness function is the objective function, which is used to define the prediction error of the CNN model. After the root mean square error (RMSE) of the CNN model is used as the objective function (fitness function) of the IFTTA optimization algorithm to evaluate each individual, the population gradually evolves and finally obtains the target hyperparameter group corresponding to the second target individual. The target hyperparameter group is the optimal hyperparameter configuration of the CNN model.

[0127] The objective function is as follows:

[0128]

[0129] in, n is the number of training samples, is the actual value, as the gas consumption label, is the predicted value of the gas consumption of the training sample. Fitness The smaller the output value, the more predicted With actual value The smaller the difference, the higher the prediction accuracy.

[0130] like Figure 4 As shown in Figure 1, the hyperparameters corresponding to the individuals in the population form a search space. The IFTTA algorithm transforms the search problem into an optimization problem, aiming to find a set of optimal hyperparameter combinations that minimize the error of the CNN model.

[0131] In the collective training stage, the population G and the number of iterations N are initialized, the first fitness value of the initial individuals (players) is calculated, and the collective training and grouping are carried out. In the collective training stage, the update of individuals is based on four strategies: approaching the best individual, moving between the best and worst individuals, opening the worst individual and approaching the best individual, and individual state fluctuations, for example, adjusting the position of the individual through the golden sine strategy.

[0132] Perform group training, update the population, determine the first target individual in the population, perform additional training on the first target individual, and use adaptive tThe distribution is used to adjust the position and update the population until the maximum number of iterations is reached. The first target individual corresponding to the updated best search agent position is used as the second target individual, and the hyperparameter group corresponding to the second target individual is the target hyperparameter group.

[0133] Before training, each individual is used as a search agent, the position of the individual (search agent) is initialized as the first search agent position, and its first fitness value is calculated. The root mean square error (RMSE) can be used as the fitness function of the optimization algorithm, and then the best and worst individuals in the population are found. Finally, the initial group is backed up, which provides a reference point for the collective training stage and ensures that the algorithm has enough information to guide the subsequent optimization process.

[0134] In this embodiment, the fitness function is used to measure the quality of each individual, and is evaluated by calculating the error of the CNN model on the training set or the validation set.

[0135] In actual implementation, at the beginning of the training, individuals (players) conduct collective training under the guidance of the coach.

[0136] The individuals (players) in the population are divided into four different types: followers, finders, thinkers and volatiles, and in each iteration, the type of individuals (players) will be randomly changed.

[0137] The four types of individuals (players) are represented as shown below.

[0138]

[0139]

[0140]

[0141]

[0142] in, Indicates i individuals (players), where k is the current iteration number, i is the individual (player) ID, It is i Individuals (Players) In Dimension j The search agent position on ; the current best and worst individuals (players) are defined as and , is the dimension of the individual (player) after training j The state of is the number of degrees of freedom for the current iteration k of tDistribute random numbers, old is the state of the individual (player) before training, new It is the state of the individual (player) after training.

[0143] In some embodiments, calculating the second search agent position of each individual includes:

[0144]

[0145] in, k is the number of iterations of the current training, For the i The individual in dimension j On k The second search agent position after iterations, For the i The individual in dimension j On k The first search agent position before the iteration, For the dimension j Previous k The best search agent position for the iteration; for A random number in the range, for A random number in the range, All are golden ratio coefficients.

[0146] In actual implementation, in the individual (player) state fluctuations of the IFTTA algorithm, collective training is one of the core ideas of the IFTTA algorithm, which means that individuals in the entire population (each individual represents a set of CNN hyperparameters) learn and imitate each other, and adjust their positions through the information of the best and worst individuals, so that the population as a whole approaches the optimal solution.

[0147] On the basis of the original football team training algorithm (FTTA), the golden sine strategy was introduced in the collective training stage, and the adaptive t distribution was introduced in the individual additional training stage to adjust the position. The FTTA algorithm was improved to form the IFTTA algorithm, which can improve the correctness and computational efficiency of the regression model and is applicable to the engineering practice of air-jet looms.

