Remote control method and system for textile equipment

Abstract

The invention discloses a remote control method and system for textile equipment, relates to the technical field of remote control of the textile equipment, and aims to solve the technical problem that the equipment is difficult to maintain in an optimal operation state all the time due to subjectivity and instability of traditional manual control, and the method comprises the following steps: S1, state parameter processing; s1.1, data acquisition: acquiring original state data from different key parts of the textile equipment by using a distributed sensor network, the sensors including but not limited to a tension sensor, a speed sensor, a temperature sensor and the like, and ensuring the comprehensiveness and accuracy of the data, the collected data cover multi-dimensional information such as a mechanical operation state, an electrical parameter and an environmental parameter of the equipment; and S1.2, characteristic value calculation: carrying out screening and normalization processing on the collected original state data, and removing abnormal data points. The device has the advantage that the device control instruction can be accurately and finely adjusted.

Description

Technical Field

[0001] The present invention relates to the technical field of remote control of textile equipment, and more specifically, to a remote control method and system for textile equipment. Background Art

[0002] In the textile industry, traditional textile equipment control methods have many drawbacks. In the past, they mainly relied on manual regular inspections of equipment, relying on the operator's experience and simple instrument readings to judge the equipment status and perform control operations. This method seems to be inadequate in the face of the increasingly complex operating environment and high-precision production requirements of modern textile equipment. On the one hand, it is difficult for humans to obtain the status parameters of each key part of the equipment in a comprehensive, timely and accurate manner, which often leads to a lag in the detection of potential problems in the equipment. On the other hand, the subjectivity and instability of manual control make it difficult for the equipment to always maintain the best operating state, increasing the energy consumption and maintenance cost of the equipment. In view of this, we propose a remote control method and system for textile equipment. Summary of the invention

[0003] The purpose of the present invention is to provide a remote control method and system for textile equipment to solve the technical problem that the subjectivity and instability of manual control make it difficult for the equipment to always maintain an optimal operating state.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: a remote control method for textile equipment, comprising the following steps: S1, state parameter processing; S1.1. Data acquisition: using a distributed sensor network to collect raw status data from different key parts of textile equipment. These sensors include but are not limited to tension sensors, speed sensors, temperature sensors, etc., to ensure the comprehensiveness and accuracy of the data. The collected data covers multi-dimensional information such as the mechanical operating status, electrical parameters, and environmental parameters of the equipment; S1.2, characteristic value calculation, screening and normalizing the collected raw state data, removing abnormal data points, and converting the raw data into loom state characteristic values ​​using statistical analysis methods and machine learning algorithms; S1.3, comprehensive parameter generation, based on the loom state eigenvalues, using data fusion technology, deeply fuses the eigenvalues ​​of different sources and types to generate comprehensive state parameters, and constructs the textile machine state vector based on the comprehensive state parameters; S2, feature vector extraction; S2.1, network classification, based on the dynamic characteristics of the comprehensive state parameters and historical data patterns, a clustering algorithm is used to adaptively classify lightweight convolutional neural networks; S2.2, channel determination, according to the comprehensive state parameter classification, determine the number of channels for the convolution operation; S2.3, feature extraction, using the autoencoder in deep learning to pre-encode the textile machine state vector, and then input the pre-encoding result into the lightweight convolutional neural network, and extract the state value importance information according to the textile machine state vector and the adjusted number of convolution operation channels through the adaptive convolution kernel, and output the convolution feature vector; S3, instruction set input judgment; S3.1, coefficient matrix generation, integrating the historical operation records and operation log information of the equipment to generate the convolution coefficient matrix at the convolution moment; S3.2, result judgment, input the convolution coefficient matrix into the trained LSTM neural network to determine whether the instructions in the control instruction set achieve the expected results; S3.3, instruction adjustment, when the expected result is achieved, a judgment result that meets the preset rules is generated, and the control instruction is fine-tuned according to the judgment result; S4, equipment control operation; S4.1, coefficient determination, according to the importance of the convolution feature value in the input convolution feature vector and the real-time production task requirements of the equipment, the convolution coefficient of the convolution kernel at the current convolution moment is determined by a fuzzy decision algorithm; S4.2, generating a result set, convolving the textile machine state vector according to the convolution coefficient to generate a convolution result set; S4.3, device operation, comparing the convolution result set with the control instruction set, and controlling the device based on the comparison result.

