A Method and System for Monitoring the Health Status of Flue Gas Desulfurization Equipment Based on Deep Neural Network

Abstract

The present invention discloses a method and system for monitoring the health status of a flue gas desulfurization device based on a deep neural network, belonging to the field of health status monitoring of flue gas desulfurization devices, including: obtaining the operation data of the flue gas desulfurization device, performing denoising processing on the basis of noise data by wavelet transform to obtain the denoised operation data; extracting multi-dimensional features of the inlet flue gas temperature, sulfur dioxide concentration, and absorbent flow rate from the precisely cleaned operation data, and constructing a feature matrix; according to the feature matrix, using Bayesian optimization to determine the optimal network structure of the deep neural network to obtain the optimized network structure; constructing a deep neural network model by using the optimized network structure, training the model by the backpropagation algorithm, and stopping training if the training error is lower than a preset threshold to obtain a trained model.

Description

Technical Field

[0001] The present invention belongs to the field of flue gas desulfurization equipment health status monitoring, and in particular, relates to a flue gas desulfurization equipment health status monitoring method and system based on a deep neural network. Background Art

[0002] In the actual operation environment of flue gas desulfurization equipment, the collected data often contains a lot of noise and outliers. Directly using it for the training of deep neural network models will lead to a decline in model performance. The data cleaning process needs to remove abnormal data, but under certain operating conditions, such as startup and shutdown or drastic load changes, the desulfurization equipment will fluctuate within the normal range. These fluctuating data are highly similar to abnormal data in certain features. How to accurately distinguish normal fluctuations from real equipment abnormalities is a difficult problem. Furthermore, when building a deep neural network model, the number of hidden layers, the number of neurons in each layer, and the choice of activation function will directly affect the performance of the model.

[0003] In the scenario of health status monitoring of flue gas desulfurization equipment, the operating status of the equipment is affected by many factors, including inlet flue gas temperature, sulfur dioxide concentration, absorbent flow rate, etc., and there are complex nonlinear relationships between these factors. How to determine an optimal network structure to accurately capture these complex nonlinear relationships and avoid overfitting or underfitting of the model is a key issue. In addition, the operating status of the flue gas desulfurization equipment is a dynamic process that changes over time, and historical data has an important impact on the current status. Although the introduction of the long short-term memory layer can capture the long-term dependencies in time series data, it has many internal parameters and a complex training process. How to determine the appropriate number of memory units and the forget gate threshold to balance the long-term memory capacity of the model and real-time computing efficiency is another technical problem that needs to be solved. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a method for monitoring the health status of flue gas desulfurization equipment based on a deep neural network, comprising:

[0005] Acquire flue gas desulfurization equipment operation data, and perform wavelet transform denoising on noise data of the flue gas desulfurization equipment operation data to obtain denoised operation data;

[0006] The denoised operation data is tested by an isolation forest algorithm, and if a data point in the operation data is determined to be abnormal, the abnormal data point is removed to obtain the operation data after preliminary cleaning;

[0007] A dynamic time warping algorithm is used to calculate the characteristic similarity between normal operating condition fluctuations and equipment abnormalities in the operation data after the preliminary cleaning, so as to obtain accurate operation data after cleaning;

[0008] Extract multi-dimensional features of the inlet flue gas temperature, sulfur dioxide concentration, and absorbent flow rate based on the precisely cleaned operation data, and construct a feature matrix;

[0009] According to the feature matrix, use Bayesian optimization to determine the optimal network structure of the deep neural network, and obtain the optimized network structure;

[0010] Construct a deep neural network model using the optimized network structure, train the model through the backpropagation algorithm. If the training error is lower than the preset threshold, stop training to obtain the trained model;

[0011] Introduce a long short-term memory layer into the trained model, determine the number of memory units according to the length of historical data. If the length of historical data is greater than the set value, increase the number of memory units, otherwise reduce the number of memory units to obtain the dynamically adjusted model;

[0012] Based on the dynamically adjusted model, monitor the operation status of the flue gas desulfurization equipment in real time. If the deviation between the monitoring result and the preset status exceeds the threshold, trigger the warning mechanism to obtain the monitoring result of the equipment health status.