[0148] The new second search agent position is calculated by the golden section coefficient and sine function to improve the global exploration ability of the search. The specific implementation is as follows:

[0149]

[0150] in, for A random number in the range, for A random number in the range, All are golden ratio coefficients.

[0151] In this embodiment, the population is updated by the golden ratio coefficient, so that the search agent can adjust its search path by the golden ratio and sinusoidal fluctuation, thereby better balancing the global and local search capabilities in complex optimization problems.

[0152] In some embodiments, performing group training on each of the grouped individuals and updating the population includes:

[0153] determining a learning probability, a communication probability, and an error probability for the group training;

[0154] Based on the learning probability, the communication probability and the error probability, group training is performed on each individual, and the training state of each individual is determined, so as to update the population according to the training state.

[0155] Therein, a new second search agent position is generated through an update step after the collective training phase.

[0156] The update of the population includes the update of the individual search agent position. After each population update, it is necessary to check the individual position to determine whether it exceeds the defined upper or lower bounds, and ensure that the individual position is within the predefined upper and lower bounds to achieve boundary control.

[0157] For example, in order to ensure that these second search proxy positions are within predefined upper and lower boundaries, boundary control of the second search proxy positions is required.

[0158] The training process enters the group training stage, where the players are divided into four categories: forwards, midfielders, defenders, and goalkeepers according to their characteristics. In this stage, an adaptive clustering method (MixGaussEM) is used to simulate the coach's decision-making process through a clustering algorithm and divide the individuals into four categories according to their characteristics.

[0159] Next, group training was defined into three training regimes: optimal learning, random learning, and random communication.

[0160] The learning probability is defined as , the probability of exchange is defined as , individuals (players) randomly select states in each training session. The formulas for the three states are defined as follows:

[0161]

[0162]

[0163]

[0164]

[0165] in, yes Group i The individual in dimension j Previous k The search agent position for the iteration; yes Group in dimension j Previous k The best search agent position for the iteration; yes Group in dimension j Previous k The search agent position of a random individual (player) at iteration .

[0166] In the process of group training, there is a certain probability of making mistakes, that is, accidentally learning the content of other individuals in other dimensions. The probability of this situation is very low and is real and objective. The error probability is defined as .

[0167]

[0168] In the above formula, yes Group k The search agent position of a random individual (player) at iterations, Representation Group , yes Group in dimension j Previous k The search agent position of a random individual (player) at iterations, is a normally distributed random number.

[0169] It can be understood that before additional training of individuals, the population is updated and the second fitness value of the current population is calculated, the individuals with better performance are retained as the first target individuals, and by comparing the optimal solution of the current iteration with that of the previous iteration, it is ensured that the population gradually approaches the global optimal solution, wherein the calculation method of the second fitness value is the same as that of the first fitness value, which will not be repeated here.

[0170] After the update, enter the individual extra training stage, select the best individual (player) as the first target individual, let him practice to become better, so that he can better drive the training status of others. The training formula is as follows:

[0171]

[0172] in, and Joint variation is used to describe individual additional training, k is the number of iterations.

[0173] In this process, adaptive t The distribution is used to adjust the best individual (the first target individual) additional training process, through adaptive t distribution to guide individuals towards the optimal solution.

[0174] In individual additional training, use t The individuals after the distribution update also need boundary control to ensure that the position is reasonable. In some embodiments, the training of the first target individual to update the population and obtain the best search agent position of the first target individual includes:

[0175]

[0176]

[0177] in, is an adaptive parameter used to adjust t The scale of the random numbers generated by the distribution is used to control the search step size of each individual in each iteration; m is a positive integer, k is the number of iterations of the current training, is the size of the population, t () is generated t Function for distributing random numbers, is the search agent position of the first target individual in the initialization phase, The first target individual is k The best search agent position before the iteration; is the The first target individual is k The best search agent position after iterations.

