Loom data monitoring method and system, computer equipment and storage medium

Abstract

The invention discloses a loom data monitoring method and system, computer equipment and a storage medium. The method comprises the following steps of S10, collecting operation data of a loom through a distributed sensor network; s20, optimizing the operation data of the weaving machine by utilizing edge computing equipment to obtain optimized operation data of the weaving machine; s30, the edge computing device transmits the optimized operation data of the weaving machine to a cloud server; s40, the cloud server analyzes the optimized operation data of the weaving machine based on a weaving machine operation state evaluation model to obtain an operation state evaluation result of the weaving machine; and S50, the cloud server performs predictive maintenance on the weaving machine based on the operation state evaluation result of the weaving machine. The technical problems of poor real-time performance, low maintenance efficiency and weak prediction performance of loom monitoring in the prior art can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of textile machinery, and in particular to a loom data monitoring method, system, computer equipment and storage medium. Background Art

[0002] Looms are important production equipment in the textile industry, and their operating status directly affects product quality and production efficiency. Traditional loom monitoring methods mainly rely on manual inspections and regular maintenance, which have the following problems:

[0003] It is impossible to grasp the running status of the loom in real time, and it is difficult to detect and handle abnormal situations in time;

[0004] The maintenance cycle is fixed, and targeted maintenance cannot be performed according to the actual operating conditions, resulting in a waste of maintenance resources or untimely maintenance;

[0005] The lack of in-depth analysis of loom operation data makes it difficult to achieve predictive maintenance.

[0006] However, how to build an efficient and intelligent loom monitoring system is still a technical problem that needs to be solved urgently. Summary of the invention

[0007] In view of this, the purpose of the present invention is to provide a loom data monitoring method, system, computer equipment and storage medium to solve the technical problems of poor real-time performance, low maintenance efficiency and weak prediction performance of loom monitoring in the prior art.

[0008] To achieve the above object, in a first aspect, a loom data monitoring method is provided, which comprises the following steps:

[0009] S10: Collecting loom operation data through a distributed sensor network;

[0010] S20: Optimizing the operation data of the loom using an edge computing device to obtain the optimized operation data of the loom;

[0011] S30: the edge computing device transmits the optimized operating data of the loom to a cloud server;

[0012] S40: the cloud server analyzes the optimized operation data of the loom based on the loom operation status evaluation model to obtain an evaluation result of the loom operation status;

[0013] S50: The cloud server performs predictive maintenance on the loom based on the evaluation result of the operating status of the loom.

[0014] In some possible implementations, step S10 specifically includes:

[0015] S11: collecting the operation data of the loom through various sensors installed on the loom, wherein the various sensors include: a temperature sensor installed at the main bearing and the motor of the loom; a humidity sensor installed in the weaving area; a vibration sensor installed on the frame of the loom; a noise sensor installed around the loom; an energy consumption sensor connected to the power supply system of the loom; and a production sensor installed at the fabric output end;

[0016] S12: Transmitting the operation data of the loom to the edge computing device through the data acquisition module.

[0017] In some possible implementations, the data acquisition module includes a data acquisition card, an industrial controller, and an industrial gateway, and step S12 specifically includes:

[0018] S121: the industrial controller configures the sampling frequency of each sensor, and controls each sensor to continuously collect the operation data of the loom according to the configured sampling frequency;

[0019] S122: the data acquisition card converts the operation data of the loom collected by each sensor into a digital signal;

[0020] S123: the industrial controller preliminarily processes the digital signals from the multiple sensors to obtain preliminarily processed data;

[0021] S124: The industrial gateway performs data standardization processing on the preliminarily processed data to obtain standardized data;

[0022] S125: The industrial gateway transmits the standardized data to the edge computing device via a wired or wireless network.

[0023] In some possible implementations, step S20 specifically includes:

[0024] S21: adding a timestamp to the operation data of the loom, wherein the timestamp is synchronized with a standard time source through a network time protocol to obtain the operation data with the timestamp;

[0025] S22: performing intelligent filtering processing based on a deep learning algorithm on the operation data with the timestamp, removing abnormal data points, and obtaining filtered operation data;

[0026] S23: Calculate the statistical characteristics of the filtered operating data to obtain statistical characteristic data;

[0027] S24: merging the filtered operation data and the statistical feature data to obtain merged data;

[0028] S25: compressing the combined data using a lossless compression algorithm to obtain optimized operating data.

[0029] The process of intelligent filtering includes the following steps:

[0030] S221: constructing a deep learning model including a long short-term memory network LSTM or an autoencoder for learning a normal pattern of loom operation data;

[0031] S222: Using the historical normal operation data of the loom to train the deep learning model so that it can capture the timing pattern and multi-dimensional feature relationship of the loom operation;

[0032] S223: Input the newly collected time-stamped operation data into the trained deep learning model to calculate the anomaly score of each data point;

[0033] S224: Identify abnormal data points based on a preset abnormal threshold. The specific process is as follows:

[0034] Two abnormality thresholds are set, including a mild abnormality threshold T1 and a severe abnormality threshold T2, and T2 is greater than T1;

[0035] Compare the calculated anomaly score of each data point with the mild anomaly threshold T1 and the severe anomaly threshold T2;

[0036] If the anomaly score is less than T1, the data point is determined to be normal;

[0037] If the anomaly score is greater than or equal to T1 but less than T2, the data point is determined to be slightly abnormal;

[0038] If the anomaly score is greater than or equal to T2, the data point is determined to be a severe anomaly;

[0039] S225: For the identified abnormal data points, perform the following processing:

[0040] If the anomaly score of a data point is greater than or equal to T1 but less than T2, it is a mild anomaly and is corrected using the predicted value of the deep learning model; the predicted value is obtained by inputting the first N normal data points into the deep learning model and predicting the value at the next time point; the corrected data point replaces the anomaly value with the predicted value of the deep learning model;

[0041] If the anomaly score of a data point is greater than or equal to T2, it is considered a severe anomaly and the abnormal data point is marked and temporarily retained for subsequent analysis;

[0042] A real-time alert is triggered if K or more mild or severe abnormal data points appear in M ​​consecutive data points, where M is greater than K.

[0043] In some possible implementations, step S40 specifically includes the following steps performed by the cloud server:

[0044] S41: Adopting adaptive neuro-fuzzy inference system ANFIS and combining loom expert knowledge, a loom operation status evaluation model is established;

[0045] S42: Analyze the optimized operation data using the loom operation status evaluation model to generate health indexes of each subsystem of the loom;

[0046] S43: Based on the health index of each subsystem of the loom, the overall operation status score of the loom is determined through a dynamic weight fusion algorithm.

[0047] In some possible implementations, step S41 specifically includes the following steps performed by the cloud server:

[0048] S411: Acquire loom expert knowledge and build fuzzy rule base;

[0049] S412: Design an ANFIS network, which includes an input layer, a fuzzy layer, a rule layer, a normalization layer and an output layer;

[0050] An input layer, for receiving input data and passing the input data to the fuzzy layer;

[0051] The fuzzy layer is used to fuzzify the input data and obtain the fuzzy set;

[0052] A rule layer, used for inferring the fuzzy set according to the fuzzy rule base to generate the activation strength of the rule;

[0053] A normalization layer, used to normalize the activation strength of the rule to obtain a normalized activation strength, and ensure that the sum of the normalized activation strengths of all rules is 1;

[0054] An output layer, used for defuzzifying the normalized activation intensity to obtain a clear output;

[0055] S413: training the ANFIS network using historical operation data of the loom, optimizing parameters of the ANFIS network, and obtaining an optimized ANFIS network;

[0056] S414: combining the optimized ANFIS network with the fuzzy rule base to form a complete loom operation status evaluation model.

[0057] In some possible implementations, step S42 specifically includes:

[0058] S421: inputting the optimized operation data into the loom operation status evaluation model;

[0059] S422: Calculating the health index of each subsystem of the loom through the forward propagation process of the ANFIS network;

[0060] S423: normalizing the calculated health index of each subsystem to obtain the normalized health index of each subsystem of the loom.