[0005] Preferably, in S1, the calculation method for generating the comprehensive state parameter is: assuming that the loom state characteristic value set is , the comprehensive state vector The calculation formula is: ; in, is the eigenvalue determined by the fuzzy logic algorithm The weight is dynamically adjusted according to the historical data and current operating status of the device and meets .

[0006] Preferably, in S2, the specific method of determining the number of channels of the convolution operation according to the obtained comprehensive state parameter classification is: When the comprehensive state parameter is a low comprehensive state parameter, the number of channels of the convolution operation is 1; When the comprehensive state parameter is the medium comprehensive state parameter, the number of channels of the convolution operation is 3; When the comprehensive state parameter is a high comprehensive state parameter, the number of channels of the convolution operation is 7.

[0007] Preferably, in S2, the state vector of the textile machine is , the convolution kernel is , the number of convolution operation channels is , the convolution feature vector , for the Convolution eigenvalues , and its calculation formula is: ; in, and , Represents the state vector of the textile machine The elements, Represents the convolution kernel The elements in .

[0008] Preferably, in S3, the convolution coefficient matrix is ​​calculated as follows: Let the convolution feature vector be , for the Convolution moments, the convolution coefficient matrix elements The calculation formula is: ; in, is the forgetting factor, and its value range is , It is the convolution coefficient matrix element of the previous moment.

[0009] Preferably, in S3, the convolution coefficient matrix is ​​input into the trained LSTM neural network, and the method for judging whether the instructions in the control instruction set achieve the expected results is as follows: The sum of the absolute values ​​of the differences between the convolution feature vector and each element of the third control instruction vector is calculated to determine whether the control instruction needs to be changed. The convolution feature vector is , the third control instruction vector is , judgment index for: ; in, Represents the third control instruction vector No. elements, Represents the convolution feature vector No. If the expected result is achieved, a judgment result that meets the preset rules is generated, and the textile equipment is controlled according to the judgment result; if the expected result is not achieved, the input is readjusted.

[0010] Preferably, in S4, the formula for determining the convolution coefficient of the convolution kernel at the current convolution moment by the fuzzy decision algorithm is: ; in, is the adjustment factor, and its value range is , is the normalized weight of the convolution feature vector element, is the current state of the device; According to the convolution coefficient, the state vector of the convolution textile machine is convolved, and a convolution result set is generated, wherein the convolution result set includes a plurality of the convolution eigenvalues, and the convolution result set The calculation formula is: ; in, represents the first elements.

[0011] Preferably, in S4, the method of comparing the convolution result set with the control instruction set is: Set control instruction set , the convolution result set is , define a matching function To measure the degree of match between the two, the calculation formula is: ; in, Indicates the control instruction set elements, Represents the convolution result set elements, Indicates the number of elements in the control instruction set and the convolution result set, which is used for normalization when calculating the matching degree; when When the two are matched, the textile equipment is controlled to work according to the command; when When the two are judged to be mismatched, the textile equipment is controlled to stop working and an error code is output.

[0012] Preferably, the control instruction set includes a start instruction, a stop instruction, a reset instruction and an error correction instruction.

[0013] A remote control system for textile equipment, comprising: Data acquisition module, used to collect original status data from different key parts of textile equipment; The data processing module is connected to the data acquisition module and is used to process the collected data, remove abnormal data points, convert the original data into loom state characteristic values ​​using statistical analysis methods and machine learning algorithms, generate comprehensive state parameters using data fusion technology, and construct the textile machine state vector based on the comprehensive state parameters; The feature extraction module uses a lightweight convolutional neural network, uses a clustering algorithm to adjust the network structure and the number of convolution channels, and uses an autoencoder and an adaptive convolution kernel to extract the convolution feature vector; The instruction judgment module uses the trained LSTM neural network to judge whether the control instruction meets the expected result based on the generated convolution coefficient matrix; The device control module determines the convolution coefficient through fuzzy decision-making according to the instruction judgment result, generates a convolution result set and compares it with the control instruction set, and controls the operating status of the device according to the comparison result and the matching function.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention can accurately generate the equipment state vector by collecting, processing and analyzing the state parameters of textile equipment, and use advanced neural network algorithms to judge the effect of control instructions, thereby realizing accurate fine-tuning of equipment control instructions. Compared with traditional control methods, it can effectively reduce problems such as unstable equipment operation and product quality fluctuations caused by control deviations, ensure that textile equipment always maintains high-precision operation under complex working conditions, and is conducive to improving product quality and pass rate.