[0178] in, This allows for extensive exploration in the early stages of the search, while focusing on a smaller area for more precise searches in the later stages. This mechanism helps improve the global search capability and convergence speed of the optimization algorithm.

[0179] m The value of can be adjusted manually. m It can effectively amplify the value of the function, help the algorithm escape from the local optimal solution, and improve the global search capability.

[0180] In this embodiment, by training the first target individual, the globally optimal best search agent position can be obtained, which effectively improves the parameter quality of the CNN model.

[0181] After training multiple first target individuals for the maximum number of iterations, the optimization is terminated, the best first target individual is selected from the multiple first target individuals as the second target individual, and the target hyperparameter group corresponding to the second target individual is migrated to the CNN model to optimize the hyperparameters of the CNN model, such as the learning rate, convolution kernel size, and number of convolutions.

[0182] After optimizing the hyperparameters of the CNN model, the CNN model is evaluated using training samples constructed using sample structural parameters and gas consumption labels of sample nozzles to obtain an airflow prediction model.

[0183] It should be noted that in the textile industry scenario, the air jet loom part uses a round hole auxiliary nozzle. Different fabrics require different auxiliary spray structure parameters and different air consumption.

[0184] A specific embodiment is described below.

[0185] In view of the defect that the air compressor in the existing air jet loom cannot accurately control the airflow, the IFTTA-CNN algorithm provided in this application optimizes the hyperparameters of the CNN model through IFTTA, and predicts the airflow of a single auxiliary nozzle through the CNN model, so as to accurately control the output flow of the air compressor. It is suitable for any form of auxiliary nozzles, including round hole outlets and elliptical hole outlets. The elliptical hole is a new type of auxiliary nozzle proposed. The airflow speed is higher than that of a single round hole, which improves the weft insertion speed, can intelligently predict the air consumption of the nozzle of the air jet loom, accurately control the overall output flow of the air compressor, and reduce air consumption and power consumption.

[0186] like Figure 5 As shown in Figure 2, data set preparation. The sample structural parameters of the sample nozzle are collected, including six input features, namely the inlet diameter, straight pipe diameter, nozzle taper, outlet major axis and outlet minor axis of the auxiliary nozzle, and the measured value of the gas consumption of the sample nozzle as the gas consumption label.

[0187] The collected data is preprocessed by outlier detection and data normalization, and the combination of the sample structure parameters after preprocessing and the gas consumption labels corresponding to the sample structure parameters is used as a training sample to obtain multiple training samples.

[0188] Multiple training samples are divided into training set and test set in a ratio of 7:3.

[0189] Design a basic CNN structure, including an input layer, at least one convolutional layer, at least one pooling layer, a fully connected layer, and an output layer, and set the hyperparameters that need to be optimized, such as learning rate, convolution kernel size, etc., and set the maximum number of training times and sample batch capacity.

[0190] Initialize the parameters of the IFTTA algorithm, including the number of populations as 15, the maximum number of iterations as 20, the number of optimized parameters as 3, and the upper and lower bounds of the parameter values, for example, the lower bound of the parameter is [1e-4, 2, 8], and the upper bound of the parameter is [1e-2, 6, 128].

[0191] An IFTTA-CNN regression model was constructed, and the hyperparameters of the CNN model were optimized using the IFTTA algorithm, including collective training, group training, and individual additional training. The CNN model was evaluated after parameter optimization using the coefficient of determination (R2) and root mean square error (RMSE) to obtain the airflow prediction model.

[0192] The air compressor of the air jet loom can control the air jet to be supplied to multiple nozzles at the same time, and the number of nozzles can be adjusted according to the weaving width of the air jet loom.