[0061] In some possible implementations, step S43 specifically includes:

[0062] S431: Determine the initial impact weight of each subsystem of the loom on the overall performance of the loom based on the current production task and historical performance data;

[0063] S432: adjusting the initial impact weight according to the impact degree of the environmental factors on each subsystem of the loom to obtain an adjusted weight coefficient;

[0064] S433: performing weighted summation of the normalized health index of each subsystem of the loom and the adjusted weight coefficient to obtain an overall health index of the loom;

[0065] S434: Determine the overall operation status score of the loom according to the overall health index of the loom and a preset scoring standard.

[0066] In some possible implementations, step S50 specifically includes:

[0067] S51: The cloud server establishes a predictive maintenance model based on the evaluation result of the operation status of the loom and in combination with the historical maintenance records;

[0068] S52: the cloud server inputs the real-time operation data of the loom and the current operation status of the loom into the predictive maintenance model to generate a loom maintenance instruction;

[0069] S53: The cloud server sends the loom maintenance instruction to the terminal device of the maintenance personnel.

[0070] In some possible implementations, the method further includes:

[0071] S60: the cloud server optimizes the working parameters of the loom according to the operating data of the loom and the operating status evaluation result;

[0072] Step S60 specifically includes:

[0073] S61: The cloud server establishes a loom parameter optimization model using a deep learning algorithm based on the real-time operation data and historical operation records of the loom;

[0074] S62: The cloud server inputs the real-time operation data and current operation status of the loom into the loom parameter optimization model to generate an adjustment instruction for the loom working parameters;

[0075] S63: The cloud server sends the adjustment instruction of the loom working parameters to the controller of the loom to adjust the working parameters of the loom.

[0076] In a second aspect, a loom data monitoring system is provided, comprising:

[0077] A distributed sensor network to collect loom operation data;

[0078] an edge computing device, communicatively connected to the distributed sensor network, for optimizing the operation data, obtaining the optimized operation data, and transmitting the optimized operation data to a cloud server;

[0079] A cloud server is communicatively connected to the edge computing device, and the cloud server includes:

[0080] A data analysis module, used for analyzing the optimized operation data based on the loom operation status evaluation model to obtain the loom operation status evaluation result;

[0081] A predictive maintenance module is used to perform predictive maintenance on the loom based on the evaluation result of the running status of the loom.

[0082] According to a third aspect, a computer device is provided, the computer device comprising:

[0083] one or more processors;

[0084] A storage device for storing one or more programs;

[0085] When the one or more programs are executed by the one or more processors, the one or more processors implement the loom data monitoring method as described below:

[0086] Obtain optimized operation data of the loom from the edge computing device;

[0087] Analyzing the optimized operation data of the loom based on the loom operation status evaluation model to obtain the loom operation status evaluation result;

[0088] Predictive maintenance is performed on the loom based on the evaluation result of the operation status of the loom.

[0089] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the loom data monitoring method described below is implemented:

[0090] Obtain optimized operation data of the loom from the edge computing device;

[0091] Analyzing the optimized operation data of the loom based on the loom operation status evaluation model to obtain the loom operation status evaluation result;

[0092] Predictive maintenance of the loom is performed based on the evaluation results of the loom's operating status.

[0093] The above technical solution has the following beneficial technical effects:

[0094] The distributed sensor network enables comprehensive and real-time collection of loom operation data, improving the comprehensiveness and real-time nature of data collection. The edge computing device is used to optimize the collected data, reducing the amount of data transmission and improving data processing efficiency. The cloud server is used for in-depth analysis, making full use of the powerful computing power of cloud computing to achieve accurate evaluation of the loom operation status. Predictive maintenance based on the evaluation results of the operation status improves the pertinence and timeliness of maintenance, and reduces equipment failure rate and maintenance costs. The overall solution organically combines the Internet of Things, edge computing and cloud computing technologies to build an efficient and intelligent loom monitoring system, which significantly improves the operation efficiency and reliability of the loom. BRIEF DESCRIPTION OF THE DRAWINGS

[0095] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0096] Figure 1 is a flow chart of a loom data monitoring method according to an embodiment of the present invention;

[0097] Figure 2 is a specific flow chart of step S12 of an embodiment of the present invention;

[0098] Figure 3 is a specific flow chart of step S20 of an embodiment of the present invention;

[0099] Figure 4 is a specific flow chart of step S40 of an embodiment of the present invention;

[0100] Figure 5 is a specific flow chart of step S50 of an embodiment of the present invention;

[0101] Figure 6 is a specific flow chart of step S60 of an embodiment of the present invention;

[0102] Figure 7 is a functional block diagram of a loom data monitoring system according to an embodiment of the present invention;

[0103] Figure 8It is a schematic diagram of the structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0104] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0105] Embodiment 1

[0106] The embodiment of the present invention provides a loom data monitoring method, which utilizes distributed sensor networks, edge computing and cloud computing technologies to achieve real-time monitoring, intelligent analysis and predictive maintenance of the loom operation status. Figure 1 The flowchart shown is used to explain the specific embodiments of the present invention in detail.

[0107] like Figure 1 As shown, the loom data monitoring method of the present invention comprises the following steps:

[0108] S10: Collecting the operation data of the loom through the distributed sensor network. Step S10 may specifically include:

[0109] S11: Collect the operation data of the loom through various sensors installed on the loom; in this step, multiple sensors are installed on the loom to form a distributed sensor network. These sensors may include temperature sensors, humidity sensors, vibration sensors, noise sensors, energy consumption sensors and output sensors. These sensors are distributed in different parts of the loom to collect various operating parameters of the loom in real time. For example, the temperature sensor is installed at the main bearing and motor of the loom to monitor the temperature changes of key components; the humidity sensor is installed in the weaving area to monitor the ambient humidity; the vibration sensor is installed on the frame of the loom to detect the vibration of the loom; the noise sensor is installed around the loom to monitor the noise level; the energy consumption sensor is connected to the power supply system of the loom to record energy consumption; the output sensor is installed at the output end of the fabric to count the production quantity. S12: The operation data of the loom is transmitted to the edge computing device through the data acquisition module. These sensors transmit the collected data to the data acquisition module by wire or wireless means, and then the data acquisition module transmits the data to the edge computing device.

[0110] In other embodiments, other sensors that can be installed on the loom include: a tension sensor for monitoring changes in fabric tension; a pressure sensor for monitoring the pressure of various components of the loom; a displacement sensor for monitoring the position and movement of loom components; a photoelectric sensor for detecting yarn breakage or other fabric defects; a speed sensor for monitoring the speed of the loom spindle and other moving parts; a current sensor for monitoring current changes in motors and other electrical equipment; an air pressure sensor for monitoring air pressure changes in looms using pneumatic systems; and a color sensor for monitoring the consistency of fabric color.

[0111] S20: Utilize the edge computing device to optimize the operating data of the loom to obtain the optimized operating data of the loom.

[0112] In this step, the edge computing device receives the original operation data from the data acquisition module and optimizes the data. The purpose of the optimization is to improve the data quality, reduce the amount of data transmission, and provide a more reliable data basis for subsequent analysis.

[0113] Optimization processing can include the following aspects: first, data cleaning, removing obviously erroneous or abnormal data points to improve data quality; second, applying appropriate compression algorithms to compress data, reducing the amount of data while ensuring that important information is not lost; then, feature extraction is used to calculate statistical features such as mean, standard deviation, peak value, etc., to provide more dimensional information for the data; finally, adding accurate timestamps to all data to ensure the time synchronization of the data, thereby maintaining the temporal integrity of the data. Through these optimization processes, edge computing devices generate optimized operating data, which is more refined and reliable, and the data volume is also effectively compressed.

[0114] S30: The edge computing device transmits the optimized operating data of the loom to the cloud server.