[0015] 2. The present invention uses a clustering algorithm to classify the adaptability of lightweight convolutional neural networks and dynamically determine the number of convolution operation channels based on comprehensive state parameters. This allows the system to automatically adjust the control strategy according to the different operating states of the textile equipment. Whether it is equipment startup, stable operation, or responding to sudden changes in operating conditions, it can quickly adapt and give reasonable control plans, thereby enhancing the control adaptability to various operating scenarios of textile equipment.

[0016] 3. The present invention also determines the convolution coefficient and generates a result set based on the fuzzy decision algorithm in the equipment control operation link, and realizes the rapid response control of the equipment by efficiently comparing the result set with the control instruction set. Compared with the traditional manual inspection and manual adjustment control method, the control decision time is greatly shortened, the real-time and automation of equipment control are improved, and the overall production efficiency is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the process of the present invention; Figure 2 Schematic diagram of the system framework of the present invention. DETAILED DESCRIPTION

[0018] In order to facilitate those skilled in the art to understand the technical solution of the present invention, the technical solution of the present invention is further described in conjunction with the accompanying drawings of the specification.

[0019] Embodiment 1, as Figure 1 As shown, the present invention provides a remote control method for a textile device, comprising the following steps: S1, state parameter processing; S1.1. Data acquisition: using a distributed sensor network to collect raw status data from different key parts of textile equipment. These sensors include but are not limited to tension sensors, speed sensors, temperature sensors, etc., to ensure the comprehensiveness and accuracy of the data. The collected data covers multi-dimensional information such as the mechanical operating status, electrical parameters, and environmental parameters of the equipment; S1.2, characteristic value calculation, screening and normalizing the collected raw state data, removing abnormal data points, and converting the raw data into loom state characteristic values ​​using statistical analysis methods and machine learning algorithms; S1.3, comprehensive parameter generation, based on the loom state eigenvalues, using data fusion technology, deeply fuses the eigenvalues ​​of different sources and types to generate comprehensive state parameters, and constructs the textile machine state vector based on the comprehensive state parameters; S2, feature vector extraction; S2.1, network classification, based on the dynamic characteristics of the comprehensive state parameters and historical data patterns, a clustering algorithm is used to adaptively classify lightweight convolutional neural networks; S2.2, channel determination, according to the comprehensive state parameter classification, determine the number of channels for the convolution operation; S2.3, feature extraction, using the autoencoder in deep learning to pre-encode the textile machine state vector, and then input the pre-encoding result into the lightweight convolutional neural network, and extract the state value importance information according to the textile machine state vector and the adjusted number of convolution operation channels through the adaptive convolution kernel, and output the convolution feature vector; S3, instruction set input judgment; S3.1, coefficient matrix generation, integrating the historical operation records and operation log information of the equipment to generate the convolution coefficient matrix at the convolution moment; S3.2, result judgment, input the convolution coefficient matrix into the trained LSTM neural network to determine whether the instructions in the control instruction set achieve the expected results; S3.3, instruction adjustment, when the expected result is achieved, a judgment result that meets the preset rules is generated, and the control instruction is fine-tuned according to the judgment result; S4, equipment control operation; S4.1, coefficient determination, according to the importance of the convolution feature value in the input convolution feature vector and the real-time production task requirements of the equipment, the convolution coefficient of the convolution kernel at the current convolution moment is determined by a fuzzy decision algorithm; S4.2, generating a result set, convolving the textile machine state vector according to the convolution coefficient, and generating a convolution result set; S4.3, device operation, comparing the convolution result set with the control instruction set, and controlling the device based on the comparison result.

