Production line monitoring method and system applied to digital production

Abstract

The invention discloses a production line monitoring method and system applied to digital production, and the method comprises the steps: installing a sensor on production line equipment, and collecting the operation data of the production line equipment in real time; the operation data is transmitted to the central monitoring module through the industrial internet, the received operation data is processed, and an equipment health state model is established by adopting a big data analysis and machine learning algorithm; using the equipment health state model to monitor the operation state of the production line equipment in real time; when it is detected that the operation parameters of the production line equipment are abnormal or the health state of the equipment is deteriorated, early warning information is sent to related personnel; according to the method, data are processed by adopting big data analysis and a machine learning algorithm, and the equipment health state model is established and optimized, so that potential abnormity and faults of the equipment can be found and early warned in time, measures are taken in advance for prevention, the problems of accidental shutdown and overlong maintenance time are avoided, and the maintenance efficiency is improved. The problems of insufficient fault detection and low production efficiency in a traditional production line are effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of production monitoring, and in particular to a production line monitoring method and system applied to digital production. Background Art

[0002] With the advent of the Industrial 4.0 era, digital transformation has become an inevitable trend in the development of the manufacturing industry, and factory automation is a key link in realizing digital factories. Therefore, it is necessary to upgrade production equipment to be intelligent to adapt to the development needs of the times. In traditional production lines, due to the lack of effective fault diagnosis methods, problems with equipment can only be discovered through manual inspections, which is not only inefficient, but also prone to missed inspections, resulting in serious damage to equipment and even safety accidents. Summary of the invention

[0003] The purpose of the present invention is to provide a production line monitoring method and system for digital production, which effectively solves the problems of insufficient fault detection and low production efficiency in traditional production lines.

[0004] The purpose of the present invention can be achieved through the following technical solutions: The present application provides a production line monitoring method applied to digital production, comprising the following steps: Installing sensors on the production line equipment, wherein the sensors are used to collect operation data of the production line equipment in real time; The operation data is transmitted to a central monitoring module via the industrial Internet; the central monitoring module processes the received operation data and establishes a device health status model using big data analysis and machine learning algorithms; Using the equipment health status model to monitor the operating status of the production line equipment in real time; When the central monitoring module detects that the operating parameters of the production line equipment are abnormal or the health status of the equipment deteriorates, it sends an early warning message to relevant personnel.

[0005] Furthermore, after receiving the operating data, the central monitoring module uses the equipment health status model to analyze the real-time acquired operating data to obtain a first result, wherein the equipment health status model is trained based on historical operating data; when the first result indicates that there is an equipment abnormality, an early warning message is sent to the maintenance terminal.

[0006] Furthermore, before the central monitoring module receives the operating data, it also includes: filtering out noise in the operating data to obtain a second result, performing dimensionality reduction processing on the second result to obtain a third result, dividing the third result into multiple sample sets, each sample set containing at least one sample; inputting the sample set into a neural network to obtain a fourth result, and adjusting the equipment health status model according to the fourth result.

[0007] Further, the noise in the operation data is filtered out to obtain a second result, which specifically includes: Perform median filtering on the original data to remove sharp noise and obtain the result after median filtering; The median filtered data is used as the observation input of the Kalman filter to remove the smooth Gaussian noise, and is predicted and updated to finally obtain the second result.

[0008] Further, the second result is subjected to dimensionality reduction processing to obtain a third result, and the dimensionality reduction processing is performed by a dimensionality reduction method of a principal component analysis method, including: Standardize the data to eliminate the impact of different dimensions and magnitudes; Calculate the covariance matrix of the standardized data to determine the relationship between data features; Calculate the eigenvalues ​​and corresponding eigenvectors of the covariance matrix; According to the size of the eigenvalue, the eigenvectors corresponding to the first m largest eigenvalues ​​are selected, and the original data are projected onto the selected principal component to obtain the reduced-dimensional data.