[0013] Preferably, the process of obtaining the denoised operation data includes:

[0014] Obtain the operation data of the flue gas desulfurization equipment, and perform denoising processing on the noise data of the operation data of the flue gas desulfurization equipment using wavelet transform to obtain denoised data;

[0015] According to the characteristics of the denoised data, determine whether the data quality meets the preset standard. If the data quality meets the standard, use the support vector machine algorithm to classify the data. If the data quality does not meet the standard, use the principal component analysis algorithm to perform dimensionality reduction processing on the data;

[0016] According to the characteristics of the data after dimensionality reduction, use the random forest algorithm to perform regression analysis on the data. Through the regression analysis result, judge whether the data accuracy reaches the preset threshold. If the data accuracy reaches the threshold, output the final processing result.

[0017] Preferably, the process of obtaining the preliminarily cleaned operation data includes:

[0018] Construct an isolation forest model, train the isolation forest model, and obtain the trained isolation forest model;

[0019] Based on the trained isolation forest model, detect the denoised operation data. If a certain data point in the operation data is determined to be an abnormal data point, mark the data point to obtain the marked operation data;

[0020] According to the labels of the data points in the operation data, the labeled data points are removed to obtain the operation data after removing abnormal data points. Based on the operation data after removing abnormal data points, the number of data points is counted. If the number is lower than a preset threshold, data points are supplemented to obtain the operation data after supplementation;

[0021] Based on the operation data after supplementation, the data point density is calculated. If the data point density is lower than a preset threshold, data point sampling is performed to obtain the operation data after sampling;

[0022] Based on the operation data after sampling, the data point variance is calculated. If the variance is greater than a preset threshold, the data points are smoothed to obtain the operation data after smoothing;

[0023] Based on the operation data after smoothing, a data feature vector is constructed, and dimensionality reduction is performed using the principal component analysis algorithm to obtain the preliminarily cleaned operation data.

[0024] Preferably, the process of obtaining the precisely cleaned operation data includes:

[0025] Based on the operation data in the initial state, a time series matrix is constructed; wherein, each row of the time series matrix represents all the data point values at a time point;

[0026] The time series matrix is segmented using the sliding window method to obtain multiple time series sub-windows;

[0027] Based on each time series sub-window, statistical features are extracted to obtain the statistical feature vector of each sub-window;

[0028] Based on the statistical feature vector, the normal operating condition fluctuation range of each time series sub-window is calculated;

[0029] The operation data of equipment anomalies is obtained, and its statistical features are extracted to obtain the feature vector of equipment anomalies;

[0030] Based on the feature vector of equipment anomalies and the normal operating condition fluctuation range, the dynamic time warping algorithm is used to calculate the feature similarity between each time series sub-window and the feature vector of equipment anomalies to obtain the feature similarity value of each time series sub-window;

[0031] Based on the feature similarity value of each time series sub-window, it is judged whether it is greater than the set threshold. If it is greater than the set threshold, the data points corresponding to the time series sub-window are retained; otherwise, the data points corresponding to the time series sub-window are removed to obtain the precisely cleaned operation data.

[0032] Preferably, the process of constructing the feature matrix includes:

[0033] Obtain multi-dimensional characteristic data of the inlet flue gas temperature, sulfur dioxide concentration, and absorbent flow rate based on the operation data after precise cleaning;

[0034] Based on the obtained multi-dimensional characteristic data, use the data standardization method to normalize the characteristic data to obtain the normalized multi-dimensional characteristic data;

[0035] Perform dimensionality reduction on the normalized multi-dimensional characteristic data through the principal component analysis algorithm, and extract the principal components that can best represent the characteristics of the original data as new features;

[0036] According to the results of the principal component analysis, select the first k principal components whose cumulative variance contribution rate reaches the preset threshold to construct a dimensionality-reduced feature matrix.

[0037] Preferably, according to the feature matrix, the process of using Bayesian optimization to determine the optimal network structure of the deep neural network and obtaining the optimized network structure includes:

[0038] Obtain different sample categories based on the feature matrix;

[0039] Construct a classification model based on the decision tree algorithm, and determine the optimal decision tree parameters through cross-validation;

[0040] Use the classification results of the decision tree classification model as the prior probability, and combine the Bayesian optimization algorithm to search for the optimal network structure parameters of the deep neural network.

[0041] Preferably, the process of obtaining the trained model includes:

[0042] Obtain a training data set. For the training data set, use the optimized network structure to construct a multi-layer deep neural network model, and set appropriate hyperparameters and activation functions;

[0043] Input the training data set into the constructed deep neural network model, and use the backpropagation algorithm to adaptively adjust the weights and thresholds of the neurons in each layer, and continuously optimize the model parameters;

[0044] In each iteration training process, calculate the training error of the current model. If the training error is lower than the preset threshold, it is determined that the model training is completed, stop the training process, and obtain the trained model.