[0115] After completing the data optimization process, the edge computing device transmits the processed data to the cloud server through the network. This step takes advantage of edge computing and completes the preliminary data processing locally, greatly reducing the amount of data that needs to be transmitted to the cloud and improving the efficiency of the entire system. During data transmission, appropriate encryption and security measures can be adopted to ensure the security and reliability of data transmission. For example, a data encryption algorithm is used to encrypt the data itself, using algorithms such as AES (Advanced Encryption Standard) and RSA (asymmetric encryption). Before data transmission, the edge computing device can first encrypt the data with these algorithms, and then decrypt and read the data after the cloud server receives it. For example, AES can perform multiple rounds of encryption on the data block, making it difficult to crack. For example, identity authentication and authorization are performed to ensure that only authorized devices and personnel can access and transmit data. This can be controlled by using API keys, OAuth authentication, etc. For example, when transmitting the running data of the loom to the cloud server, the identity of the device can be verified first to ensure that the transmission is from a trusted edge device. For example, integrity verification (Checksum, Hashing) is performed, which verifies the data before and after data transmission to ensure that the data has not been tampered with during transmission. Hash algorithms such as MD5 and SHA can be used. For example, the hash value of the data is calculated before transmission, and then calculated again and compared after transmission. If the values ​​are consistent, the data has not been modified.

[0116] S40: The cloud server analyzes the optimized operation data of the loom based on the loom operation status evaluation model to obtain an evaluation result of the loom operation status.

[0117] After receiving the optimized operation data, the cloud server uses the pre-established loom operation status evaluation model to conduct in-depth analysis on the data. This evaluation model can be trained based on a machine learning algorithm (such as a support vector machine, random forest, or deep neural network, etc.), which can identify the loom's operation status from multi-dimensional operation data.

[0118] The analysis process can include the following steps: first, data normalization is performed to unify different types of data to the same scale for unified processing; second, feature selection is performed to select the most representative features from a large number of features for analysis; then, the processed data is input into the evaluation model, and the status evaluation results of each component of the loom are obtained through model prediction; finally, based on the status of each component, the overall operating status of the loom is comprehensively evaluated to form a comprehensive operation analysis report. Through this step, the cloud server can obtain the current operating status evaluation results of the loom, including the health status of each key component and the operating status score of the entire machine.

[0119] S50: The cloud server performs predictive maintenance on the loom based on the evaluation result of the operating status of the loom.

[0120] Based on the operating status evaluation results obtained in the previous step, the cloud server performs predictive maintenance analysis. The purpose of this step is to predict possible problems and take corresponding maintenance measures before equipment failure occurs.

[0121] Predictive maintenance can include the following aspects: first, perform fault prediction, based on the current status and historical data, predict the type and time of possible faults; second, generate the optimal maintenance plan based on the prediction results, determine the maintenance time and maintenance items; at the same time, perform spare parts management, and prepare parts that may need to be replaced in advance according to the predicted fault type; finally, provide maintenance personnel with detailed maintenance guidance, including fault diagnosis and specific repair steps, to ensure efficient and accurate maintenance operations.

[0122] The cloud server sends the generated maintenance plan and guidance to relevant personnel so that maintenance operations can be carried out in a timely manner. Through the above steps, the present invention realizes comprehensive monitoring and intelligent maintenance of the loom operation status, greatly improves the operation efficiency and reliability of the loom, reduces unplanned downtime, and reduces maintenance costs.

[0123] like Figure 2 As shown, in some embodiments, the data acquisition module includes a data acquisition card, an industrial controller and an industrial gateway, and step S12 specifically includes:

[0124] S121: The industrial controller configures the sampling frequency of each sensor, and controls each sensor to continuously collect the operation data of the loom according to the configured sampling frequency.

[0125] In this step, the industrial controller first configures an appropriate sampling frequency for each sensor installed on the loom according to the specific operating characteristics and data requirements of the loom. For example, the sensor that monitors the spindle speed of the loom can be configured to sample 10 times per second, while the sensor that monitors the fabric tension can be configured to sample 100 times per second. After the configuration is completed, the industrial controller will continue to send sampling instructions to each sensor to ensure that the sensor continuously collects the operating data of the loom at the preset sampling frequency. This can ensure the continuity and integrity of the data and provide a reliable data basis for subsequent analysis.

[0126] S122: The data acquisition card converts the operation data of the loom collected by each sensor into a digital signal.

[0127] In this step, the data acquisition card receives analog signals from various sensors, which can be voltage, current or other forms of physical quantities. The built-in analog-to-digital converter (ADC) of the data acquisition card converts these analog signals into digital signals. For example, a 16-bit ADC can convert analog signals into digital values ​​in the range of 0-65535, thereby realizing the digital representation of the loom operation data. This conversion makes subsequent data processing and transmission more convenient and efficient.

[0128] S123: The industrial controller preliminarily processes the digital signals from the multiple sensors to obtain preliminarily processed data.

[0129] After receiving the digital signal converted by the data acquisition card, the industrial controller will perform a series of preliminary processing. First, the data is filtered to remove noise and outliers, such as using a median filter to remove instantaneous spike interference; then, the data with little change is compressed to reduce the amount of data, such as using differential encoding technology to optimize data storage and transmission; at the same time, an accurate timestamp is added to each set of data, such as a millisecond timestamp, to ensure the timing of the data for subsequent analysis; finally, the data from different sensors is classified and sorted, and the temperature, humidity, pressure and other data are classified separately for more targeted processing and analysis.

[0130] S124: The industrial gateway performs data standardization processing on the data after the preliminary processing to obtain standardized data.

[0131] After receiving the initially processed data from the industrial controller, the industrial gateway will perform data standardization to ensure data consistency. This process includes: converting data from different sources into a unified format, such as JSON or XML, for unified processing; converting data in different units, such as converting Fahrenheit to Celsius, to ensure data unit standardization; organizing data into a structured form according to a predefined data model for subsequent processing and analysis; and encoding the data, such as using UTF-8 encoding, to ensure data compatibility with subsequent systems. Through these standardized processing steps, it is ensured that data from different sensors and looms can be processed and analyzed in a consistent format.

[0132] S125: The industrial gateway transmits the standardized data to the edge computing device via a wired or wireless network.

[0133] In this last step, the industrial gateway transmits the standardized data to the edge computing device. The transmission can be carried out through a wired network (such as Ethernet) or a wireless network (Wi-Fi, 4G / 5G). The industrial gateway will select the most suitable transmission protocol, such as the message queue telemetry transmission protocol MQTT, OPC UA, etc., according to the network conditions and data characteristics. For example, for data with high real-time requirements, the MQTT protocol is selected for transmission; and for large quantities of historical data, the HTTP protocol is selected for batch transmission. This ensures the efficiency and reliability of data transmission and provides timely and accurate data input for subsequent edge computing and intelligent analysis.

[0134] like Figure 3 As shown, in some embodiments, the optimization process of step S20 specifically includes the following steps:

[0135] S21: Add a timestamp to the operation data of the loom, and the timestamp is synchronized with the standard time source through the network time protocol to obtain the operation data with the timestamp. In this step, the system uses the network time protocol (NTP) to synchronize with the standard time source (such as the National Time Service Center) to ensure the accuracy of the timestamp. For example, each data point will be assigned a timestamp accurate to milliseconds.

[0136] S22: Perform intelligent filtering based on a deep learning algorithm on the time-stamped operation data to remove abnormal data points and obtain filtered operation data. This step involves a complex deep learning algorithm, and the specific process is as follows:

[0137] S221: constructing a deep learning model including a long short-term memory network LSTM or an autoencoder for learning a normal pattern of loom operation data;

[0138] In this embodiment, the LSTM network is selected. LSTM is particularly suitable for processing time series data and can capture long-term dependencies. For example, this embodiment can construct a network structure including 3 layers of LSTM, each layer has 64 neurons, followed by a fully connected layer.

[0139] S222: Using the historical normal operation data of the loom to train the deep learning model so that it can capture the timing pattern and multi-dimensional feature relationship of the loom operation;

[0140] During the training process, this embodiment can use the normal operation data of the past 6 months. These data include various parameters of the loom, such as speed, tension, temperature, etc. The training process may require dozens of iterations until the loss function of the model converges to a stable low value.

[0141] S223: Input the newly collected time-stamped operation data into the trained deep learning model to calculate the anomaly score of each data point;

[0142] For each new data point, the model predicts its expected value and then calculates the difference between the actual value and the predicted value as an anomaly score. This score can be based on mean squared error or other distance metrics.

