A spinning frame front roller speed intelligent control system and method

Abstract

The present invention belongs to the technical field of spinning control, and discloses a system and method for intelligently controlling the rotational speed of the front roller of a spinning frame, comprising: collecting data on the production process of a target spinning frame, obtaining and storing corresponding spinning frame information; the spinning frame information includes operating data and production data; processing the obtained spinning frame information, and constructing a corresponding multi-objective optimization function based on the data processing result; constructing a corresponding production parameter tuning model based on the obtained target optimization function; and performing parameter control on the target spinning frame in combination with the production parameter tuning model; the present invention is of great significance for improving the production efficiency and product quality of spinning frames and reducing energy consumption and labor costs.

Description

Technical Field

[0001] The present invention relates to the technical field of spinning control, and more specifically, to a system and method for intelligently controlling the rotational speed of a front roller of a spinning frame. Background Art

[0002] As the textile industry's requirements for production efficiency and product quality continue to increase, the automation and intelligence levels of spinning frames also need to be improved accordingly; some existing detection and control methods cannot meet the high-precision and high-efficiency requirements of modern textile production, and more intelligent and automated systems are needed to achieve real-time monitoring and control; in recent years, with the rapid development of the Internet of Things, big data, and artificial intelligence technologies, intelligent control technologies have gradually been applied to the speed control of textile equipment; however, most of the existing intelligent speed control methods only optimize a single speed parameter, ignoring the many factors involved in the operation of the spinning frame, such as energy consumption, output, and quality, making it difficult to achieve the global optimal control effect.

[0003] Compared with existing technologies, in traditional spinning production processes, the front roller speed is usually manually adjusted by the operator or controlled in a fixed mode. These methods are difficult to adapt to different yarn types and changes in production environments, resulting in low production efficiency, high energy consumption, and unstable product quality. These problems seriously affect the production efficiency and product quality of the spinning frame.

[0004] In view of this, the present invention proposes a spinning frame front roller speed intelligent control system and method to solve the above problems. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art and achieve the above-mentioned objectives, the present invention provides the following technical solutions:

[0006] A method for intelligently controlling the rotation speed of a front roller of a spinning frame, comprising:

[0007] Step 1: collecting data on the production process of the target spinning frame, obtaining and storing corresponding spinning frame information; the spinning frame information includes operation data and production data;

[0008] Step 2: Process the obtained spinning frame information and construct the corresponding multi-objective optimization function based on the data processing results;

[0009] Step three: constructing a corresponding production parameter tuning model based on the obtained target optimization function; and performing parameter control on the target spinning frame in combination with the production parameter tuning model.

[0010] Furthermore, the process of collecting data on the production process of the target spinning frame, obtaining corresponding spinning frame information and storing the information includes:

[0011] Setting a data collection node, collecting relevant data in the production process of the corresponding spinning frame based on the data collection node, and obtaining corresponding spinning frame information; the spinning frame information includes spinning operation data, spinning energy consumption data and spinning production data;

[0012] The collected spinning frame information is subjected to data preprocessing; after the data preprocessing is completed, the spinning frame information after the data preprocessing is stored in a pre-built information database; wherein, the information database also stores corresponding spinning transaction information.

[0013] Furthermore, the process of data processing the obtained spinning frame information includes:

[0014] A two-dimensional rectangular coordinate system is constructed with time as the horizontal axis and spinning frame information as the vertical axis, and the obtained rectangular coordinate system is mapped into a corresponding two-dimensional rectangular coordinate system to obtain a corresponding spinning change curve; the spinning change curve includes a spinning operation data curve, a spinning energy consumption data curve, and a spinning production data curve;

[0015] The obtained spinning operation data curve, spinning energy consumption data curve and spinning production data curve are sampled respectively, and corresponding operation matrix, energy consumption matrix and production matrix are constructed based on the obtained data;

[0016] Normalize the obtained energy consumption matrix and operation matrix respectively, and extract the components of the normalized energy consumption matrix and operation matrix based on the principal component analysis algorithm to obtain the corresponding feature pairs;

[0017] Based on the above feature pair acquisition process, and by randomly sampling the spinning operation data curve and the spinning energy consumption data curve to update the operation matrix and the feature matrix, and repeating the feature pair extraction process, a plurality of feature pairs are obtained; and a corresponding first feature space matrix is ​​constructed based on the feature pairs;

[0018] Based on the method for obtaining the first characteristic space matrix, a second characteristic space matrix corresponding to the corresponding spinning operation data and the corresponding spinning production data is obtained;

[0019] Acquire the collected spinning transaction information, and construct a corresponding spinning profit curve based on the information, and then perform data fitting between the corresponding spinning profit curve and the spinning operation curve based on the collection time to obtain a corresponding operation-profit fitting curve;

[0020] Furthermore, a corresponding multi-objective optimization function is constructed based on the operation-profit fitting curve, the first feature space matrix and the second feature space matrix.