[0045] Preferably, the process of obtaining the dynamically adjusted model includes:

[0046] Obtain the trained model, introduce a long short-term memory layer, and construct a dynamic optimization model;

[0047] Obtain the historical data to be processed and determine the historical length of the data;

[0048] Determine whether the length of the historical data exceeds a preset threshold. If it exceeds the threshold, dynamically increase the number of memory units;

[0049] If it does not exceed the threshold, dynamically reduce the number of memory units;

[0050] According to the adjusted number of memory units, reconfigure the structure and parameters of the long short-term memory layer;

[0051] Embed the reconfigured long short-term memory layer into the initial model to obtain an optimized model with a dynamic memory mechanism introduced.

[0052] On the other hand, the present invention also provides a flue gas desulfurization equipment health status monitoring system based on a deep neural network, including:

[0053] A data acquisition module for acquiring the operation data of the flue gas desulfurization equipment;

[0054] A denoising processing module for performing wavelet transform denoising processing on the noise data of the operation data of the flue gas desulfurization equipment to obtain denoised operation data;

[0055] An anomaly detection module for detecting the denoised operation data through an isolation forest algorithm. If a data point in the operation data is determined to be abnormal, the abnormal data point is removed to obtain preliminarily cleaned operation data;

[0056] A data cleaning module for calculating the feature similarity between the normal operating condition fluctuations and equipment anomalies in the preliminarily cleaned operation data by using a dynamic time warping algorithm to obtain precisely cleaned operation data;

[0057] A feature extraction module for extracting multi-dimensional features of the inlet flue gas temperature, sulfur dioxide concentration, and absorbent flow rate based on the precisely cleaned operation data and constructing a feature matrix;

[0058] A network optimization module for determining the optimal network structure of the deep neural network by using Bayesian optimization according to the feature matrix to obtain an optimized network structure;

[0059] A model training module for constructing a deep neural network model by using the optimized network structure and training the model through a backpropagation algorithm. If the training error is lower than a preset threshold, stop training to obtain a trained model;

[0060] A status monitoring module is used to introduce a long short-term memory layer into the trained model, determine the number of memory units according to the length of historical data. If the length of historical data is greater than a set value, the number of memory units is increased; otherwise, the number of memory units is decreased, resulting in a dynamically adjusted model. Based on the dynamically adjusted model, the operating status of the flue gas desulfurization equipment is monitored in real time. If the deviation between the monitoring result and the preset status exceeds the threshold, an early warning mechanism is triggered to obtain the monitoring result of the equipment health status.

[0061] Compared with the prior art, the present invention has the following advantages and technical effects:

[0062] The present invention discloses a method for monitoring the health status of a flue gas desulfurization equipment based on a deep neural network. The method first performs wavelet transform denoising and anomaly detection on the equipment operation data, and then uses the dynamic time warping algorithm for precise data cleaning. Next, multi-dimensional features are extracted from the cleaned data to construct a feature matrix. Based on this feature matrix, Bayesian optimization is used to determine the optimal structure of the deep neural network, and a long short-term memory layer is introduced to dynamically adjust the number of memory units according to the length of historical data. Finally, the dynamically adjusted model is used to monitor the operating status of the flue gas desulfurization equipment in real time, and when the deviation between the monitoring result and the preset status exceeds the threshold, an early warning mechanism is triggered. By combining multiple data processing and intelligent algorithms, the present invention realizes precise monitoring and timely warning of the operating status of the flue gas desulfurization equipment, effectively improving the safety and stability of the equipment operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The drawings forming a part of this application are used to provide a further understanding of this application. The schematic embodiments and descriptions thereof of this application are used to explain this application and do not constitute an improper limitation to this application. In the drawings:

[0064] Figure 1 is a flowchart of the method for monitoring the health status of a flue gas desulfurization equipment based on a deep neural network according to an embodiment of the present invention;

[0065] Figure 2 is a schematic diagram of the system structure according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] It should be noted that, without conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The following will refer to the drawings and combine the embodiments to detail this application.

[0067] It should be noted that the steps shown in the flowchart of the drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0068] Example 1

[0069] As Figure 1 shown, in this embodiment, a method for monitoring the health status of a flue gas desulfurization device based on a deep neural network is provided, including:

[0070] S101. Obtain the operation data of the flue gas desulfurization device, and perform denoising processing on the noise data by using wavelet transform to obtain the denoised operation data.