[0009] Further, the third result is divided into a plurality of sample sets, and the sample sets are input into the neural network to obtain the fourth result. Specifically, the sample sets are input into the network as training data, and forward propagation and back propagation are performed to update the network parameters, which specifically includes: Use forward propagation to calculate the output of each layer until the output layer, and get the prediction results for each sample set; Dividing the third result into multiple sample sets, and inputting each sample set into an input layer of a neural network; wherein the number of nodes in the input layer is equal to the number of features of each sample, and the neural network includes one or more hidden layers, and each hidden layer is nonlinearly transformed through weights and activation functions; Calculate the loss function, which measures the difference between the model prediction and the actual value; Calculate the gradient and update the weights through back propagation, and continuously adjust the parameters to optimize the prediction results; The process of forward propagation, loss calculation, backpropagation, and parameter update is repeated until the performance of the model on the validation set reaches a predetermined number of iterations.

[0010] Furthermore, when the first result indicates that there is an equipment abnormality, an early warning message is sent to the maintenance terminal, including: judging whether the equipment operating parameters are within a threshold range according to preset rules; if they are within the threshold range, continuing monitoring; if they are not within the threshold range, recording the occurrence time and parameter information of the abnormal event; counting the number of consecutive abnormal events N and / or the number of consecutive abnormal events whose time intervals are less than a first set value T1 M; if N ≥ the first set value or M ≥ the second set value, issuing an early warning signal; if N < the first set value and M < the second set value, continuing monitoring.

[0011] A production line monitoring system applied to digital production, applied to a production line monitoring method applied to digital production, specifically comprising: a data acquisition module, a data transmission module, a data processing and analysis module, a real-time monitoring module and an early warning information generation module; The data acquisition module is used to collect operation data from sensors installed on production line equipment in real time; The data transmission module transmits the collected operation data to the central monitoring module via the industrial Internet; The data processing and analysis module is used to process the received operation data and establish a device health status model using big data analysis and machine learning algorithms; The real-time monitoring module uses the equipment health status model to monitor the operating status of the production line equipment in real time, inputs the third result after dimensionality reduction into the neural network, performs forward propagation and back propagation, and optimizes the model parameters; The warning information generation module sends warning information to relevant personnel when it detects that the equipment operating parameters are abnormal or the equipment health status deteriorates.

[0012] Furthermore, a device health status model is established, and the sharp noise in the data is removed by median filtering, and the Gaussian noise is further eliminated by combining Kalman filtering to obtain the second result; The data is then reduced in dimension through principal component analysis to extract key features to obtain the third result, and the historical operating data is then used to train the equipment health status model.

[0013] Furthermore, the warning information generation module also includes judging whether the equipment operating parameters are within a threshold range according to preset rules; if they are within the threshold range, continuing monitoring; if they are not within the threshold range, recording the occurrence time and parameter information of the abnormal event; counting the number of consecutive abnormal events N and / or the number of consecutive abnormal events whose time intervals are less than a first set value T1 M; if N ≥ the first set value or M ≥ the second set value, issuing a warning signal; if N < the first set value and M < the second set value, continuing monitoring.

[0014] The beneficial effects of the present invention are: (1) Big data analysis and machine learning algorithms are used to process data, establish and optimize equipment health status models, and be able to promptly detect and warn of potential abnormalities and failures of equipment, so as to take preventive measures in advance, avoid unexpected downtime and excessive maintenance time, and significantly improve production efficiency and equipment stability. At the same time, the application of equipment health status models provides more accurate and reliable equipment status assessment, enhances the predictive ability and response speed of the monitoring process, ensures the continuity and safety of the production process, and effectively solves the problems of insufficient fault detection and low production efficiency in traditional production lines; (2) When the health status of the equipment deteriorates, it can issue early warnings in a timely manner, allowing the maintenance team to take measures before the problem develops into a serious failure. By statistically analyzing the number N and time interval M of consecutive abnormal events, it can predict potential risks and arrange maintenance in advance, thereby reducing unexpected downtime, avoiding major losses, and effectively reducing maintenance costs. It effectively solves the problem that fault diagnosis methods lead to serious equipment damage, high maintenance costs, and possible safety accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] For better understanding and implementation, the technical solution of the present application is described in detail below with reference to the accompanying drawings.