[0021] Furthermore, the formula for obtaining feature pairs is: Where, and Respectively represent the feature u within the a-th feature pair a and v a The corresponding eigenvector; ρ a Represents the unit eigenvector corresponding to the matrix TM when the a-th feature pair is extracted; Indicates the feature v within the a-th feature pair a The corresponding weight vector; ; Where A and B represent the energy consumption matrix and the operation matrix respectively; θ represents the maximum eigenvalue corresponding to the matrix TM; where TM=A T BB T A;

[0022] Multi-objective optimization function ; G(E1) and G(E2) refer to the linear functions defined by the corresponding first feature space matrix and the second feature space matrix respectively; F(b) is the defining linear function of the corresponding operation-profit fitting curve; E1, E2 and b respectively represent the functional independent variables of the linear function defined by the corresponding first feature space matrix, the second feature space matrix and the operation-profit fitting curve.

[0023] Furthermore, the process of constructing a corresponding production parameter tuning model based on the obtained target optimization function includes:

[0024] The network framework of the production parameter tuning model is defined as an improved convolutional neural network, which is used to learn the nonlinear relationship between the spinning operation data and the spinning production data in the spinning frame information;

[0025] The basic architecture of the improved convolutional neural network is an input layer, a multi-scale convolution layer and an output layer;

[0026] The output layer is used to receive input training samples and convert the corresponding training samples into a one-dimensional data vector;

[0027] The multi-scale convolution layer consists of a convolution layer and a pooling layer, and a plurality of multi-scale convolution layers are provided in the corresponding improved convolutional neural network; the convolution layer is used to perform a convolution operation on the one-dimensional data vector output by the input layer to obtain the output features corresponding to the corresponding convolution layer; the pooling layer is used to perform a pooling operation on the output features of the convolution layer;

[0028] A fully connected layer is provided in the output layer, and the fully connected layer is used to map the output features learned by the improved convolutional neural network to the corresponding sample space, and the output layer outputs the results;

[0029] Define the loss function ;in, The total number of training samples input to the production parameter tuning model, z is the index variable used to traverse all training samples, Generate the output of the parameter tuning model for the zth training example, where Y(z) is the label of the zth training example;

[0030] Based on the multi-objective optimization function, the weights in each layer of the improved convolutional neural network are initialized and set, and the initialization setting is completed; based on the historical spinning frame information, a corresponding training data set is constructed, and the training samples in the training data set are divided into several training batches, and based on it, the corresponding production parameter tuning model is iteratively trained, and the value of the corresponding loss function is recorded. When the loss function in several consecutive training batches no longer decreases or changes, the model parameters at this time are saved, and the construction process of the corresponding production parameter tuning model is completed.

[0031] Furthermore, the formula for defining the convolution operation is: Where, Indicates the The j-th output feature in the convolution layer within the multi-scale convolution layer; Represents the j-th output feature of the previous layer in the multi-scale convolutional layer; (i.e., The output of the pooling layer within the multi-scale convolutional layer; >0 and is an integer); M j Indicates the The total number of output results of the previous convolution layer in a multi-scale convolution layer; Indicates the The weight matrix of the convolution kernel in the convolution layer; Represents the bias term of the convolutional layer; Represents the convolution operation, f() represents the selected activation function;

[0032] The formula defining the pooling operation is: Where, Indicates the The downsampling parameter of the i-th output feature in the pooling layer within the multi-scale convolutional layer, H() represents the downsampling function; Represents the bias term of the pooling layer; Indicates the The i-th output feature of the pooling layer in the multi-scale convolutional layer; , i and j are all natural numbers, representing the layer indexes of the multi-scale convolutional layer and the pooling layer and convolution layer within the multi-scale convolutional layer, respectively.