[0071] Obtain the operation data of the flue gas desulfurization device, perform denoising processing on the noise part of the data by using wavelet transform to obtain the denoised operation data. According to the characteristics of the denoised data, determine whether the data quality meets the preset standard. If the data quality meets the standard, use the support vector machine algorithm to classify the data. If the data quality does not meet the standard, use the principal component analysis algorithm to reduce the dimension of the data. According to the characteristics of the dimension-reduced data, use the random forest algorithm to perform regression analysis on the data. Through the regression analysis result, judge whether the data accuracy reaches the preset threshold. If the data accuracy reaches the threshold, output the final processing result. If the data accuracy does not reach the threshold, re-obtain the data and perform a new round of processing.

[0072] Specifically, the acquisition of flue gas desulfurization equipment operation data is an important part of environmental monitoring. These data usually include key indicators such as sulfur dioxide concentration, pH value, liquid-gas ratio, etc. However, the original data often has noise interference, which affects the data quality and analysis accuracy. Wavelet transform is an effective denoising method, which can analyze signals in the time and frequency domains simultaneously, retain useful information and remove noise. For example, for sulfur dioxide concentration data, the db4 wavelet basis function can be selected, 5-layer decomposition, and soft threshold method denoising can be used to effectively eliminate high-frequency noise and retain low-frequency trends. The quality assessment of the denoised data is a key step to ensure the reliability of subsequent analysis. The preset standards may include aspects such as data integrity, consistency, and accuracy. If the data quality meets the standards, the support vector machine (SVM) algorithm can be used for data classification. For example, the desulfurization efficiency can be classified according to different operating conditions (such as load level, fuel type), a multi-class SVM model can be established, the kernel function parameters can be optimized, and the classification accuracy can be improved. When the data quality does not meet the standards, principal component analysis (PCA) can be used for dimensionality reduction. PCA can extract the main features in the data and reduce redundant information. In the flue gas desulfurization system, there may be multiple parameters with strong correlation, such as absorption tower pressure, slurry density, etc. Through PCA, these high-dimensional data can be reduced to 2 to 3 principal components, retaining more than 80% of the information, simplifying subsequent analysis. The random forest algorithm is a powerful ensemble learning method suitable for regression analysis. In the prediction of desulfurization efficiency, influencing factors such as inlet sulfur dioxide concentration, absorption tower pH value, liquid-gas ratio, etc. can be selected as feature variables to establish a random forest model. By setting the appropriate number of trees (such as 500 trees) and the minimum number of node samples, stable and reliable prediction results can be obtained. Data accuracy assessment is an important part of model application. The prediction error can be set to less than 5% as the threshold standard. If the threshold is reached, it means that the model performance is good, and the final processing result can be output to guide the optimized operation of the desulfurization equipment. If the threshold is not reached, the data needs to be acquired again, and the sampling frequency, sensor position, etc. may need to be adjusted to improve the data quality.

[0073] S102, using an isolation forest algorithm to detect abnormal data in the denoised operating data, if a data point in the operating data is determined to be abnormal, removing the abnormal data point to obtain preliminarily cleaned operating data.

[0074] Obtain the denoised operation data, train a model according to the Isolation Forest algorithm to obtain the trained Isolation Forest model. Apply the trained Isolation Forest model to detect the operation data. If a certain data point in the operation data is determined to be an abnormal data point, mark this data point to obtain the marked operation data. According to the marks of the data points in the operation data, eliminate the marked data points to obtain the operation data after eliminating abnormal data points. According to the operation data after eliminating abnormal data points, count the number of data points. If the number is lower than a certain preset threshold, supplement the data points to obtain the supplemented operation data. According to the supplemented operation data, calculate the data point density. If the data point density is lower than a certain preset threshold, perform data point sampling to obtain the sampled operation data. According to the sampled operation data, calculate the data point variance. If the variance is greater than a certain preset threshold, perform smoothing processing on the data points to obtain the operation data after smoothing processing. According to the operation data after smoothing processing, construct a data feature vector and perform dimensionality reduction processing using the Principal Component Analysis algorithm to obtain the preliminarily cleaned operation data.