[0020] In the embodiment of the present invention, in S1, the calculation method for generating the comprehensive state parameter is as follows: assuming that the loom state feature value set is , the comprehensive state vector The calculation formula is: ; in, is the eigenvalue determined by the fuzzy logic algorithm The weight is dynamically adjusted according to the historical data and current operating status of the device and meets .

[0021] In the embodiment of the present invention, in S2, the specific method of determining the number of channels of the convolution operation according to the obtained comprehensive state parameter classification is: When the comprehensive state parameter is a low comprehensive state parameter, the number of channels of the convolution operation is 1; When the comprehensive state parameter is the medium comprehensive state parameter, the number of channels of the convolution operation is 3; When the comprehensive state parameter is a high comprehensive state parameter, the number of channels of the convolution operation is 7; In S2, the state vector of the textile machine is assumed to be , the convolution kernel is , the number of convolution operation channels is , the convolution feature vector , for the Convolution eigenvalues , and its calculation formula is: ; in, and , Represents the state vector of the textile machine The elements, Represents the convolution kernel The elements in .

[0022] In an embodiment of the present invention, in S3, the convolution coefficient matrix is ​​calculated as follows: Let the convolution feature vector be , for the Convolution moments, the convolution coefficient matrix elements The calculation formula is: ; in, is the forgetting factor, and its value range is , It is the convolution coefficient matrix element of the previous moment; In S3, the convolution coefficient matrix is ​​input into the trained LSTM neural network. The method for judging whether the instructions in the control instruction set achieve the expected results is as follows: The sum of the absolute values ​​of the differences between the convolution feature vector and each element of the third control instruction vector is calculated to determine whether the control instruction needs to be changed. The convolution feature vector is , the third control instruction vector is , judgment index for: ; The third control instruction is a set of pre-set instructions for comparison with the convolution feature vector, which represents the expected device operation instruction state. Its purpose is to determine whether the current control instruction needs to be adjusted or modified by comparing with the current convolution feature vector to ensure that the control of the device meets the expected goals.

[0023] in, Represents the third control instruction vector No. elements, Represents the convolution feature vector No. If the expected result is achieved, a judgment result that meets the preset rules is generated, and the textile equipment is controlled according to the judgment result; if the expected result is not achieved, the input is readjusted.

[0024] In an embodiment of the present invention, in S4, the formula for determining the convolution coefficient of the convolution kernel at the current convolution moment by the fuzzy decision algorithm is: ; in, is the adjustment factor, and its value range is , is the normalized weight of the convolution feature vector element, is the current state of the device; According to the convolution coefficient and the state vector of the convolution textile machine, a convolution result set is generated, and the convolution result set includes multiple convolution eigenvalues ​​and the convolution result set. The calculation formula is: ; in, represents the first elements; In S4, the method of comparing the convolution result set with the control instruction set is: Set control instruction set , the convolution result set is , define a matching function To measure the degree of match between the two, the calculation formula is: ; in, Indicates the control instruction set elements, Represents the convolution result set elements, Indicates the number of elements in the control instruction set and the convolution result set, which is used for normalization when calculating the matching degree; when When the two are matched, the textile equipment is controlled to work according to the command; when When the two are judged to be mismatched, the textile equipment is controlled to stop working and an error code is output.

[0025] In an embodiment of the present invention, the control instruction set includes a start instruction, a stop instruction, a reset instruction and an error correction instruction.

[0026] Embodiment 2, as Figure 2 As shown, the present invention provides a remote control system for a textile device, comprising: Data acquisition module, used to collect original status data from different key parts of textile equipment; The data processing module is connected to the data acquisition module and is used to process the collected data, remove abnormal data points, convert the original data into loom state characteristic values ​​using statistical analysis methods and machine learning algorithms, generate comprehensive state parameters using data fusion technology, and construct the textile machine state vector based on the comprehensive state parameters; The feature extraction module uses a lightweight convolutional neural network, uses a clustering algorithm to adjust the network structure and the number of convolution channels, and uses an autoencoder and an adaptive convolution kernel to extract the convolution feature vector; The instruction judgment module uses the trained LSTM neural network to judge whether the control instruction meets the expected result based on the generated convolution coefficient matrix; The device control module determines the convolution coefficient through fuzzy decision-making according to the instruction judgment result, generates a convolution result set and compares it with the control instruction set, and controls the operating status of the device according to the comparison result and the matching function.