[0016] Figure 1 A schematic diagram of a process flow of a production line monitoring method for digital production provided in Example 1 of the present application; Figure 2 A schematic diagram of a process for filtering out noise in operation data in a production line monitoring method for digital production provided in Example 1 of the present application; Figure 3 A schematic diagram of a process of performing dimensionality reduction processing by a principal component analysis method in a production line monitoring method for digital production provided in Example 1 of the present application; Figure 4 A schematic diagram of a model flow of inputting a sample set into a neural network for training in a production line monitoring method for digital production provided in Example 1 of the present application. DETAILED DESCRIPTION

[0017] In order to further explain the technical means and effects taken by the present invention to achieve the predetermined invention purpose, exemplary embodiments will be described in detail here, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are only examples of methods and systems consistent with some aspects of the present application as detailed in the attached claims.

[0018] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0019] The specific implementation methods, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0020] Example 1 See also Figure 1-Figure 4 , this embodiment provides a production line monitoring method and system applied to digital production, which effectively solves the problems of insufficient fault detection and low production efficiency in traditional production lines.

[0021] The present invention provides a production line monitoring method applied to digital production, comprising the following steps: S1. Install sensors on the production line equipment, and the sensors are used to collect the operation data of the production line equipment in real time; the operation data of the production line equipment includes parameters such as temperature, vibration, and power consumption; S2. Transmitting the operation data to a central monitoring module via the industrial Internet; the central monitoring module processes the received operation data and establishes a device health status model using big data analysis and machine learning algorithms; S3. Using the equipment health status model to monitor the operating status of the production line equipment in real time; S4. When the central monitoring module detects that the operating parameters of the production line equipment are abnormal or the health status of the equipment deteriorates, it sends an early warning message to relevant personnel.

[0022] Specifically, the method of this embodiment can analyze the operating data of production line equipment during the digital production process and establish an equipment health status model, thereby realizing intelligent operation and maintenance of production line equipment.

[0023] Furthermore, installing sensors on production line equipment includes arranging a variety of sensor devices on the production line, such as thermometers and hygrometers, infrared thermal imagers, acceleration sensors, etc., so as to collect various physical and chemical property data in real time.

[0024] Furthermore, after receiving the operating data, the central monitoring module uses the equipment health status model to analyze the real-time acquired operating data to obtain a first result, wherein the equipment health status model is trained based on historical operating data; when the first result indicates that there is an equipment abnormality, an early warning message is sent to the maintenance terminal.

[0025] Among them, the first result is the result obtained by preliminary analysis of real-time operation data based on the equipment health status model, which can evaluate the current health status of the equipment. This result is directly related to the real-time monitoring and maintenance decisions of the production line equipment. It is the preliminary output of the entire method and reflects the transformation from data to information.

[0026] Specifically, the equipment health status model can be trained through deep learning, supervised learning, or unsupervised learning, and this embodiment does not limit this.

[0027] Furthermore, before the central monitoring module receives the operating data, it also includes: filtering out noise in the operating data to obtain a second result, performing dimensionality reduction processing on the second result to obtain a third result, dividing the third result into multiple sample sets, each sample set containing at least one sample; inputting the sample set into a neural network to obtain a fourth result, and adjusting the equipment health status model according to the fourth result.

[0028] Among them, the second result is that before conducting equipment health status analysis, the data must first be preprocessed to remove noise. The third result is that the second result is processed by dimensionality reduction. The third result reflects the transformation from high-dimensional data to low-dimensional features and is the preparatory stage for model training.

[0029] Specifically, due to the inevitable existence of various interference factors in the actual production environment, the collected data often contains a lot of noise. Therefore, in order to improve the accuracy of the model, the original data needs to be preprocessed first, that is, the noise is filtered out.