[0033] Furthermore, the process of initializing the weights in each layer of the improved convolutional neural network includes:

[0034] Obtain the mean square error of the network performance function in the corresponding improved convolutional neural network, and construct the corresponding fitness function based on the obtained multi-objective optimization function ; Where MSE represents the mean square error of the network performance function, and Function weights representing the multi-objective optimization function and mean square error;

[0035] At the same time, a particle swarm set is initialized based on the learning mode of the improved convolutional neural network, and the number of particles and the spatial dimension corresponding to the corresponding particle swarm set are defined;

[0036] Obtaining the initial velocity and initial position of each particle in the corresponding particle group, marking them as Sp and Wp respectively; obtaining the fitness value corresponding to the corresponding particle based on the fitness function, and updating the initial position and initial velocity of the corresponding particle based on the fitness value;

[0037] Based on the fitness function, the initial position and initial velocity of each particle in the corresponding particle swarm are continuously updated until a preset iteration stop condition is met, thereby obtaining the corresponding global optimal solution; the obtained global optimal solution is mapped to the corresponding improved convolutional neural network, and then a new fitness value is obtained through the positive feedback output of the improved convolutional neural network, and the weights of the improved convolutional neural network are updated based on the new fitness value.

[0038] Furthermore, the update formula for the initial position is defined as:

[0039] Where, represents the particle position of particle p at the k+1th iteration; represents the particle velocity of particle p at the kth iteration; and They represent the individual optimal solution and the global optimal solution at the kth iteration respectively; r1 and r2 are constants, and both r1 and r2 are ∈ [0,1];

[0040] ω represents the adaptive inertia weight; Where, ω max and ω min Respectively represent the maximum and minimum values ​​of the preset inertia weight; k and k max Represent the current number of iterations and the maximum number of iterations respectively; α and ε represent the speed coefficients, which are fixed constants;

[0041] c1 and c2 represent learning factors; where ; Where c10, c20 and c12, c22 represent the initial and final values ​​of the learning factors c1 and c2 respectively.

[0042] Furthermore, the process of controlling the parameters of the target spinning frame in combination with the production parameter tuning model includes:

[0043] Based on the information database, real-time spinning frame information corresponding to the target spinning frame is obtained, and the real-time spinning frame information is input into the corresponding production parameter tuning model to obtain a corresponding model output result; the model output result is the optimal data after optimizing the target spinning operation data;

[0044] The obtained model output result is fed back to the control terminal of the corresponding spinning frame; further, the control terminal adjusts the spinning operation parameters of the corresponding spinning frame based on the model output result until the actual operation parameters are the same as the model output result.

[0045] Furthermore, a spinning frame front roller speed intelligent control system includes:

[0046] A data acquisition module is used to collect data on the production process of the target spinning frame, obtain corresponding spinning frame information and store it; the spinning frame information includes operation data and production data;

[0047] A data processing module is used to process the obtained spinning frame information and construct a corresponding multi-objective optimization function based on the data processing results;

[0048] The data control module is used to construct a corresponding production parameter tuning model based on the obtained target optimization function; and to control the parameters of the target spinning frame in combination with the production parameter tuning model.

[0049] The technical effects and advantages of the intelligent control system and method for the front roller speed of a spinning frame of the present invention are as follows:

[0050] 1. The multi-objective optimization function constructed based on data processing results can comprehensively consider multiple factors such as energy consumption, output and profit to achieve the optimal configuration of spinning frame operating parameters. It helps to reduce energy consumption and increase profits while meeting production needs, achieving a win-win situation in economic and social benefits.

[0051] 2. The production parameter tuning model constructed using an improved convolutional neural network can learn the nonlinear relationship between the spinning operation data and the spinning production data within the spinning frame information, thereby achieving more accurate speed control. It not only has high prediction accuracy and generalization ability, but can also adapt to changes in different production environments and process requirements, thereby improving the flexibility and adaptability of control.

[0052] 3. The present invention realizes the intelligent control of the rotation speed of the front roller of the spinning frame, and can automatically adjust the rotation speed according to real-time data and optimization goals, making the operation of the spinning frame more stable and efficient; it not only improves production efficiency and product quality, but also reduces operation difficulty and labor costs, bringing significant economic benefits and competitive advantages to the enterprise. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 Schematic diagram of a method for intelligently controlling the rotational speed of the front roller of a spinning frame according to the present invention;

[0054] Figure 2 This is a schematic diagram of an intelligent control system for the front roller speed of a spinning frame according to the present invention. DETAILED DESCRIPTION

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0056] Example 1

[0057] See also Figure 1 As shown, the method for intelligently controlling the rotation speed of the front roller of a spinning frame described in this embodiment includes:

[0058] Step 1: collecting data on the production process of the target spinning frame, obtaining and storing corresponding spinning frame information; the spinning frame information includes operation data and production data;

[0059] Step 2: Process the obtained spinning frame information and construct the corresponding multi-objective optimization function based on the data processing results;

[0060] Step 3: constructing a corresponding production parameter tuning model based on the obtained target optimization function; and performing parameter control on the target spinning frame in combination with the production parameter tuning model;