[0193] For each nozzle, a sensitivity analysis is performed on the air flow prediction model, and the air flow prediction model is optimized using the particle swarm optimization (PSO) algorithm to obtain the minimum air consumption of each nozzle and the optimal structural parameters corresponding to the minimum air consumption.

[0194] According to the optimal structural parameters of the nozzle to be tested, a nozzle with a corresponding structure is selected, and the output power of the air compressor to the nozzle is controlled according to the minimum air consumption.

[0195] In the process of optimizing the airflow prediction model using the PSO algorithm, the optimization goal is defined, for example, the minimum air consumption, and the PSO parameters are set, including the number of particles, the maximum number of iterations, the inertia weight, the individual learning factor, and the social learning factor;

[0196] Represent each particle as a vector containing hyperparameters of the airflow prediction model, such as the learning rate, number of convolutional layers, and activation function type;

[0197] Randomly initialize the parameter value of each particle to cover the entire search space;

[0198] The airflow prediction model is trained using the parameter settings of the current particles, and the performance of the airflow prediction model is evaluated. The fitness value is calculated, and the speed and position of each particle are updated according to the fitness value. This operation is repeated until the maximum number of iterations is reached or other termination conditions are met. The optimal particle position is selected according to the fitness value as the final hyperparameter configuration of the airflow prediction model to optimize the airflow prediction model.

[0199] In this embodiment, not only can the air consumption of the nozzle be accurately predicted and the optimal structural parameters of the nozzle be obtained, but the output power of the air compressor in the air jet loom can also be accurately controlled.

[0200] like Figure 6 As shown, the horizontal axis is the sample size of the training sample in the training set, the unit is pieces, and the vertical axis is the gas consumption, the unit is L·min -1 ,like Figure 7 As shown, the horizontal axis is the sample size of the training samples in the test set, the unit is pieces, and the vertical axis is the gas consumption, the unit is L·min -1 .

[0201] The sample optimization results of the IFTTA-CNN model are shown in Table 1:

[0202] Table 1 Sample optimization results of IFTTA-CNN model

[0203]

[0204] After training and evaluating the CNN model, the airflow prediction model can be optimized through the random forest algorithm or the adaptive learning rate algorithm. For example, the importance of features can be evaluated to help identify the features that contribute most to the prediction, or the data stream can be processed step by step and the model can be updated in real time to adapt to dynamically changing data, so as to assist the airflow prediction model in predicting the air consumption of nozzles with different structures.

[0205] In this embodiment, the CNN model parameters are automatically optimized through IFTTA, which improves the prediction accuracy and adaptability of the CNN model, overcomes the problems of low prediction accuracy and difficulty in tuning model parameters in the prior art, and thus achieves high-precision prediction of nozzle gas consumption.

[0206] For example, a jet loom has 20 auxiliary nozzles, and the auxiliary nozzles consume 70% of the air. The process of the air compressor supplying air to the auxiliary nozzles consumes electrical energy, and the auxiliary nozzles are the largest gas- and energy-consuming components. After constructing the IFTTA-CNN regression model, the present application arbitrarily inputs the structural parameters of an auxiliary nozzle to be measured into the air flow prediction model, such as the inlet diameter D1, the straight pipe diameter D2, the elliptical outlet major axis d1, and the elliptical outlet minor axis d2, to predict the nozzle air consumption of this auxiliary nozzle. Similarly, the nozzle air consumption of 20 auxiliary nozzles is obtained, thereby achieving precise control of the overall output flow of the air compressor and achieving the purpose of energy saving and consumption reduction.

[0207] like Figure 8 As shown, the horizontal axis is the auxiliary nozzle's auxiliary spray structure characteristic parameters, including inlet diameter D1, straight pipe diameter D2, nozzle taper α, the major axis d1 of the elliptical outlet and the minor axis d2 of the elliptical outlet. The ordinate is the sensitivity index. The structural parameter of the elliptical hole auxiliary spray that has the greatest impact on the gas consumption is the major diameter d1 of the elliptical hole, followed by the minor diameter d2 of the elliptical hole. The structural parameters that have a greater impact on the gas consumption are: the inlet diameter D1 and the straight pipe diameter D2 have a greater impact on the gas consumption. The nozzle taper α The impact on gas consumption is minimal.