[0143] S224: Identify abnormal data points based on a preset abnormal threshold. The specific process is as follows:

[0144] Two abnormality thresholds are set, including a mild abnormality threshold T1 and a severe abnormality threshold T2, and T2 is greater than T1; the calculated abnormality score of each data point is compared with the mild abnormality threshold T1 and the severe abnormality threshold T2; if the abnormality score is less than T1, the data point is determined to be normal; if the abnormality score is greater than or equal to T1 but less than T2, the data point is determined to be mildly abnormal; if the abnormality score is greater than or equal to T2, the data point is determined to be severely abnormal; for example, in this embodiment, T1 can be set to 3 times the standard deviation and T2 to 5 times the standard deviation. This means that if the abnormality score of a data point exceeds 3 times the standard deviation but is less than 5 times the standard deviation, it will be marked as a mild abnormality; if it exceeds 5 times the standard deviation, it will be marked as a severe abnormality.

[0145] S225: For the identified abnormal data points, the following processing is performed: if the anomaly score of the data point is greater than or equal to T1 but less than T2, it is a mild anomaly and is corrected using the predicted value of the deep learning model; the predicted value is obtained by inputting the first N normal data points into the deep learning model and predicting the value at the next time point; the corrected data point replaces the abnormal value with the predicted value of the deep learning model; if the anomaly score of the data point is greater than or equal to T2, it is a severe anomaly, the abnormal data point is marked and temporarily retained for subsequent analysis; if K or more mild or severe abnormal data points appear in M ​​consecutive data points, a real-time alarm is triggered, where M is greater than K.

[0146] In this step, this embodiment can set N=10, that is, use the first 10 normal data points to predict the value of the next point. For severe anomalies, this embodiment will mark them as severe anomalies and save them in a separate log. For real-time alarms, this embodiment can set M=100, K=5, that is, if 5 or more abnormal points appear in 100 consecutive data points, an alarm will be triggered.

[0147] S23: Calculate the statistical characteristics of the filtered operating data to obtain statistical characteristic data;

[0148] In this step, the present embodiment calculates the following statistical features: first, the mean and standard deviation are calculated to describe the central tendency and dispersion of the data; second, the maximum and minimum values ​​are calculated to understand the range of the data; at the same time, the median and quartiles are calculated to provide the distribution of the data; finally, the skewness and kurtosis are calculated to measure the symmetry and peak degree of the data. These statistical features can provide an overall overview of the data distribution, which is helpful for subsequent analysis and decision-making.

[0149] S24: merging the filtered operation data and the statistical feature data to obtain merged data;

[0150] During the merging process, this embodiment can create a data structure for each time window (per hour or per shift), including the original data and the corresponding statistical features. This structured data format is convenient for subsequent storage and analysis.

[0151] S25: compressing the combined data using a lossless compression algorithm to obtain optimized operating data.

[0152] In this last step, the present embodiment may choose to use lossless compression algorithms such as Huffman coding. These lossless compression algorithms can significantly reduce data storage space while ensuring data integrity. For example, Huffman coding can reduce data size by 50-70% without losing any information.

[0153] Through the above steps, this embodiment realizes comprehensive optimization processing of loom operation data, including time synchronization, anomaly detection, data correction, feature extraction and data compression. This processing not only improves data quality, but also provides a reliable basis for subsequent intelligent analysis and decision-making.

[0154] like Figure 4 As shown, in one embodiment, the loom operation status evaluation process of step S40 is performed by a cloud server, and specifically includes the following steps:

[0155] S41: Adaptive Neuro-Fuzzy Inference System (ANFIS) is used to establish a loom operation status evaluation model in combination with loom expert knowledge; this step is the basis of the entire evaluation process and specifically includes the following sub-steps:

[0156] S411: Acquire loom expert knowledge and build fuzzy rule base;

[0157] In this step, the embodiment obtains a series of expert knowledge about the operating status of the loom. For example, a fuzzy rule can be: if the spindle speed is low and the fabric tension is high, the health of the weft insertion system may be poor. These rules are systematically organized into a fuzzy rule base containing hundreds of rules.

[0158] S412: Design an ANFIS network, which includes an input layer, a fuzzy layer, a rule layer, a normalization layer and an output layer;

[0159] The design of the ANFIS network is as follows: Input layer, used to receive multi-dimensional input data, such as spindle speed, fabric tension, temperature, etc. Each input corresponds to a node. Fuzzy layer, used to fuzzify each input. For example, the spindle speed can be fuzzified into three fuzzy sets of low, medium and high using Gaussian membership function. Rule layer, used to generate rule activation strength based on the fuzzy rule base. Assuming there are 100 rules, this layer has 100 nodes. Normalization layer, used to normalize the rule activation strength to ensure that the sum of the normalized activation strengths of all rules is 1. Output layer, used to perform defuzzification to obtain clear numerical output, such as the health index of the subsystem.

[0160] The fuzzification process first processes the input data and receives data from the input layer, such as spindle speed, fabric tension, temperature, etc., in preparation for fuzzy conversion. Next, a predefined Gaussian membership function is applied to each input variable. For example, the spindle speed can be divided into three membership functions: "low", "medium", and "high". Then, the membership value of the data of each input layer under each membership function is calculated, for example, the membership of a spindle speed value in the three fuzzy sets of "low", "medium", and "high" is calculated. Finally, all the calculated membership values ​​are combined to form a fuzzy set as the output of the fuzzy layer.

[0161] The defuzzification process first receives the normalized results of the rule activation strength from the normalization layer to prepare for generating a crisp output. Then, the output value of each rule is multiplied by its corresponding normalized activation strength through the weighted average calculation, and these weighted values ​​are summed. Next, the above weighted sum is divided by the sum of all normalized activation strengths to obtain the final crisp numerical output. Finally, the calculated crisp output is interpreted as the corresponding subsystem status indicator, such as the health index of the subsystem, thereby completing the quantitative evaluation of the loom system status.

[0162] Through the above-mentioned fuzzification and defuzzification processing, the ANFIS network can effectively handle the uncertainty and ambiguity existing in the loom system, use expert knowledge and data learning to provide reliable status assessment results, and provide strong support for system optimization and maintenance decisions.

[0163] The following example illustrates the working principle of ANFIS network:

[0164] In one example, the ANFIS network is designed to assess the health of multiple subsystems of a loom using three input variables: spindle speed (450 rpm), fabric tension (80 N), and temperature (35 ° C). At the fuzzy layer, each input variable has three Gaussian membership functions (low, medium, and high). The membership functions of the spindle speed include low speed (A1, center value 300 rpm, standard deviation 50), medium speed (A2, center value 450 rpm, standard deviation 50), and high speed (A3, center value 600 rpm, standard deviation 50). The membership functions of fabric tension include low tension (B1, center value 60 N, standard deviation 10), medium tension (B2, center value 80 N, standard deviation 10), and high tension (B3, center value 100 N, standard deviation 10). The membership functions of temperature include low temperature (C1, center value 25 ° C, standard deviation 5), medium temperature (C2, center value 35 ° C, standard deviation 5), and high temperature (C3, center value 45 ° C, standard deviation 5). Through calculation, the membership of each input variable under different membership functions is obtained: the membership of the spindle speed x1 (450rpm) is μA1≈0.05, μA2≈1.00, μA3≈0.05; the membership of the fabric tension x2 (80N) is μB1≈0.14, μB2≈1.00, μB3≈0.14; the membership of the temperature x3 (35℃) is μC1≈0.14, μC2≈1.00, μC3≈0.14.

[0165] The rule layer contains 27 rules, combining different states of the three input variables. For example, Rule 1 states: If the spindle speed is medium (A2), the fabric tension is medium (B2), and the temperature is medium (C2), then the health of subsystem 1 is high (y1=0.9), the health of subsystem 2 is medium-high (y2=0.85), and the health of subsystem 3 is medium (y3=0.8). Rule 2 states: If the spindle speed is medium (A2), the fabric tension is medium (B2), and the temperature is low (C1), then the health of subsystem 1 is medium-high (y1=0.8), the health of subsystem 2 is medium (y2=0.75), and the health of subsystem 3 is medium-low (y3=0.7). Rule 3 states that if the spindle speed is medium (A2), the fabric tension is low (B1), and the temperature is medium (C2), then the health of subsystem 1 is medium (y1=0.7), the health of subsystem 2 is medium-low (y2=0.65), and the health of subsystem 3 is low (y3=0.6). The activation strength of each rule is calculated by taking the minimum value of the corresponding membership of the input variables, for example, the activation strength of rule 1 is w1=min(μA2,μB2,μC2)=min(1.00,1.00,1.00)=1.00, the activation strength of rule 2 is w2=min(μA2,μB2,μC1)=min(1.00,1.00,0.14)=0.14, and the activation strength of rule 3 is w3=min(μA2,μB1,μC2)=min(1.00,0.14,1.00)=0.14.