[0061] It should be further explained that, in the specific implementation process, the process of collecting data on the production process of the target spinning frame, obtaining corresponding spinning frame information and storing it includes:

[0062] Setting a data collection node, wherein the data collection node is set in the spinning frame equipment and the control terminal of the corresponding spinning frame according to actual needs; collecting relevant data in the production process of the corresponding spinning frame based on the data collection node to obtain corresponding spinning frame information; the spinning frame information includes spinning operation data, spinning energy consumption data and spinning production data;

[0063] The spinning operation data is composed of a plurality of spinning operation indicators, including the spinning frame front roller speed, spindle speed and other indicator data related to the operation of the corresponding spinning frame equipment; the spinning energy consumption data is composed of a plurality of spinning energy consumption indicators, including electric energy indicators, voltage indicators and other indicator data related to the energy consumption of the spinning frame; the spinning production data is composed of a plurality of spinning production indicators, including output indicators, quality indicator parameters, etc.

[0064] Furthermore, data preprocessing is performed on the collected spinning frame information, and the data processing includes data cleaning, data conversion and format conversion. The data cleaning is used to eliminate or repair erroneous data and missing data in the corresponding spinning frame information. The data conversion is used to unify the data units in the corresponding spinning frame information. The format conversion is used to standardize the format of discrete data in the corresponding spinning frame information and perform normalization processing to facilitate subsequent data processing and analysis.

[0065] After the data preprocessing is completed, the spinning frame information after the data preprocessing is stored in a pre-built information database; wherein, the information database also stores corresponding spinning transaction information.

[0066] It should be further explained that, in the specific implementation process, the process of processing the obtained spinning frame information and constructing the corresponding multi-objective optimization function based on the data processing results includes:

[0067] A two-dimensional rectangular coordinate system is constructed with time as the horizontal axis and spinning frame information as the vertical axis, and the obtained rectangular coordinate system is mapped into a corresponding two-dimensional rectangular coordinate system to obtain a corresponding spinning change curve; the spinning change curve includes a spinning operation data curve, a spinning energy consumption data curve, and a spinning production data curve;

[0068] Furthermore, data sampling is performed on the obtained spinning operation data curve, spinning energy consumption data curve, and spinning production data curve, and corresponding operation matrix, energy consumption matrix, and production matrix are constructed based on the obtained data; wherein the column vectors of the operation matrix, energy consumption matrix, and production matrix are the various indicator data in the corresponding operation data curve, spinning energy consumption data curve, and spinning production data curve, respectively; and the row vectors are the collected values ​​corresponding to the corresponding indicator data;

[0069] Taking the operation matrix and energy consumption matrix as examples, the obtained energy consumption matrix and operation matrix are normalized respectively, and the components of the normalized energy consumption matrix and operation matrix are extracted based on the principal component analysis algorithm to obtain the corresponding feature pairs; Where, and Respectively represent the feature u within the a-th feature paira and v a The corresponding eigenvector; ρ a Indicates the unit eigenvector corresponding to the matrix TM when the a-th feature pair is extracted, which is used to represent the feature u a The weight vector of Indicates the feature v within the a-th feature pair a The corresponding weight vector; ; Where A and B represent the energy consumption matrix and the operation matrix respectively; θ represents the maximum eigenvalue corresponding to the matrix TM; where TM=A T BB T A;

[0070] Based on the above-mentioned feature pair acquisition process, and by randomly sampling the spinning operation data curve and the spinning energy consumption data curve to update the operation matrix and the feature matrix, and repeating the feature pair extraction process, a plurality of feature pairs are obtained; a corresponding first feature space matrix is ​​constructed based on the feature pairs; the first feature space matrix retains the original feature space data information, realizes the orthogonality of the feature space, and retains the mapping relationship between the spinning operation data and the spinning energy consumption data in the spinning process to the greatest extent;

[0071] Based on the method for obtaining the first characteristic space matrix, a second characteristic space matrix corresponding to the corresponding spinning operation data and the corresponding spinning production data is obtained;

[0072] Acquire the collected spinning transaction information, and construct a corresponding spinning profit curve based on the information, and then perform data fitting between the corresponding spinning profit curve and the spinning operation curve based on the collection time to obtain a corresponding operation-profit fitting curve;

[0073] Furthermore, a corresponding multi-objective optimization function is constructed based on the operation-profit fitting curve, the first feature space matrix and the second feature space matrix; ; G(E1) and G(E2) refer to the linear functions defined by the corresponding first feature space matrix and the second feature space matrix respectively; F(b) is the definition linear function of the corresponding operation-profit fitting curve; Y is a 3×1 vector containing energy consumption, output and profit; E1, E2 and b represent the functional variables of the linear function defined by the corresponding first feature space matrix, the second feature space matrix and the operation-profit fitting curve respectively; wherein, the functional variables involved in the linear function all belong to the spinning operation data.