[0030] Further, the noise in the operation data is filtered out to obtain a second result, which specifically includes: S21, performing median filtering on the original data to remove sharp noise; Select an odd-sized window and slide the window to each point in the data set. For each data point in the window, find the median and replace the current data point with the median in the window. The specific median filter formula is: in, is the result after median filtering.

[0031] S22. Use the median filtered data as observation input of the Kalman filter, further remove the smoothed Gaussian noise, perform prediction and update, and finally obtain the second result.

[0032] The specific Kalman filter includes: Status Assessment: ;in is the predicted state at step k, is the state transition matrix, is the control matrix, is the control input, is the process noise covariance; Forecast error covariance: ;in, The forecast error covariance matrix before time step k; Mann gain calculation: ,in is the Kalman gain, is the observation model matrix, is the observation noise covariance; Status Update: ,in is the updated state after time step k, is the observed value after median filtering; Covariance update: ,in is the updated error covariance after time step k.

[0033] It should be explained that median filtering is a nonlinear filtering technique used to remove sharp noise in data. Median filtering is achieved by replacing each data point with the median of all data points in the neighborhood of that point. This method is very effective in removing impulse noise (i.e., sudden, short-lived noise). Kalman filtering is a linear filtering technique that uses a series of observations to estimate the state of a dynamic system. The combination of median filtering and Kalman filtering can effectively remove different types of noise from the data, thereby providing more accurate data for subsequent data analysis and model training.

[0034] Further, the second result is subjected to dimensionality reduction processing to obtain a third result, and the dimensionality reduction processing is performed by a dimensionality reduction method of a principal component analysis method, including: S31. Standardize the data to eliminate the impact of different dimensions and magnitudes; The standardized formula is expressed as: ,in is the standardized data, is the original data, is the mean of the ith feature, is the standard deviation of the ith feature.

[0035] S32, calculating the covariance matrix of the standardized data to determine the relationship between the data features; The covariance matrix formula is expressed as: , where C is the covariance matrix, n is the number of samples, is the standardized mean of the ith feature.

[0036] S33, calculating the eigenvalues ​​and corresponding eigenvectors of the covariance matrix, wherein the eigenvectors determine the directions of the principal components, and the eigenvalues ​​determine the sizes of the principal components; The eigenvalues ​​and eigenvectors are expressed as: ,in, is the ith eigenvalue, is the corresponding eigenvector.

[0037] S34. Select the eigenvectors corresponding to the first m largest eigenvalues ​​according to the size of the eigenvalues, where m is the dimension to which the dimensionality is desired to be reduced, and project the original data onto the selected principal component to obtain the data after dimensionality reduction.

[0038] The dimensionality reduction formula is expressed as: ,in It is the data after dimensionality reduction.

[0039] It should be explained that principal component analysis (PCA) is a statistical method used to transform data into a new coordinate system through orthogonal transformation, so as to maximize the variance on the new coordinate axis and achieve dimensionality reduction. In the production line monitoring method of digital production, principal component analysis (PCA) is used as a dimensionality reduction method, mainly because PCA can efficiently extract key features from multidimensional data and project the data to the direction with the largest variance through orthogonal transformation, thereby reducing the data dimension and reducing the complexity of subsequent processing. The benefit of PCA is that it can not only remove redundancy and noise in the data and improve data quality, but also retain the most important information in the data set, which is crucial for real-time monitoring systems because it can speed up data processing, optimize resource allocation, and improve the model's sensitivity and response speed to changes in equipment status, thereby ensuring the stability and efficiency of the production process.

[0040] Further, the third result is divided into a plurality of sample sets, and the sample sets are input into the neural network to obtain the fourth result. Specifically, the sample sets are input into the network as training data, and forward propagation and back propagation are performed to update the network parameters, which specifically includes: S41, use forward propagation to calculate the output of each layer until the output layer, and obtain the prediction result of each sample set; The third result is divided into multiple sample sets, and each sample set is input into an input layer of a neural network; wherein the number of nodes in the input layer is equal to the number of features of each sample, and the neural network includes one or more hidden layers, and each hidden layer is nonlinearly transformed through weights and activation functions.