[0166] At the normalization layer, the normalized activation strength of each rule is calculated. Assuming the activation strengths of the first three rules are as described above, the sum of the activation strengths of the remaining rules is 0.05, and the sum is 1.33 (1.00 + 0.14 + 0.14 + 0.05). Therefore, the normalized activation strength of rule 1 is Normalized activation strength of rule 2 Normalized activation strength of rule 3 The sum of the normalized activation intensities of the remaining rules This step ensures that the sum of all rule activation strengths is 1.

[0167] Finally, defuzzification is performed at the output layer. For each subsystem, the weighted average method is used to calculate the final health index. The health index of subsystem 1 is calculated as follows: y1 = (0.9*0.7519) + (0.8*0.1053) + (0.7*0.1053) + (0.6*0.0375) ≈ 0.8556, indicating a relatively high health state. The health index of subsystem 2 is calculated as follows: y2 = (0.85*0.7519) + (0.75*0.1053) + (0.65*0.1053) + (0.55*0.0375) ≈ 0.8056, indicating a medium-high health state. The health index of subsystem 3 is calculated as follows: y3 = (0.8*0.7519) + (0.7*0.1053) + (0.6*0.1053) + (0.5*0.0375) ≈ 0.7556, indicating a medium health state. This approach allows the use of the same ANFIS network to simultaneously evaluate the health status of multiple related subsystems based on the same input variables (spindle speed, fabric tension, and temperature), providing a comprehensive health assessment for the maintenance and management of the loom.

[0168] In an alternative embodiment, step S41 is replaced by the following steps performed by the cloud server:

[0169] S411': Obtain expert knowledge of each subsystem of the loom and construct multiple fuzzy rule bases, each fuzzy rule base corresponds to a subsystem;

[0170] S412': Design an ANFIS network for each subsystem, each ANFIS network includes an input layer, a fuzzy layer, a rule layer, a normalization layer and an output layer; the input layer is used to receive input data of the corresponding subsystem and pass the input data to the fuzzy layer; the fuzzy layer is used to fuzzify the input data to obtain a fuzzy set; the rule layer is used to infer the fuzzy set according to the corresponding fuzzy rule base to generate the activation strength of the rule; the normalization layer is used to normalize the activation strength of the rule to obtain the normalized activation strength, and ensure that the sum of the normalized activation strengths of all rules is 1; the output layer is used to defuzzify the normalized activation strength to obtain a clear output;

[0171] S413': using the historical operation data of each subsystem of the loom to train the corresponding ANFIS network respectively, optimizing the parameters of each ANFIS network, and obtaining multiple optimized ANFIS networks;

[0172] S414': combining each optimized ANFIS network with the corresponding fuzzy rule base to form a complete set of evaluation models for the running status of each subsystem of the loom.

[0173] Furthermore, step S42' specifically includes: S421': inputting the optimized subsystem operation data into the corresponding subsystem operation status evaluation model respectively; S422': calculating the health index of each subsystem of the loom respectively through the forward propagation process of each ANFIS network; S423': normalizing the calculated health index of each subsystem to obtain the normalized health index of each subsystem of the loom.

[0174] Each subsystem can have its own dedicated ANFIS network to assess the health of the subsystem. Each ANFIS network can be optimized for a specific subsystem to provide a more accurate health assessment. The ANFIS network for each subsystem can be designed and adjusted independently, including selecting different input variables, fuzzy rules, and membership functions to ensure that the characteristics of each subsystem are fully considered. In addition, this design has good scalability. When adding or modifying a subsystem, only the corresponding ANFIS network needs to be adjusted, and other parts will not be affected. Multiple ANFIS networks can run in parallel, thereby increasing processing speed. Fault isolation is another advantage. If there is a problem with the ANFIS network of a subsystem, it will not affect the evaluation of other subsystems, ensuring the robustness of the system. Finally, in terms of maintenance and updating, the model can be updated and optimized for a single subsystem without retraining the entire system, which greatly reduces maintenance costs and complexity.

[0175] S413: training the ANFIS network using historical operation data of the loom, optimizing parameters of the ANFIS network, and obtaining an optimized ANFIS network;

[0176] In this step, the present embodiment uses the loom operation data of the past year for training. The training process uses the back propagation algorithm to optimize the parameters in the ANFIS network, such as the parameters of the membership function and the rule weights. The training requires more than thousands of iterations until the error between the network output and the actual historical evaluation results drops to an acceptable level (mean square error less than 0.01).

[0177] S414: combining the optimized ANFIS network with the fuzzy rule base to form a complete loom operation status evaluation model.

[0178] The optimized ANFIS network is combined with the fuzzy rule base to form a complete evaluation model that is capable of adaptive learning and contains expert knowledge.

[0179] The fuzzy rule base is a set of IF-THEN rules pre-defined by domain experts based on their experience and knowledge. These rules include the possible health states of the loom subsystems under various input conditions. For example, if the spindle speed is high, the fabric tension is medium, and the temperature is low, the weaving system health is medium.

[0180] The integration process includes multiple steps. The first is rule structure alignment, which ensures that the rule structure in the ANFIS network is consistent with the rule structure in the fuzzy rule base, which involves adjusting the number or structure of rules in the network to match the expert rules. The second is parameter fusion, which converts the expert knowledge in the fuzzy rule base into the initial parameters or constraints of the ANFIS network, such as using the expert-defined membership function as the initial value of the network membership function or optimizing the boundary. Then there is rule weight adjustment, which adjusts the weights of the corresponding rules in the ANFIS network according to the importance of the rules in the fuzzy rule base to ensure that the expert knowledge is fully reflected. Finally, there is output calibration, which uses typical cases in the fuzzy rule base to calibrate the output of the ANFIS network to ensure that the output is consistent with expert expectations.

[0181] The integrated model can be validated through a test data set to ensure that the model reflects both the patterns of historical data and expert knowledge. The validation process verifies the reliability and accuracy of the integrated model by evaluating its predictive ability for new data, thereby verifying its performance in actual applications.

[0182] During the integration process, the adaptive learning ability of the ANFIS network is retained, so that the model can continue to learn and adjust in actual use to adapt to changes in the loom operating environment. This adaptive mechanism allows the model to dynamically adjust its own parameters when faced with new data or changing operating conditions, maintaining the ability to accurately evaluate the loom's operating status.

[0183] S42: Analyze the optimized operation data using the loom operation status evaluation model to generate health indexes of each subsystem of the loom. This step specifically includes:

[0184] S421: inputting the optimized operation data into the loom operation status evaluation model;

[0185] For example, in this embodiment, the operating data of the last 24 hours can be input, including multiple parameters such as spindle speed, fabric tension, temperature, humidity, etc.

[0186] S422: Calculating the health index of each subsystem of the loom through the forward propagation process of the ANFIS network;

[0187] Assume that the loom has five main subsystems: weft insertion subsystem, warp control subsystem, beating subsystem, take-up subsystem and parking subsystem. The parking subsystem is mainly responsible for controlling the stop of the loom in abnormal situations or at the end of operation, including immediate parking in case of yarn breakage or other failures. The ANFIS network calculates a health index for each subsystem, which can range from 0 to 100. The following example describes the working process of step S422 in detail:

[0188] The input layer receives real-time data of the loom operation, including four key variables: spindle speed, fabric tension, temperature and vibration amplitude. In this example, the specific input values ​​are: spindle speed x1 = 480rpm, fabric tension x2 = 85N, temperature x3 = 38℃, vibration amplitude x4 = 0.15mm.

[0189] The fuzzification layer uses predefined Gaussian membership functions to convert precise numerical inputs into fuzzy sets. Each input variable has three membership functions, representing low, medium, and high states. Taking the spindle speed (x1 = 480rpm) as an example, the calculated membership is: μA1_low(480)≈0.01 (low speed), μA1_medium(480)≈0.85 (medium speed), μA1_high(480)≈0.14 (high speed). This shows that the current speed is mainly considered to be medium speed, but it is also slightly biased towards high speed. Similar fuzzification is performed on other input variables.