[0074] It should be further explained that, in the specific implementation process, the process of constructing the corresponding production parameter tuning model based on the obtained target optimization function includes:

[0075] The network framework of the production parameter tuning model is defined as an improved convolutional neural network, which is used to learn the nonlinear relationship between the spinning operation data and the spinning production data in the spinning frame information;

[0076] The basic architecture of the improved convolutional neural network is an input layer, a multi-scale convolution layer and an output layer;

[0077] The output layer is used to receive input training samples and convert the corresponding training samples into a one-dimensional data vector;

[0078] The multi-scale convolution layer consists of a convolution layer and a pooling layer, and a corresponding improved convolutional neural network is provided with several multi-scale convolution layers; the convolution layer is used to perform a convolution operation on the one-dimensional data vector output by the input layer to obtain the output features corresponding to the corresponding convolution layer; the pooling layer is used to perform a pooling operation on the output features of the convolution layer to retain the main features and reduce the parameter vector; and prevent overfitting;

[0079] A fully connected layer is provided in the output layer, and the fully connected layer is used to map the output features learned by the improved convolutional neural network to the corresponding sample space, and the output layer outputs the results;

[0080] The formula defining the convolution operation is: Where, Indicates the The j-th output feature in the convolution layer within the multi-scale convolution layer; Represents the j-th output feature of the previous layer in the multi-scale convolutional layer; (i.e., The output of the pooling layer within the multi-scale convolutional layer; >0 and is an integer); M j Indicates the The total number of output results of the previous convolution layer in a multi-scale convolution layer; Indicates the The weight matrix of the convolution kernel in the convolution layer; Represents the bias term of the convolutional layer; Represents the convolution operation, f() represents the selected activation function;

[0081] The formula defining the pooling operation is: Where, Indicates the The downsampling parameter of the i-th output feature in the pooling layer within the multi-scale convolutional layer, H() represents the downsampling function; Represents the bias term of the pooling layer; Indicates the The i-th output feature of the pooling layer in the multi-scale convolutional layer; , i and j are all natural numbers, representing the layer indexes of the multi-scale convolutional layer and the pooling layer and convolution layer within the multi-scale convolutional layer, respectively.

[0082] Define the loss function for the production parameter tuning model ;in, The total number of training samples input to the production parameter tuning model, z is the index variable used to traverse all training samples, Generate the output of the parameter tuning model for the zth training example, where Y(z) is the label of the zth training example;

[0083] Initializing the weights in each layer of the improved convolutional neural network until the initialization is completed; constructing a corresponding training data set based on the historical spinning frame information, dividing the training samples in the training data set into a plurality of training batches, and iteratively training the corresponding production parameter tuning model based on the training data set, and recording the value of the corresponding loss function. When the loss function in a plurality of consecutive training batches no longer decreases or changes, the model parameters at this time are saved, and the construction process of the corresponding production parameter tuning model is completed;

[0084] It should be further explained that, in a specific implementation process, the process of initializing the weights in each layer of the improved convolutional neural network includes:

[0085] Obtain the mean square error of the network performance function in the corresponding improved convolutional neural network, and construct the corresponding fitness function based on the obtained multi-objective optimization function ; Where MSE represents the mean square error of the network performance function, and Representing the function weights of the multi-objective optimization function and the mean square error; at the same time, initializing a particle swarm set based on the learning model of the improved convolutional neural network, and defining the number of particles and spatial dimensions corresponding to the corresponding particle swarm set;

[0086] Then, the initial velocity and initial position of each particle in the corresponding particle group are obtained and marked as Sp and Wp respectively; then, the fitness value corresponding to the corresponding particle is obtained based on the fitness function, and the initial position and initial velocity of the corresponding particle are updated based on the fitness value; where Sp∈[Vmin,Vmax]; Vmax and Vmin represent the maximum particle velocity and minimum particle velocity allowed for the corresponding particle respectively; Wp∈[Ld,Ud]; Ld and Ud represent the lower bound and upper bound of the spatial dimension of the corresponding particle respectively;

[0087] The update formula for defining the initial position is:

[0088] Where, represents the particle position of particle p at the k+1th iteration; represents the particle velocity of particle p at the kth iteration; and They represent the individual optimal solution and the global optimal solution at the kth iteration respectively; r1 and r2 are constants, and both r1 and r2 are ∈ [0,1];