[0208] like Fig. 9 As shown, the horizontal axis is the number of iteration steps, and the vertical axis is the Fitness value (minimum gas consumption). In the process of obtaining the optimal structural parameters and the predicted nozzle gas consumption, the PSO algorithm is used to take the minimum gas consumption as the optimization goal to find the minimum gas consumption and the corresponding structural parameters. The constraints are the structural geometric parameters, among which the inlet diameter parameter range is , straight pipe diameter , the major axis of the ellipse outlet , the minor axis of the ellipse outlet The PSO algorithm was used to find the minimum gas consumption value of 19.6600L / min after 20 iterations. The corresponding elliptical hole auxiliary spray structure parameters are inlet diameter D1=8.2mm, straight pipe diameter D2=5.7mm, elliptical outlet major axis d1=1.4mm, elliptical outlet minor axis d2=0.8mm, nozzle taper .

[0209] The air jet loom auxiliary jet air consumption prediction method provided in the embodiment of the present application can be executed by an air jet loom auxiliary jet air consumption prediction device. In the embodiment of the present application, the air jet loom auxiliary jet air consumption prediction method is performed by an air jet loom auxiliary jet air consumption prediction device as an example to illustrate the air jet loom auxiliary jet air consumption prediction device provided in the embodiment of the present application.

[0210] An embodiment of the present application also provides a device for predicting the auxiliary jet air consumption of an air jet loom.

[0211] like Fig.10 As shown, the device for predicting the auxiliary jet air consumption of an air jet loom includes: an acquisition module 1010 and a processing module 1020.

[0212] An acquisition module 1010 is used to acquire the structural parameters to be measured of the nozzle to be measured in the air jet loom, wherein the structural parameters to be measured include an inlet diameter, a straight tube diameter, a nozzle taper, an elliptical outlet major axis, and an elliptical outlet minor axis;

[0213] The processing module 1020 is used to input the structural parameters to be tested into the air flow prediction model to obtain the nozzle air consumption of the nozzle to be tested output by the air flow prediction model. The air flow prediction model is trained based on the sample structural parameters and air consumption labels of the sample nozzle.

[0214] It can be understood that the device for predicting the auxiliary jet air consumption of the air jet loom can be a host computer.

[0215] In the upper computer, the acquisition module 1010 is connected to the detection chip (lower computer), the detection chip (lower computer) receives the original data collected by the sensor and standardizes the data, and the processing module 1020 executes the process of training or applying the airflow prediction model based on the standardized data.

[0216] According to the device for predicting the auxiliary jet air consumption of a jet loom provided in the embodiment of the present application, the air flow prediction model is used to predict the structural parameters to be measured that affect the air consumption, and the nozzle air consumption of the nozzle to be measured is obtained, thereby improving the prediction accuracy and efficiency. The device is more suitable for complex industrial scenarios such as textiles, and can accurately predict the nozzle air consumption in a complex industrial environment, providing data reference for the energy consumption management of the jet loom.

[0217] The device for predicting the auxiliary air consumption of the air jet loom in the embodiment of the present application can be an electronic device, or a component in the electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal, or other devices other than a terminal. For example, the electronic device can be a personal computer (PC), a server, a workstation, a mobile phone, a host computer, a single-chip microcomputer, an industrial computer, etc., which is not specifically limited in the embodiment of the present application.

[0218] The device for predicting the auxiliary jet air consumption of an air jet loom in the embodiment of the present application may be a device having an operating system. The operating system may be a Windows operating system, a Linux operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0219] The device for predicting the auxiliary jet air consumption of an air jet loom provided in the embodiment of the present application can achieve Figures 1 to 9 To avoid repetition, the various processes implemented by the embodiment of the method for predicting the auxiliary jet air consumption of an air jet loom are not described here.