[0190] The rule layer contains 81 IF-THEN rules, which reflect the relationship between the input variables and the health status of each subsystem. For example, Rule 1 can be: If the speed is medium AND the tension is medium AND the temperature is medium AND the vibration is low, then the health of the weft insertion system is high, the health of the warp control system is medium-high, the health of the beating system is medium, the health of the take-up system is high, and the health of the parking subsystem is medium-high. The activation strength of Rule 1 is calculated as: w1 = min(0.85, μB1_medium(85), μC1_medium(38), μD1_low(0.15)), where 0.85 is the medium speed membership of the speed, and the other items are the membership of the corresponding input under their respective membership functions.

[0191] The normalization layer calculates the relative importance of each rule. Assuming that the sum of the activation strengths of all rules is S, the normalized activation strength of rule 1 is: This step ensures that the total impact of all rules remains consistent, making the contribution of each rule proportional to its activation strength.

[0192] The defuzzification layer converts the results of fuzzy reasoning into precise numerical outputs. Taking the weft insertion subsystem as an example, the calculation formula for its health index is: where 90, 85, etc. are the weft insertion system health corresponding to each rule (converted to a scale of 0-100). This weighted average approach takes into account the contribution of all rules, weighted according to their activation strength.

[0193] The output layer finally gives the health index of the five subsystems, ranging from 0 to 100. In this example, the calculated results are: the health index of the weft insertion subsystem is 87.5, the health index of the warp control subsystem is 82.3, the health index of the beating-up subsystem is 79.8, the health index of the winding subsystem is 85.6, and the health index of the parking subsystem is 81.2. These indexes intuitively reflect the health status of each subsystem and provide a quantitative basis for loom maintenance and management. For example, a score of 87.5 for the weft insertion subsystem indicates that the system is in good condition, while a score of 79.8 for the beating-up subsystem may indicate the need for preventive maintenance.

[0194] S423: normalizing the calculated health index of each subsystem to obtain the normalized health index of each subsystem of the loom. The normalization process ensures that the health index of all subsystems is between 0 and 1, which is convenient for subsequent weight calculation and comprehensive evaluation.

[0195] S43: Based on the health index of each subsystem of the loom, determine the overall operation status score of the loom through a dynamic weight fusion algorithm. This step is further divided into:

[0196] S431: Determine the initial impact weight of each subsystem of the loom on the overall performance of the loom based on the current production task and historical performance data;

[0197] For example, if a high-density fabric is currently being produced, the initial weight of the warp control system will be set higher, such as 0.3, while the other systems will each be 0.175.

[0198] S432: adjusting the initial impact weight according to the impact degree of the environmental factors on each subsystem of the loom to obtain an adjusted weight coefficient;

[0199] Assume that the ambient humidity is detected to be high, which has a greater impact on the weft insertion subsystem. The system will adjust the weight of the weft insertion subsystem from 0.175 to 0.2, and reduce the weights of other subsystems accordingly.

[0200] S433: performing weighted summation of the normalized health index of each subsystem of the loom and the adjusted weight coefficient to obtain an overall health index of the loom;

[0201] For example, if the normalized health indexes of the five subsystems are 0.9, 0.85, 0.95, 0.88, and 0.92, respectively, and the adjusted weights are 0.2, 0.3, 0.15, 0.2, and 0.15, respectively, then the overall health index of the loom is:

[0202] 0.9*0.2+0.85*0.3+0.95*0.15+0.88*0.2+0.92*0.15=0.8935.

[0203] S434: Determine the overall operation status score of the loom according to the overall health index of the loom and a preset scoring standard.

[0204] In this embodiment, the following scoring criteria can be set: when the score is between 0.9 and 1.0, it is evaluated as excellent and gets 5 points; when the score is between 0.8 and 0.9, it is evaluated as good and gets 4 points; when the score is between 0.7 and 0.8, it is evaluated as average and gets 3 points; when the score is between 0.6 and 0.7, it is evaluated as poor and gets 2 points; when the score is lower than 0.6, it is evaluated as terrible and gets 1 point. According to the overall health index of 0.8935 obtained by the above calculation, the overall operation status of the loom is scored as 4 points, which is in a good state.

[0205] Through the above steps, this embodiment realizes a comprehensive and intelligent evaluation of the loom operation status. This evaluation method not only takes into account expert knowledge, but also utilizes the adaptive ability of machine learning, and at the same time considers the influence of current production tasks and environmental factors through a dynamic weight fusion algorithm, thereby obtaining a comprehensive, accurate and dynamic loom operation status score, which provides a reliable decision-making basis for production management and equipment maintenance of weaving enterprises.

[0206] like Figure 5 As shown, in a specific embodiment of the present invention, the loom maintenance instruction generation process of step S50 is executed by the cloud server, and specifically includes the following steps:

[0207] S51: The cloud server establishes a predictive maintenance model based on the evaluation results of the loom's operating status and in combination with historical maintenance records. In this step, the cloud server uses a machine learning algorithm, especially a random forest algorithm, to build a predictive maintenance model. The model building process includes data collection, feature engineering, model training, model verification, and model optimization. In the data collection phase, the system collects the evaluation results of the loom's operating status and the corresponding maintenance records for the past three years, including the health index of each subsystem (weft insertion subsystem, warp control subsystem, beating subsystem, winding subsystem, and warp delivery subsystem), the overall operating status score of the loom, the type of maintenance operation, the maintenance time interval, and the degree of performance improvement after maintenance. In the feature engineering phase, meaningful features are extracted from the collected data, such as the rate of change of the health index of each subsystem, the trend of the overall operating status score, the interval from the last maintenance to the current time, the type of fabric produced, and environmental factors. Subsequently, the prediction model is trained using the random forest algorithm, with the goal of predicting the best time for the next maintenance and the specific maintenance operations that need to be performed. Finally, the model performance was evaluated by cross-validation method to ensure that the accuracy reached at least 85%, and the model performance was optimized by adjusting the hyperparameters.

[0208] S52: The cloud server inputs the real-time operation data of the loom and the current operation status of the loom into the predictive maintenance model to generate a loom maintenance instruction. In this step, the cloud server continuously receives the real-time operation data of the loom and performs a series of operations regularly (for example, every hour). First, the real-time operation data is cleaned and standardized to ensure that the data format is consistent with the data used when training the model. Then, the same features as those used in training are extracted from the preprocessed data. Next, the ANFIS model established in step S40 is used to evaluate the current operation status of the loom. After the evaluation is completed, the extracted features and the current operation status evaluation results are input into the predictive maintenance model. The model output includes the next recommended maintenance time and a list of recommended maintenance operations, each of which has a priority score. Finally, based on the model output, the system generates specific maintenance instructions, such as "It is recommended to check and adjust the weft insertion subsystem within 36 hours, with a high priority", "It is recommended to replace the tension sensor of the warp control subsystem within 72 hours, with a medium priority", "It is recommended to lubricate the warp let-off device at the next shutdown, with a low priority", etc.

[0209] S53: The cloud server sends the loom maintenance instruction to the terminal device of the maintenance personnel. In this step, the system first sorts the instructions according to the priority of each maintenance instruction. Then, the system matches the appropriate maintenance personnel according to the type and complexity of the maintenance instruction, for example, assigning more complex maintenance tasks to more experienced technicians. Next, the system sends the maintenance instruction to the terminal device of the corresponding maintenance personnel in a variety of ways, including push notifications to the maintenance personnel's mobile phone app, sending detailed maintenance instruction emails, updating the task list in the maintenance personnel's work management system, etc. The system is also equipped with a confirmation mechanism, requiring the maintenance personnel to confirm receipt of the instruction and submit feedback after performing the maintenance task. For high-priority maintenance instructions, the system will also notify the relevant person in charge by SMS or phone to ensure timely processing. In this way, the present embodiment realizes intelligent predictive maintenance based on the operation status of the loom, improves maintenance efficiency, reduces unplanned downtime, and thus improves the overall production efficiency and fabric quality of the loom.