[0027] Example 3: The situation where the control instruction set matches the convolution result set: Assume the convolution result set , control instruction set , the preset matching value ,; According to the matching function , the calculation process is as follows: for : ; for : ; for : ; for : ; but ; because , so it is considered that the control instruction set and the convolution result set match, and the textile equipment is controlled to work according to the command.

[0028] Example 4: The situation where the control instruction set does not match the convolution result set: Assume the convolution result set , control instruction set , the preset matching value ,; According to the matching function , the calculation process is as follows: for : ; for : ; for : ; for : ; but ; because , so it is considered that the control instruction set and the convolution result set do not match, and the textile equipment is controlled to stop working.

[0029] The embodiments of the present invention disclose preferred embodiments, but are not limited thereto. A person skilled in the art can easily understand the spirit of the present invention based on the above embodiments and make different extensions and changes. However, as long as they do not deviate from the spirit of the present invention, they are all within the protection scope of the present invention.

Claims

1. A remote control method for textile equipment, characterized in that: The following steps are involved: S1, state parameter processing; S1.

1. Data acquisition: using a distributed sensor network to collect raw status data from different key parts of textile equipment. These sensors include but are not limited to tension sensors, speed sensors, temperature sensors, etc., to ensure the comprehensiveness and accuracy of the data. The collected data covers multi-dimensional information such as the mechanical operating status, electrical parameters, and environmental parameters of the equipment; S1.2, characteristic value calculation, screening and normalizing the collected raw state data, removing abnormal data points, and converting the raw data into loom state characteristic values ​​using statistical analysis methods and machine learning algorithms; S1.3, comprehensive parameter generation, based on the loom state eigenvalues, using data fusion technology, deeply fuses the eigenvalues ​​of different sources and types to generate comprehensive state parameters, and constructs the textile machine state vector based on the comprehensive state parameters; S2, feature vector extraction; S2.1, network classification, based on the dynamic characteristics of the comprehensive state parameters and historical data patterns, a clustering algorithm is used to adaptively classify lightweight convolutional neural networks; S2.2, channel determination, according to the comprehensive state parameter classification, determine the number of channels for the convolution operation; S2.3, feature extraction, using the autoencoder in deep learning to pre-encode the textile machine state vector, and then input the pre-encoding result into the lightweight convolutional neural network, and extract the state value importance information according to the textile machine state vector and the adjusted number of convolution operation channels through the adaptive convolution kernel, and output the convolution feature vector; S3, instruction set input judgment; S3.1, coefficient matrix generation, integrating the historical operation records and operation log information of the equipment to generate the convolution coefficient matrix at the convolution moment; S3.2, result judgment, input the convolution coefficient matrix into the trained LSTM neural network to determine whether the instructions in the control instruction set achieve the expected results; S3.3, instruction adjustment, when the expected result is achieved, a judgment result that meets the preset rules is generated, and the control instruction is fine-tuned according to the judgment result; S4, equipment control operation; S4.1, coefficient determination, according to the importance of the convolution feature value in the input convolution feature vector and the real-time production task requirements of the equipment, the convolution coefficient of the convolution kernel at the current convolution moment is determined by a fuzzy decision algorithm; S4.2, generating a result set, convolving the textile machine state vector according to the convolution coefficient to generate a convolution result set; S4.3, device operation, comparing the convolution result set with the control instruction set, and controlling the device based on the comparison result.

2. A remote control method for textile equipment according to claim 1, characterized in that: In S1, the calculation method for generating the comprehensive state parameter is as follows: assuming that the set of loom state characteristic values ​​is , the comprehensive state vector The calculation formula is: ； in, is the eigenvalue determined by the fuzzy logic algorithm The weight is dynamically adjusted according to the historical data and current operating status of the device and meets .