[0220] In some embodiments, Fig.11 As shown, an embodiment of the present application also provides an electronic device 1100, including a processor 1101, a memory 1102, and a computer program stored in the memory 1102 and executable on the processor 1101. When the program is executed by the processor 1101, each process of the above-mentioned jet loom auxiliary jet air consumption prediction method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0221] The electronic device 1100 may include a host computer.

[0222] It should be noted that the electronic devices in the embodiments of the present application include the mobile electronic devices and non-mobile electronic devices mentioned above.

[0223] An embodiment of the present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, each process of the above-mentioned jet loom auxiliary jet air consumption prediction method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0224] The processor is the processor in the electronic device in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0225] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned method for predicting the auxiliary jet air consumption of an air jet loom.

[0226] The processor is the processor in the electronic device in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk.

[0227] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, which are coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above-mentioned jet loom auxiliary jet air consumption prediction method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0228] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0229] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0230] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment method can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, disk, CD), and includes a number of instructions for a terminal (which can be a mobile phone, computer, server, or network device, etc.) to execute the air consumption prediction method for the auxiliary jet of the air jet loom in each embodiment of the present application.

[0231] In the description of this application, "first feature" or "second feature" may include one or more of the features.

[0232] In the description of the present application, “plurality” means two or more.

[0233] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

[0234] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0235] Although the embodiments of the present application have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the claims and their equivalents.

Claims

1. A method for predicting the air consumption of auxiliary jet of an air jet loom, characterized in that: include: Acquire the structural parameters to be measured of the nozzle to be measured in the air jet loom, wherein the structural parameters to be measured include the inlet diameter, the straight tube diameter, the nozzle taper, the elliptical outlet major axis and the elliptical outlet minor axis; Inputting the structure parameters to be tested into an air flow prediction model to obtain the nozzle air consumption of the nozzle to be tested output by the air flow prediction model, wherein the air flow prediction model is trained based on sample structure parameters and air consumption labels of sample nozzles; The airflow prediction model includes an input layer, a hidden layer and an output layer, wherein the hidden layer includes a convolutional layer, a pooling layer and a fully connected layer; Receiving the structural parameters to be measured through the input layer; Extracting features of the structural parameters to be measured through the convolution layer to obtain a first morphological feature of the nozzle to be measured; Performing data compression on the first morphological feature through the pooling layer, and performing feature extraction on the compressed first morphological feature through the convolution layer to obtain a second morphological feature of the nozzle to be tested; The first morphological feature and the second morphological feature are fused and mapped to the nozzle flow feature through the fully connected layer to generate the nozzle air consumption; Outputting the nozzle gas consumption through the output layer; Determine a fitness function for optimizing the convolutional neural network model, and obtain a first fitness value corresponding to a first search agent position of each individual in the population, wherein the individual is characterized as a hyperparameter group to be optimized for the convolutional neural network model; Training each of the individuals, calculating a second search agent position of each of the individuals, and grouping the population based on the first fitness value; Performing group training on each of the grouped individuals, and updating the population, so as to obtain a second fitness value of each of the individuals based on the second search agent position and the fitness function; Based on the second fitness value, determining a first target individual in the population; Training the first target individual to update the population and obtain an optimal search agent position of the first target individual; Repeat to obtain multiple optimal search agent positions until the maximum number of iterations is reached to obtain a second target individual, and optimize the learning rate, convolution kernel size and number of convolutions of the convolutional neural network model through the target hyperparameter group corresponding to the second target individual.

2. The method for predicting the auxiliary jet air consumption of an air jet loom according to claim 1, characterized in that: The airflow prediction model is obtained based on the following steps: Constructing a convolutional neural network model based on the input layer, the hidden layer and the output layer; Optimizing the learning rate of the convolutional neural network model, as well as the convolution kernel size and the number of convolutions of the convolutional layer; The optimized convolutional neural network model is trained to obtain the airflow prediction model.