[0210] like Figure 6 As shown, in another specific embodiment of the present invention, the method further includes step S60, that is, the cloud server optimizes the working parameters of the loom according to the operating data and operating status evaluation results of the loom. The specific implementation process of this step is as follows:

[0211] S61: The cloud server uses a deep learning algorithm to establish a loom parameter optimization model based on the real-time operation data and historical operation records of the loom. In this step, the cloud server uses a deep reinforcement learning (DRL) algorithm, especially a deep deterministic policy gradient (DDPG) algorithm to build a loom parameter optimization model. The model building process includes several stages: data preparation, model design, model training, and model verification. First, in the data preparation stage, the system collects and organizes the loom operation data of the past two years, including various working parameters (such as weft tension, warp tension, weft beating force, fabric density, etc.), corresponding production efficiency data, and fabric quality data. Then, in the model design stage, a DDPG model structure including an Actor network and a Critic network is constructed. The Actor network is used to select the optimal parameter adjustment action according to the current state, and the Critic network is used to evaluate the value of the selected action. Then, in the model training stage, historical data is used to simulate the loom operation environment, and the parameter selection strategy of the model is optimized through repeated trials and learning. Finally, in the model verification phase, the model is tested and fine-tuned using actual operating data from the last three months to ensure the effectiveness of the model in the actual production environment. In deep reinforcement learning, especially in the Actor-Critic method, the Actor is responsible for selecting the best action based on the current state. It learns a strategy to decide what action should be taken in a given state. The Critic is responsible for evaluating the value of the action selected by the Actor. It learns a value function to estimate the expected return of taking an action in a given state.

[0212] S62: The cloud server inputs the real-time operation data and current operation status of the loom into the loom parameter optimization model to generate adjustment instructions for the loom working parameters. In this step, the cloud server continuously receives the real-time operation data and current operation status of the loom, and performs a parameter optimization process at a certain time interval (for example, every 15 minutes). First, the system preprocesses the received real-time data, including data cleaning, standardization and feature extraction. Then, the processed data is input into the DDPG model trained in step S61. The Actor network of the model will output a set of recommended parameter adjustment values ​​based on the input status information, such as "increase the weft tension by 0.2N", "reduce the warp tension by 0.1N", "increase the weft beating force by 5%", etc. The Critic network evaluates the expected benefits of these adjustment suggestions. The system comprehensively considers the output of the Actor network and the evaluation of the Critic network to generate the final parameter adjustment instructions. These instructions include not only specific parameter adjustment values, but also the priority and expected effect of the adjustment. For example, "adjust the weft tension from 10.5N to 10.7N, with a high priority, and it is expected to increase the fabric strength by 2%."

[0213] S63: The cloud server sends the adjustment instruction of the loom working parameter to the controller of the loom to adjust the working parameters of the loom. In this step, the system first performs a security check on the generated parameter adjustment instruction to ensure that all adjustments are within the preset safety range. Then, the system determines the execution order according to the priority and expected effect of the adjustment instruction. For adjustments with high priority and significant expected effect, the system will immediately send the instruction to the controller of the loom. For adjustments with low priority or less obvious expected effect, the system will temporarily store it and wait for the next appropriate time (such as a small interval of the loom or a shift change) to execute it. When sending the adjustment instruction, the system uses an encrypted communication protocol to ensure the security of instruction transmission. After receiving the instruction, the loom controller will gradually implement parameter adjustment to avoid sudden and large changes affecting production. At the same time, the controller will monitor the various indicators in the adjustment process in real time. If an abnormal situation is found, the adjustment will be immediately terminated and restored to the previous parameter settings. After the adjustment is completed, the controller will feedback the execution results to the cloud server, including the actual adjusted parameter values ​​and the immediate effect data after adjustment.

[0214] Through the above steps, this embodiment realizes the automatic optimization of loom working parameters based on deep learning. This method can dynamically adjust the working parameters according to the real-time operating status and historical performance of the loom, which not only improves the production efficiency of the loom, but also continuously optimizes the quality of the fabric. At the same time, due to the use of deep reinforcement learning algorithms, the system can continuously learn and improve from new operating data, making the parameter optimization strategy more and more accurate and effective over time. This intelligent parameter adjustment method greatly reduces the need for manual intervention and improves the automation level of the loom.

[0215] Embodiment 2

[0216] like Figure 7 As shown, this embodiment provides a loom data monitoring system 700, which includes:

[0217] A distributed sensor network 710 for collecting operation data of the loom;

[0218] The edge computing device 720 is used to optimize the operation data of the loom, obtain the optimized operation data of the loom, and transmit the optimized operation data of the loom to the cloud server;

[0219] The cloud server 730 is used to analyze the optimized operation data of the loom based on the loom operation status evaluation model, obtain the loom operation status evaluation result, and perform predictive maintenance on the loom based on the loom operation status evaluation result.

[0220] The cloud server includes: a data analysis module, which is used to analyze the optimized operation data based on the loom operation status evaluation model to obtain the loom operation status evaluation result; and a predictive maintenance module, which is used to perform predictive maintenance on the loom based on the loom operation status evaluation result.

[0221] In some embodiments, the distributed sensor network includes: a temperature sensor installed at the main bearing and motor of the loom; a humidity sensor installed in the weaving area; a vibration sensor installed on the frame of the loom; a noise sensor installed around the loom; an energy consumption sensor connected to the power supply system of the loom; a production sensor installed at the fabric output end; and a data acquisition module for transmitting the operating data of the loom to an edge computing device.

[0222] In some embodiments, the data acquisition module includes:

[0223] A data acquisition card, used to convert the operation data of the loom collected by various sensors into digital signals;

[0224] The industrial controller is used to configure the sampling frequency of each sensor, control each sensor to continuously collect the operation data of the loom according to the configured sampling frequency, and preliminarily process the digital signals from multiple sensors to obtain preliminarily processed data;

[0225] The industrial gateway is used to perform data standardization processing on the data after the preliminary processing to obtain standardized data, and transmit the standardized data to the edge computing device through a wired or wireless network.

[0226] In some embodiments, the edge computing device includes:

[0227] A timestamp adding module is used to add a timestamp to the operation data of the loom, wherein the timestamp is synchronized with a standard time source through a network time protocol to obtain the operation data with a timestamp;

[0228] An intelligent filtering module, used to perform intelligent filtering processing on the operating data with timestamps based on a deep learning algorithm, remove abnormal data points, and obtain filtered operating data;

[0229] A statistical feature calculation module, used to calculate the statistical features of the filtered operating data to obtain statistical feature data;

[0230] A data merging module, used to merge the filtered operating data and the statistical feature data to obtain merged data;

[0231] The data compression module is used to compress the combined data using a lossless compression algorithm to obtain optimized operating data.

[0232] In some embodiments, the intelligent filtering module includes:

[0233] Deep learning models, including long short-term memory networks (LSTMs) or autoencoders, are used to learn normal patterns in loom operation data;

[0234] an anomaly detection unit, for calculating an anomaly score for each data point and identifying anomaly data points based on a preset anomaly threshold;

[0235] The data processing unit is used to process the identified abnormal data points.

[0236] In some embodiments, the cloud server includes:

[0237] A model building module is used to establish a loom operation status assessment model using an adaptive neuro-fuzzy inference system ANFIS combined with loom expert knowledge;

[0238] A health index generating module, used for analyzing the optimized operation data by using the loom operation status evaluation model to generate the health index of each subsystem of the loom;

[0239] The status scoring module is used to determine the overall operation status score of the loom based on the health index of each subsystem of the loom through a dynamic weight fusion algorithm.

[0240] In some embodiments, the model building module includes:

[0241] Knowledge acquisition unit, used to acquire loom expert knowledge and build a fuzzy rule base;

[0242] ANFIS network design unit, used to design the ANFIS network including input layer, fuzzy layer, rule layer, normalization layer and output layer;

[0243] A network training unit, used to train the ANFIS network using historical operation data of the loom, optimize the parameters of the ANFIS network, and obtain an optimized ANFIS network;

[0244] The model integration unit is used to combine the optimized ANFIS network with the fuzzy rule base to form a complete loom operation status evaluation model.