3. A remote control method for textile equipment according to claim 2, characterized in that: In S2, according to the obtained comprehensive state parameter classification, the specific method of determining the number of channels of the convolution operation is: When the comprehensive state parameter is a low comprehensive state parameter, the number of channels of the convolution operation is 1; When the comprehensive state parameter is the medium comprehensive state parameter, the number of channels of the convolution operation is 3; When the comprehensive state parameter is a high comprehensive state parameter, the number of channels of the convolution operation is 7.

4. A remote control method for textile equipment according to claim 3, characterized in that: In S2, the state vector of the textile machine is assumed to be , the convolution kernel is , the number of convolution operation channels is , the convolution feature vector , for the Convolution eigenvalues , and its calculation formula is: ； in, and , Represents the state vector of the textile machine The elements, Represents the convolution kernel The elements in .

5. A remote control method for textile equipment according to claim 4, characterized in that: In S3, the convolution coefficient matrix is ​​calculated as follows: Let the convolution feature vector be , for the Convolution moments, the convolution coefficient matrix elements The calculation formula is: ； in, is the forgetting factor, and its value range is , It is the convolution coefficient matrix element of the previous moment.

6. A remote control method for textile equipment according to claim 5, characterized in that: In S3, the convolution coefficient matrix is ​​input into the trained LSTM neural network, and the method for judging whether the instructions in the control instruction set achieve the expected results is as follows: The sum of the absolute values ​​of the differences between the convolution feature vector and each element of the third control instruction vector is calculated to determine whether the control instruction needs to be changed. The convolution feature vector is , the third control instruction vector is , judgment index for: ； in, Represents the third control instruction vector No. elements, Represents the convolution feature vector No. If the expected result is achieved, a judgment result that meets the preset rules is generated, and the textile equipment is controlled according to the judgment result; if the expected result is not achieved, the input is readjusted.

7. A remote control method for textile equipment according to claim 6, characterized in that: In S4, the formula for determining the convolution coefficient of the convolution kernel at the current convolution moment by the fuzzy decision algorithm is: ； in, is the adjustment factor, and its value range is , is the normalized weight of the convolution feature vector element, is the current state of the device; According to the convolution coefficient, the state vector of the convolution textile machine is convolved, and a convolution result set is generated, wherein the convolution result set includes a plurality of the convolution eigenvalues, and the convolution result set The calculation formula is: ； in, represents the first elements.

8. A remote control method for textile equipment according to claim 7, characterized in that: In S4, the method of comparing the convolution result set with the control instruction set is: Set control instruction set , the convolution result set is , define a matching function To measure the degree of match between the two, the calculation formula is: ； in, Indicates the control instruction set elements, Represents the convolution result set elements, Indicates the number of elements in the control instruction set and the convolution result set, which is used for normalization when calculating the matching degree; when When the two are matched, the textile equipment is controlled to work according to the command; when When the two are judged to be mismatched, the textile equipment is controlled to stop working and an error code is output.

9. A remote control method for textile equipment according to claim 8, characterized in that: The control instruction set includes a start instruction, a stop instruction, a reset instruction and an error correction instruction.

10. A remote control system for textile equipment according to claim 9, characterized in that: include: Data acquisition module, used to collect original status data from different key parts of textile equipment; The data processing module is connected to the data acquisition module and is used to process the collected data, remove abnormal data points, convert the original data into loom state characteristic values ​​using statistical analysis methods and machine learning algorithms, generate comprehensive state parameters using data fusion technology, and construct the textile machine state vector based on the comprehensive state parameters; The feature extraction module uses a lightweight convolutional neural network, uses a clustering algorithm to adjust the network structure and the number of convolution channels, and uses an autoencoder and an adaptive convolution kernel to extract the convolution feature vector; The instruction judgment module uses the trained LSTM neural network to judge whether the control instruction meets the expected result based on the generated convolution coefficient matrix; The device control module determines the convolution coefficient through fuzzy decision-making according to the instruction judgment result, generates a convolution result set and compares it with the control instruction set, and controls the operating status of the device according to the comparison result and the matching function.

Images