[0089] ω represents the adaptive inertia weight; Where, ω max and ω min Respectively represent the maximum and minimum values ​​of the preset inertia weight; k and k max Represent the current number of iterations and the maximum number of iterations respectively; α and ε represent the speed coefficients, which are fixed constants;

[0090] c1 and c2 represent learning factors; where ; Where c10, c20 and c12, c22 represent the initial and final values ​​of learning factors c1 and c2 respectively;

[0091] Furthermore, the initial position and initial velocity of each particle in the corresponding particle swarm are scaled and updated based on the fitness function until the pre-set iterative stopping condition is met, and the corresponding global optimal solution is obtained; then, the obtained global optimal solution is mapped to the corresponding improved convolutional neural network, and then a new fitness value is obtained through the positive feedback output of the improved convolutional neural network, and the weights of the improved convolutional neural network are updated based on it; wherein, the iterative stopping condition is that the maximum number of iterations is reached or the individual optimal solution obtained is inferior to the global optimal solution for P1 consecutive times, and P1 is a fixed constant.

[0092] It should be further explained that, in a specific implementation process, the process of controlling the parameters of the target spinning frame in combination with the production parameter tuning model includes:

[0093] Based on the information database, real-time spinning frame information corresponding to the target spinning frame is obtained, and the real-time spinning frame information is input into the corresponding production parameter tuning model to obtain a corresponding model output result; the model output result is the optimal data after optimizing the target spinning operation data;

[0094] Feeding back the obtained model output result to the control terminal of the corresponding spinning frame; further, the control terminal adjusts the spinning operation parameters of the corresponding spinning frame based on the model output result until the actual operation parameters are the same as the model output result;

[0095] Taking the front roller speed of a spinning frame as an example, the control terminal obtains the corresponding optimal front roller speed based on the output result of the model;

[0096] The control terminal generates a corresponding control signal and adjusts the corresponding front roller speed based on the control signal; at the same time, the real-time speed of the front roller of the current spinning machine is collected in real time, and the deviation is calculated between it and the optimal front roller speed to obtain a corresponding deviation value, and the corresponding control signal is adjusted based on the deviation value until the real-time speed of the front roller meets the model output result; at the same time, the control terminal obtains the correlation between the corresponding front roller speed and other operating parameters, and based on the correlation and in combination with the optimal front roller speed, proportional parameter adjustment is performed on other operating parameters.

[0097] The present invention closely links the steps of data collection, data processing, multi-objective optimization function construction, production parameter tuning model construction to parameter control, which can realize intelligent regulation of the rotation speed of the front roller of the spinning frame, thereby improving the production efficiency and product quality of the spinning frame; and has broad application prospects, which is of great significance to the intelligent transformation and upgrading of the textile industry.

[0098] Example 2

[0099] See also Figure 2 As shown, for the parts not described in detail in this embodiment, please refer to the description of embodiment 1. A spinning frame front roller speed intelligent control system is provided, comprising:

[0100] A data acquisition module is used to collect data on the production process of the target spinning frame, obtain corresponding spinning frame information and store it; the spinning frame information includes operation data and production data;

[0101] A data processing module is used to process the obtained spinning frame information and construct a corresponding multi-objective optimization function based on the data processing results;

[0102] A data control module is used to construct a corresponding production parameter tuning model based on the obtained target optimization function; and to control the parameters of the target spinning frame in combination with the production parameter tuning model;

[0103] The modules are connected via wired and / or wireless means to achieve data transmission between modules.

[0104] Example 3

[0105] This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the operation mode of the above-mentioned intelligent control system and method for the front roller speed of a spinning machine is realized.

[0106] Since the electronic device introduced in this embodiment is an electronic device used to implement a spinning frame front roller speed intelligent control system and method in the embodiment of this application, based on the spinning frame front roller speed intelligent control system and method introduced in the embodiment of this application, technical personnel in this field can understand the specific implementation of the electronic device of this embodiment and its various variations, so how the electronic device implements the method in the embodiment of this application will not be described in detail here. As long as technical personnel in this field implement the electronic device used in the spinning frame front roller speed intelligent control system and method in the embodiment of this application, it falls within the scope of protection of this application.