[0245] In some embodiments, the cloud server further includes:

[0246] A predictive maintenance module, for establishing a predictive maintenance model based on the evaluation result of the operation status of the loom and in combination with the historical maintenance records, and inputting the real-time operation data of the loom and the current operation status of the loom into the predictive maintenance model to generate a loom maintenance instruction;

[0247] The instruction sending module is used to send the loom maintenance instruction to the terminal device of the maintenance personnel.

[0248] In some embodiments, the cloud server further includes:

[0249] The parameter optimization module is used to optimize the working parameters of the loom according to the operating data of the loom and the operating status evaluation result.

[0250] In some embodiments, the parameter optimization module includes:

[0251] A model building unit, used to build a loom parameter optimization model using a deep learning algorithm based on the real-time operation data and historical operation records of the loom;

[0252] An instruction generation unit, used for inputting the real-time operation data and current operation status of the loom into the loom parameter optimization model to generate adjustment instructions for the loom working parameters;

[0253] The instruction sending unit is used to send the adjustment instruction of the working parameters of the loom to the controller of the loom to adjust the working parameters of the loom.

[0254] In some embodiments, it also includes: a loom controller for receiving and executing adjustment instructions for loom working parameters; and a maintenance personnel terminal device for receiving loom maintenance instructions.

[0255] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0256] Embodiment 3

[0257] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, any one of the loom data monitoring methods described above is implemented.

[0258] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. Of course, there are other ways of readable storage media, such as quantum memory, graphene memory, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media does not include electrical carrier signals and telecommunication signals.

[0259] Embodiment 4

[0260] like Figure 8 As shown, the present invention also provides a computer device. The computer device of the embodiment of the present invention includes: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the loom data monitoring method provided by the present invention.

[0261] Reference below Figure 8 , which shows a schematic diagram of the structure of a computer device 800 suitable for implementing an embodiment of the present invention. Figure 8 The computer device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0262] like Figure 8As shown, the computer device 800 includes a central processing unit (CPU) 801, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 802 or the program loaded from the storage part 808 to the random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the computer system 800 are also stored. The CPU 801, the ROM 802, and the RAM 803 are connected to each other through a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0263] The following components are connected to the I / O interface 805: an input section 806 including a keyboard, a mouse, etc.; an output section 807 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a LAN card, a modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the I / O interface 805 as needed. A removable medium 811, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 810 as needed so that a computer program read therefrom is installed into the storage section 808 as needed.

[0264] In particular, according to the embodiments disclosed in the present invention, the process described in the main step diagram above can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the main step diagram. In the above embodiment, the computer program can be downloaded and installed from the network through the communication part 809, and / or installed from the removable medium 811. When the computer program is executed by the central processing unit 801, the above functions defined in the system of the present invention are executed.

[0265] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer readable signal media may also be any computer readable medium other than computer readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0266] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the above-mentioned module, program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0267] The units involved in the embodiments of the present invention may be implemented by software or hardware. The units described may also be arranged in a processor, and the names of these units do not constitute limitations on the units themselves in some cases.

[0268] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A loom data monitoring method, characterized in that: The following steps are involved: S10: Collecting loom operation data through a distributed sensor network; S20: Optimizing the operation data of the loom using an edge computing device to obtain optimized operation data of the loom; S30: the edge computing device transmits the optimized operating data of the loom to a cloud server; S40: the cloud server analyzes the optimized operation data of the loom based on the loom operation status evaluation model to obtain an evaluation result of the loom operation status; S50: The cloud server performs predictive maintenance on the loom based on the evaluation result of the operating status of the loom.

2. The loom data monitoring method according to claim 1, characterized in that: Step S40 specifically includes the following steps performed by the cloud server: S41: Adopting adaptive neuro-fuzzy inference system ANFIS and combining loom expert knowledge, a loom operation status evaluation model is established; S42: Analyze the optimized operation data using the loom operation status evaluation model to generate health indexes of each subsystem of the loom; S43: Based on the health index of each subsystem of the loom, the overall operation status score of the loom is determined through a dynamic weight fusion algorithm.

3. The loom data monitoring method according to claim 2, characterized in that: Step S41 specifically includes the following steps performed by the cloud server: S411: Acquire loom expert knowledge and build fuzzy rule base; S412: Design an ANFIS network, which includes an input layer, a fuzzy layer, a rule layer, a normalization layer and an output layer; An input layer, for receiving input data and passing the input data to the fuzzy layer; The fuzzy layer is used to fuzzify the input data and obtain the fuzzy set; A rule layer, used for inferring the fuzzy set according to the fuzzy rule base to generate the activation strength of the rule; A normalization layer, used to normalize the activation strength of the rule to obtain a normalized activation strength, and ensure that the sum of the normalized activation strengths of all rules is 1; An output layer, used for defuzzifying the normalized activation intensity to obtain a clear output; S413: training the ANFIS network using historical operation data of the loom, optimizing parameters of the ANFIS network, and obtaining an optimized ANFIS network; S414: combining the optimized ANFIS network with the fuzzy rule base to form a complete loom operation status evaluation model.

4. The loom data monitoring method according to claim 3, characterized in that: Step S42 specifically includes: S421: inputting the optimized operation data into the loom operation status evaluation model; S422: calculating the health index of each subsystem of the loom through the forward propagation process of the ANFIS network; S423: normalizing the calculated health index of each subsystem to obtain the normalized health index of each subsystem of the loom.

5. The loom data monitoring method according to claim 4, characterized in that: Step S43 specifically includes: S431: Determine the initial impact weight of each subsystem of the loom on the overall performance of the loom based on the current production task and historical performance data; S432: adjusting the initial impact weight according to the impact degree of the environmental factors on each subsystem of the loom to obtain an adjusted weight coefficient; S433: performing weighted summation of the normalized health index of each subsystem of the loom and the adjusted weight coefficient to obtain an overall health index of the loom; S434: Determine the overall operation status score of the loom according to the overall health index of the loom and a preset scoring standard.

6. The loom data monitoring method according to claim 1, characterized in that: Step S50 specifically includes: S51: The cloud server establishes a predictive maintenance model based on the evaluation result of the operation status of the loom and in combination with the historical maintenance records; S52: the cloud server inputs the real-time operation data of the loom and the current operation status of the loom into the predictive maintenance model to generate a loom maintenance instruction; S53: The cloud server sends the loom maintenance instruction to the terminal device of the maintenance personnel.

7. The loom data monitoring method according to claim 1, characterized in that: The method further comprises: S60: the cloud server optimizes the working parameters of the loom according to the operating data of the loom and the operating status evaluation result; Step S60 specifically includes: S61: The cloud server establishes a loom parameter optimization model using a deep learning algorithm based on the real-time operation data and historical operation records of the loom; S62: the cloud server inputs the real-time operation data and current operation status of the loom into the loom parameter optimization model to generate an adjustment instruction for the loom working parameters; S63: The cloud server sends the adjustment instruction of the loom working parameters to the controller of the loom to adjust the working parameters of the loom.

8. A loom data monitoring system, characterized in that: include: A distributed sensor network to collect loom operation data; an edge computing device, communicatively connected to the distributed sensor network, for optimizing the operation data, obtaining the optimized operation data, and transmitting the optimized operation data to a cloud server; A cloud server is communicatively connected to the edge computing device, and the cloud server includes: A data analysis module, used for analyzing the optimized operation data based on the loom operation status evaluation model to obtain the loom operation status evaluation result; A predictive maintenance module is used to perform predictive maintenance on the loom based on the evaluation result of the running status of the loom.

9. A computer device, characterized in that: The computer device comprises: one or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the loom data monitoring method as described below: Obtain optimized operation data of the loom from the edge computing device; Analyzing the optimized operation data of the loom based on the loom operation status evaluation model to obtain the loom operation status evaluation result; Predictive maintenance is performed on the loom based on the evaluation result of the operation status of the loom.

10. A computer-readable storage medium, characterized in that: The computer readable storage medium stores a computer program, and when the computer program is executed by the processor, the following loom data monitoring method is implemented: Obtain optimized operation data of the loom from the edge computing device; Analyzing the optimized operation data of the loom based on the loom operation status evaluation model to obtain the loom operation status evaluation result; Predictive maintenance is performed on the loom based on the evaluation result of the operation status of the loom.

Images