[0041] There are L hidden layers, and the output of each layer is expressed as: in, It is The output of the layer, It is The weight matrix of the layer, It is Layer bias vector, f is the activation function; The output of the last hidden layer is passed to the output layer to obtain the final prediction result. The calculation formula of the output layer is: in, is the output result of the sample set, is the weight matrix of the output layer, is the bias vector of the output layer.

[0042] S42. Calculate the loss function, expressed as: , where L is the loss function, which measures the difference between the model prediction and the actual value, y is the actual label or output value, and N is the total number of samples used to calculate the mean.

[0043] S43, calculate the gradient and update the weight through back propagation, and continuously adjust the parameters to optimize the prediction results; Use the loss function to calculate the gradient of network parameters, including the output layer weight gradient and the output layer bias gradient; Among them, the output layer weight gradient is expressed as: The output layer bias gradient is expressed as: in, is the partial derivative of the loss function with respect to the predicted output, is the error gradient of the output layer, which can be used to predict the output through the loss function The partial derivative of is obtained.

[0044] Recursively calculate the gradient of each hidden layer using the chain rule; The recursive formula is expressed as: ,in, It is The error gradient of the layer, represents the Hadamard product, is the derivative of the activation function, Representative The previous activation value of the layer, that is, the linear combination result before the activation function is applied.

[0045] The hidden layer weight gradient is expressed as: ; The hidden layer bias gradient is expressed as: ; Update the network's weights and biases based on the gradient and the optimization algorithm.

[0046] The weight update is expressed as: ; The bias update is expressed as: ;in, is the learning rate.

[0047] S44. Repeat the process of forward propagation, loss calculation, back propagation, and parameter update until the performance of the model on the validation set reaches a predetermined number of iterations.

[0048] Among them, the fourth result and the third result are used as training data for the neural network. Through the process of forward propagation and back propagation, the network parameters are continuously optimized until the performance of the model on the validation set reaches the predetermined standard. The fourth result is the final output of the neural network training, which reflects the model's ability to predict the health status of the equipment and is the core of realizing intelligent monitoring in the entire method.

[0049] Specifically, a neural network is a computational model that simulates the working principle of biological neurons. It can automatically extract useful features from a large amount of data and use them to perform tasks such as classification and regression. In this embodiment, the use of a neural network for modeling can improve the generalization ability of the model and make the model more suitable for different application scenarios.

[0050] Furthermore, when the first result indicates that there is an equipment abnormality, an early warning message is sent to the maintenance terminal, including: judging whether the equipment operating parameters are within a threshold range according to preset rules; if they are within the threshold range, continuing monitoring; if they are not within the threshold range, recording the occurrence time and parameter information of the abnormal event; counting the number of consecutive abnormal events N and / or the number of consecutive abnormal events whose time intervals are less than a first set value T1 M; if N ≥ the first set value or M ≥ the second set value, issuing an early warning signal; if N < the first set value and M < the second set value, continuing monitoring; wherein, the first set value and the second set value can be set according to actual conditions.

[0051] Specifically, the method of this embodiment can also count and analyze abnormal events, so as to discover potential risk points in advance and take timely measures to avoid major accidents.

[0052] In summary, the method provided by the present invention can establish an equipment health status model based on the operating data of the production line equipment, using big data analysis and machine learning algorithms, thereby realizing intelligent operation and maintenance of the production line equipment, improving the operation and maintenance efficiency and accuracy of the production line equipment, and reducing labor costs.