3. The method for predicting the auxiliary jet air consumption of an air jet loom according to claim 1, characterized in that: The calculating the second search agent position of each individual includes: ； in, k is the number of iterations of the current training, For the i The individual in dimension j On k The second search agent position after iterations, For the i The individual in dimension j On k The first search agent position before the iteration, For the k Iterations in dimension j Best search proxy locations on; for A random number in the range, for A random number in the range, All are golden ratio coefficients.

4. The method for predicting the auxiliary jet air consumption of an air jet loom according to claim 1, characterized in that: The step of performing group training on each of the grouped individuals and updating the population comprises: determining a learning probability, a communication probability, and an error probability for the group training; Based on the learning probability, the communication probability and the error probability, group training is performed on each individual, and the training state of each individual is determined, so as to update the population according to the training state.

5. The method for predicting auxiliary jet air consumption of an air jet loom according to claim 1, characterized in that: The step of training the first target individual to update the population and obtain the best search agent position of the first target individual includes: ； ； in, is an adaptive parameter used to adjust t The scale of the random numbers generated by the distribution is used to control the search step size of each individual in each iteration; m is a positive integer, k is the number of iterations of the current training, is the size of the population, t () is generated t Function for distributing random numbers, is the search agent position of the first target individual in the initialization phase, The first target individual is k The best search agent position before iterations; The first target individual is k The best search agent position after iterations.

6. A device for predicting the air consumption of auxiliary jet of an air jet loom, characterized in that: include: An acquisition module is used to acquire the structural parameters to be measured of the nozzle to be measured in the air jet loom, wherein the structural parameters to be measured include the inlet diameter, the straight tube diameter, the nozzle taper, the elliptical outlet major axis and the elliptical outlet minor axis; A processing module, used for inputting the structural parameters to be tested into an air flow prediction model to obtain the nozzle air consumption of the nozzle to be tested output by the air flow prediction model, wherein the air flow prediction model is trained based on sample structural parameters and air consumption labels of sample nozzles; The airflow prediction model includes an input layer, a hidden layer and an output layer, wherein the hidden layer includes a convolutional layer, a pooling layer and a fully connected layer; Receiving the structural parameters to be measured through the input layer; Extracting features of the structural parameters to be measured through the convolution layer to obtain a first morphological feature of the nozzle to be measured; Performing data compression on the first morphological feature through the pooling layer, and performing feature extraction on the compressed first morphological feature through the convolution layer to obtain a second morphological feature of the nozzle to be tested; The first morphological feature and the second morphological feature are fused and mapped to the nozzle flow feature through the fully connected layer to generate the nozzle air consumption; Outputting the nozzle gas consumption through the output layer; Determine a fitness function for optimizing the convolutional neural network model, and obtain a first fitness value corresponding to a first search agent position of each individual in the population, wherein the individual is characterized as a hyperparameter group to be optimized for the convolutional neural network model; Training each of the individuals, calculating a second search agent position of each of the individuals, and grouping the population based on the first fitness value; Performing group training on each of the grouped individuals, and updating the population, so as to obtain a second fitness value of each of the individuals based on the second search agent position and the fitness function; Based on the second fitness value, determining a first target individual in the population; Training the first target individual to update the population and obtain an optimal search agent position of the first target individual; Repeat to obtain multiple optimal search agent positions until the maximum number of iterations is reached to obtain a second target individual, and optimize the learning rate, convolution kernel size and number of convolutions of the convolutional neural network model through the target hyperparameter group corresponding to the second target individual.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method for predicting the auxiliary jet air consumption of an air jet loom as described in any one of claims 1 to 5 is implemented.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for predicting the auxiliary jet air consumption of an air jet loom as described in any one of claims 1 to 5 is implemented.

Images