[0107] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0108] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for users of ordinary skill in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for intelligently controlling the rotation speed of the front roller of a spinning frame, characterized in that: include: Step 1: collecting data on the production process of the target spinning frame, obtaining and storing corresponding spinning frame information; the spinning frame information includes operation data and production data; Step 2: Process the obtained spinning frame information and construct the corresponding multi-objective optimization function based on the data processing results; The process of data processing the obtained spinning frame information includes: A two-dimensional rectangular coordinate system is constructed with time as the horizontal axis and spinning frame information as the vertical axis, and the obtained rectangular coordinate system is mapped into a corresponding two-dimensional rectangular coordinate system to obtain a corresponding spinning change curve; the spinning change curve includes a spinning operation data curve, a spinning energy consumption data curve, and a spinning production data curve; The obtained spinning operation data curve, spinning energy consumption data curve and production data curve are sampled respectively, and the corresponding operation matrix, energy consumption matrix and production matrix are constructed based on the obtained data; Normalize the obtained energy consumption matrix and operation matrix respectively, and extract the components of the normalized energy consumption matrix and operation matrix based on the principal component analysis algorithm to obtain the corresponding feature pairs; By randomly sampling the spinning operation data curve and the spinning energy consumption data curve to update the operation matrix and the feature matrix, and repeating the feature pair extraction process to obtain a plurality of feature pairs; constructing a corresponding first feature space matrix based on the feature pairs; Based on the method for obtaining the first characteristic space matrix, a second characteristic space matrix corresponding to the corresponding spinning operation data and the corresponding spinning production data is obtained; Acquire the collected spinning transaction information, and construct a corresponding spinning profit curve based on the information, perform data fitting on the corresponding spinning profit curve and the spinning operation curve based on the collection time, and obtain a corresponding operation-profit fitting curve; Constructing a corresponding multi-objective optimization function based on the operation-profit fitting curve, the first feature space matrix and the second feature space matrix; Step three: constructing a corresponding production parameter tuning model based on the obtained target optimization function; and performing parameter control on the target spinning frame in combination with the production parameter tuning model.

2. The method for intelligently controlling the front roller speed of a spinning frame according to claim 1, characterized in that: The process of collecting data from the production process of the target spinning frame, obtaining corresponding spinning frame information and storing it includes: Setting a data collection node, collecting relevant data in the production process of the corresponding spinning frame based on the data collection node, and obtaining corresponding spinning frame information; the spinning frame information includes spinning operation data, spinning energy consumption data and spinning production data; The collected spinning frame information is subjected to data preprocessing; after the data preprocessing is completed, the spinning frame information after the data preprocessing is stored in a pre-built information database; wherein, the information database also stores corresponding spinning transaction information.

3. The method for intelligently controlling the front roller speed of a spinning frame according to claim 1, characterized in that: The formula for obtaining feature pairs is: Where, and Respectively represent the feature u within the a-th feature pair a and v a The corresponding eigenvector; ρ a represents the unit eigenvector corresponding to the matrix TM when the ath feature pair is extracted; γ a Indicates the feature v within the a-th feature pair a The corresponding weight vector; ; Where A and B represent the energy consumption matrix and the operation matrix respectively; θ represents the maximum eigenvalue corresponding to the matrix TM; where TM=A T BB T A; Multi-objective optimization function Y=[G(E1),G(E2),F(b)] T ; G(E1) and G(E2) refer to the linear functions defined by the corresponding first feature space matrix and the second feature space matrix respectively; F(b) is the defining linear function of the corresponding operation-profit fitting curve; E1, E2 and b respectively represent the functional independent variables of the linear function defined by the corresponding first feature space matrix, the second feature space matrix and the operation-profit fitting curve.

4. The method for intelligently controlling the front roller speed of a spinning frame according to claim 1, characterized in that: The process of building a corresponding production parameter tuning model based on the obtained target optimization function includes: The network framework of the production parameter tuning model is defined as an improved convolutional neural network; the basic architecture of the improved convolutional neural network is an input layer, a multi-scale convolution layer, and an output layer; The output layer is used to receive input training samples and convert the corresponding training samples into a one-dimensional data vector; The multi-scale convolution layer consists of a convolution layer and a pooling layer, and a plurality of multi-scale convolution layers are provided in the corresponding improved convolutional neural network; the convolution layer is used to perform a convolution operation on the one-dimensional data vector output by the input layer to obtain the output features corresponding to the corresponding convolution layer; the pooling layer is used to perform a pooling operation on the output features of the convolution layer; A fully connected layer is provided in the output layer, and the fully connected layer is used to map the output features learned by the improved convolutional neural network to the corresponding sample space, and the output layer outputs the results; Define the loss function ; Where N' is the total number of training samples input into the production parameter tuning model, and z is an index variable used to traverse all training samples. Generate the output of the parameter tuning model for the zth training example, where Y(z) is the label of the zth training example; Based on the multi-objective optimization function, the weights in each layer of the improved convolutional neural network are initialized and set until the initialization is completed; based on the historical spinning frame information, a corresponding training data set is constructed, and the training samples in the training data set are divided into several training batches, and based on the corresponding production parameter tuning model, iterative training is performed, and the value of the corresponding loss function is recorded. When the loss function in several consecutive training batches no longer decreases or changes, the model parameters at this time are saved, and the construction process of the corresponding production parameter tuning model is completed.