[0053] The equipment health status model of this embodiment is established based on historical data to predict the future operating status of the equipment. This model learns the characteristics of the equipment in normal and abnormal states by analyzing various patterns and trends in historical data. The purpose of the model is to predict changes in the equipment status and analyze the current status, including possible failure development and performance degradation of the equipment. When the first result indicates that there is an equipment abnormality, the early warning logic relationship of sending early warning information to the maintenance terminal is designed based on the output of the equipment health status model, and defines under what circumstances the early warning signal should be triggered. The early warning logic uses preset rules or thresholds to determine whether the current status of the equipment is normal. The purpose is to monitor the equipment status in real time and immediately issue an alarm when an abnormality or potential problem is detected so that timely maintenance or repair measures can be taken.

[0054] Example 2 This embodiment is a production line monitoring system applied to digital production, which is applied to a production line monitoring method applied to digital production. It can realize intelligent operation and maintenance of production line equipment, improve production efficiency, reduce maintenance costs, and ensure the stability and safety of the production process.

[0055] Specifically include: data acquisition module, data transmission module, data processing and analysis module, real-time monitoring module and early warning information generation module; Among them, the data acquisition module is used to collect operation data from sensors installed on production line equipment in real time; the equipment operation data of the production line is collected through a variety of sensor devices, such as thermometers and hygrometers, infrared thermal imagers, acceleration sensors, etc.

[0056] Data transmission module, which transmits the collected operation data to the central monitoring module through the industrial Internet; Data processing and analysis module, used to process the received operation data and establish the equipment health status model using big data analysis and machine learning algorithms; Among them, the equipment health status model is established, the sharp noise in the data is removed by median filtering, and the Gaussian noise is further eliminated by combining Kalman filtering to obtain clean data, which is the second result; The data is then reduced in dimension through principal component analysis to extract key features to obtain the third result, and the historical operating data is then used to train the equipment health status model.

[0057] The real-time monitoring module uses the equipment health status model to monitor the operating status of the production line equipment in real time, inputs the third-party results after dimensionality reduction into the neural network, performs forward propagation and back propagation, and optimizes the model parameters; The early warning information generation module sends early warning information to relevant personnel when it detects abnormal equipment operating parameters or deterioration of equipment health status.

[0058] Specifically: determine whether the equipment operating parameters are within the threshold range according to preset rules; if they are within the threshold range, continue monitoring; if they are not within the threshold range, record the occurrence time and parameter information of the abnormal event; count the number of consecutive abnormal events N and / or the number of consecutive abnormal events whose time intervals are less than the first set value T1 M; if N ≥ the first set value or M ≥ the second set value, issue a warning signal; if N < the first set value and M < the second set value, continue monitoring; wherein, the first set value and the second set value can be set according to actual conditions.

[0059] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. Production line monitoring method applied to digital production, Features: Installing sensors on the production line equipment, wherein the sensors are used to collect operation data of the production line equipment in real time; The operation data is transmitted to a central monitoring module via the industrial Internet; the central monitoring module processes the received operation data and establishes a device health status model using big data analysis and machine learning algorithms; Using the equipment health status model to monitor the operating status of the production line equipment in real time; When the central monitoring module detects that the operating parameters of the production line equipment are abnormal or the health status of the equipment deteriorates, it sends an early warning message to relevant personnel.

2. The production line monitoring method for digital production according to claim 1, Features: After receiving the operating data, the central monitoring module uses the equipment health status model to analyze the real-time acquired operating data to obtain a first result, wherein the equipment health status model is trained based on historical operating data; when the first result indicates that there is an equipment abnormality, an early warning message is sent to the maintenance terminal.

3. The production line monitoring method for digital production according to claim 1, Features: Before the central monitoring module receives the operating data, it also includes: filtering out noise in the operating data to obtain a second result, performing dimensionality reduction processing on the second result to obtain a third result, dividing the third result into multiple sample sets, each sample set containing at least one sample; inputting the sample set into a neural network to obtain a fourth result, and adjusting the equipment health status model according to the fourth result.

4. The production line monitoring method for digital production according to claim 3, Features: Filtering out the noise in the operation data to obtain a second result specifically includes: Perform median filtering on the original data to remove sharp noise and obtain the result after median filtering; The median filtered data is used as the observation input of the Kalman filter to remove the smooth Gaussian noise, and is predicted and updated to finally obtain the second result.