5. The method for intelligently controlling the front roller speed of a spinning frame according to claim 4, characterized in that: The formula defining the convolution operation is: Where, Indicates the The first convolution layer in the multi-scale convolution layer output features; Indicates the first layer of the previous convolution layer in the multi-scale convolution layer. output features; Indicates the The total number of output results of the previous convolution layer in a multi-scale convolution layer; Indicates the The weight matrix of the convolution kernel in the convolution layer; Represents the bias term of the convolutional layer; Represents the convolution operation, f() represents the selected activation function; The formula defining the pooling operation is: Where, represents the pooling layer in the multi-scale convolutional layer The downsampling parameters of the output features, H() represents the downsampling function; Represents the bias term of the pooling layer; Indicates the The first pooling layer in the multi-scale convolutional layer output features; 、 and Both are natural numbers, representing the layer indexes of the multi-scale convolution layer and the pooling layer and convolution layer within the multi-scale convolution layer.

6. The method for intelligently controlling the front roller speed of a spinning frame according to claim 4, characterized in that: The process of initializing the weights in each layer of the improved convolutional neural network includes: Obtain the mean square error of the network performance function in the corresponding improved convolutional neural network, and combine the obtained multi-objective optimization function to construct the corresponding fitness function Fit=φ1×Y+φ2×MSE; where MSE represents the mean square error of the network performance function, and φ1 and φ2 represent the function weights of the multi-objective optimization function and the mean square error; At the same time, a particle swarm set is initialized based on the learning mode of the improved convolutional neural network, and the number of particles and the spatial dimension corresponding to the corresponding particle swarm set are defined; Obtaining the initial velocity and initial position of each particle in the corresponding particle group, marking them as Sp and Wp respectively; obtaining the fitness value corresponding to the corresponding particle based on the fitness function, and updating the initial position and initial velocity of the corresponding particle based on the fitness value; Based on the fitness function, the initial position and initial velocity of each particle in the corresponding particle swarm are continuously updated until a preset iteration stop condition is met, thereby obtaining the corresponding global optimal solution; the obtained global optimal solution is mapped to the corresponding improved convolutional neural network, and then a new fitness value is obtained through the positive feedback output of the improved convolutional neural network, and the weights of the improved convolutional neural network are updated based on the new fitness value.

7. The method for intelligently controlling the front roller speed of a spinning frame according to claim 6, characterized in that: The update formula for defining the initial position is: Where, represents the particle position of particle p at the k+1th iteration; represents the particle velocity of particle p at the kth iteration; and They represent the individual optimal solution and the global optimal solution at the kth iteration respectively; r1 and r2 are constants, and both r1 and r2 are in [0,1] ; ω represents the adaptive inertia weight; Where, ω max and ω min Respectively represent the maximum and minimum values ​​of the preset inertia weight; k and k max Represent the current number of iterations and the maximum number of iterations respectively; α and ε represent the speed coefficients, which are fixed constants; c1 and c2 represent learning factors; where ; Where c10, c20 and c12, c22 represent the initial and final values ​​of the learning factors c1 and c2 respectively.

8. The method for intelligently controlling the front roller speed of a spinning frame according to claim 6, characterized in that: The process of controlling the parameters of the target spinning frame in combination with the production parameter tuning model includes: Based on the information database, real-time spinning frame information corresponding to the target spinning frame is obtained, and the real-time spinning frame information is input into the corresponding production parameter tuning model to obtain a corresponding model output result; the model output result is the optimal data after optimizing the target spinning operation data; The obtained model output result is fed back to the control terminal of the corresponding spinning frame; the control terminal adjusts the spinning operation parameters of the corresponding spinning frame based on the model output result until the actual operation parameters are the same as the model output result.

9. A spinning frame front roller speed intelligent control system, which is used to implement the spinning frame front roller speed intelligent control method according to any one of claims 1 to 8, comprising: The data acquisition module is used to collect data on the production process of the target spinning frame, obtain corresponding spinning frame information and store it; The spinning frame information includes operation data and production data; A data processing module is used to process the obtained spinning frame information and construct a corresponding multi-objective optimization function based on the data processing results; The data control module is used to construct a corresponding production parameter tuning model based on the obtained target optimization function; and to control the parameters of the target spinning frame in combination with the production parameter tuning model.

Images