5. The production line monitoring method for digital production according to claim 4, Features: Performing dimensionality reduction processing on the second result to obtain a third result, specifically performing dimensionality reduction processing by a dimensionality reduction method of a principal component analysis method, including: Standardize the data to eliminate the impact of different dimensions and magnitudes; Calculate the covariance matrix of the standardized data to determine the relationship between data features; Calculate the eigenvalues ​​and corresponding eigenvectors of the covariance matrix; According to the size of the eigenvalue, the eigenvectors corresponding to the first m largest eigenvalues ​​are selected, and the original data are projected onto the selected principal component to obtain the reduced-dimensional data.

6. The production line monitoring method for digital production according to claim 5, Features: The third result is divided into a plurality of sample sets, and the sample sets are input into a neural network to obtain a fourth result. Specifically, the sample sets are input into the network as training data, and forward propagation and back propagation are performed to update network parameters, which specifically includes: Use forward propagation to calculate the output of each layer until the output layer, and get the prediction results for each sample set; Dividing the third result into multiple sample sets, and inputting each sample set into an input layer of a neural network; wherein the number of nodes in the input layer is equal to the number of features of each sample, and the neural network includes one or more hidden layers, and each hidden layer is nonlinearly transformed through weights and activation functions; Calculate the loss function, which measures the difference between the model prediction and the actual value; Calculate the gradient and update the weights through back propagation, and continuously adjust the parameters to optimize the prediction results; The process of forward propagation, loss calculation, backpropagation, and parameter update is repeated until the performance of the model on the validation set reaches a predetermined number of iterations.

7. The production line monitoring method for digital production according to claim 2, Features: When the first result indicates that there is an equipment abnormality, sending an early warning message to the maintenance terminal includes: judging whether the equipment operating parameters are within a threshold range according to preset rules; if they are within the threshold range, continuing monitoring; if they are not within the threshold range, recording the occurrence time and parameter information of the abnormal event; counting the number of consecutive abnormal events N and / or the number of consecutive abnormal events whose time intervals are less than a first set value T1 M; if N≥the first set value or M≥the second set value, issuing an early warning signal; if N<the first set value and M<the second set value, continuing monitoring.

8. A production line monitoring system for digital production, applied to a production line monitoring method for digital production as claimed in any one of claims 1 to 7, Features: It includes data acquisition module, data transmission module, data processing and analysis module, real-time monitoring module and early warning information generation module; The data acquisition module is used to collect operation data from sensors installed on production line equipment in real time; The data transmission module transmits the collected operation data to the central monitoring module via the industrial Internet; The data processing and analysis module is used to process the received operation data and establish a device health status model using big data analysis and machine learning algorithms; The real-time monitoring module uses the equipment health status model to monitor the operating status of the production line equipment in real time, inputs the third result after dimensionality reduction into the neural network, performs forward propagation and back propagation, and optimizes the model parameters; The warning information generation module sends warning information to relevant personnel when it detects that the equipment operating parameters are abnormal or the equipment health status deteriorates.

9. The production line monitoring system for digital production according to claim 8, Features: Establish a device health status model, remove sharp noise in the data through median filtering, and further eliminate Gaussian noise by combining Kalman filtering to obtain the second result; The data is then reduced in dimension through principal component analysis to extract key features to obtain the third result, and the historical operating data is then used to train the equipment health status model.

10. The production line monitoring system for digital production according to claim 8, Features: The warning information generation module also includes determining whether the equipment operating parameters are within the threshold range according to preset rules; if they are within the threshold range, continuing monitoring; if they are not within the threshold range, recording the occurrence time and parameter information of the abnormal event; Count the number of consecutive abnormal events N and / or the number of times the time interval of consecutive abnormal events is less than the first set value T1 M; if N ≥ the first set value or M ≥ the second set value, issue a warning signal; if N < the first set value and M < the second set value, continue monitoring.

Images