[0075] Specifically, the Isolation Forest algorithm is an anomaly detection method that identifies outliers in a dataset by constructing decision trees. In the processing of flue gas desulfurization equipment operation data, applying this algorithm can effectively detect abnormal data. For example, the pH value of a certain desulfurization equipment usually ranges from 6.5 to 7.5. If data points with pH values of 3 or 10 appear, the Isolation Forest model will mark them as abnormal. After marking the abnormal data points, these points need to be eliminated to ensure data quality. Suppose there are 20 data points marked as abnormal among 1000 data points. After elimination, 980 valid data points remain. If the preset threshold is 950, there is no need to supplement data points; but if the threshold is 1000, 20 data points need to be supplemented to meet the requirements. The data point density reflects the degree of uniformity of data distribution. Taking the desulfurization efficiency data as an example, if within the range of 0% to 100%, the data points are mainly concentrated in the interval of 40% to 60%, and the data in other intervals is sparse, sampling may be required to increase the overall data density. The sampling method can be simple random sampling or stratified sampling to ensure an equal number of data points in each interval. The data point variance reflects the degree of dispersion of the data. For the flue gas flow data, if the variance is too large, it may indicate unstable equipment operation or measurement errors. At this time, smoothing processing is required, such as using the moving average method, to reduce abnormal fluctuations and make the data smoother. Feature vector construction is a key step in data analysis. For the flue gas desulfurization system, the inlet SO2 concentration, desorbent dosage, desulfurization efficiency, etc. can be selected as features. Principal Component Analysis (PCA) can reduce the data dimension and extract the most important features. For example, reducing 10 features to 3 - 4 principal components not only retains most of the information but also simplifies the subsequent analysis process. Through this series of processes, operation data with higher quality and greater representativeness is obtained.

[0076] S103. For the operation data after the preliminary cleaning, use the dynamic time warping algorithm to calculate the feature similarity between the normal operating condition fluctuations and equipment anomalies. If the feature similarity is higher than the preset threshold, retain the data point; otherwise, eliminate the data point to obtain the operation data after precise cleaning.

[0077] Based on the operation data in the initial state, construct a time series matrix. Each row of the time series matrix represents the values of all data points at a time point. Use the sliding window method to divide the time series matrix to obtain multiple time series sub-windows. Each sub-window contains the operation data within a continuous period of time. For each time series sub-window, extract statistical features to obtain the statistical feature vector of each sub-window. The statistical feature vector includes the mean and variance. According to the statistical feature vector, calculate the normal operating condition fluctuation range of each time series sub-window. The normal operating condition fluctuation range is a numerical interval. Obtain the operation data of equipment anomalies, extract its statistical features to get the feature vector of equipment anomalies. The feature vector of equipment anomalies includes the mean and variance. According to the feature vector of equipment anomalies and the normal operating condition fluctuation range, use the dynamic time warping algorithm to calculate the feature similarity between each time series sub-window and the equipment anomalies, and obtain the feature similarity value of each time series sub-window. According to the feature similarity value of each time series sub-window, judge whether it is greater than the set threshold. If it is greater than the set threshold, retain the data points corresponding to the time series sub-window; otherwise, eliminate the data points corresponding to the time series sub-window, and finally obtain the operation data after precise cleaning.

[0078] S104. Extract the multi-dimensional features of the inlet flue gas temperature, sulfur dioxide concentration, and absorbent flow rate from the operation data after precise cleaning, and construct a feature matrix.

[0079] According to the operation data after precise cleaning, obtain multi-dimensional feature data such as the inlet flue gas temperature, sulfur dioxide concentration, and absorbent flow rate. For the obtained multi-dimensional feature data, use the data standardization method to normalize the feature data to eliminate the dimensional differences between different features. Use the principal component analysis algorithm to perform dimensionality reduction on the normalized multi-dimensional feature data, and extract the principal components that can best represent the characteristics of the original data as new features. According to the results of the principal component analysis, select the first k principal components whose cumulative variance contribution rate reaches the preset threshold to construct a dimensionality-reduced feature matrix. Use the K-means clustering algorithm to perform clustering analysis on the dimensionality-reduced feature matrix, and divide the data into different operating conditions according to the clustering results. For each operating condition, use the decision tree algorithm to establish a non-linear relationship model between the inlet flue gas temperature, sulfur dioxide concentration, absorbent flow rate, and desulfurization efficiency. Apply the established desulfurization efficiency prediction model to the actual production process, and predict the desulfurization efficiency based on the real-time collected inlet flue gas parameters to provide a basis for optimized control.

[0080] Specifically, in the desulfurization system, the operation data after precise cleaning lays the foundation for subsequent analysis. First, multi-dimensional characteristic data such as inlet flue gas temperature, sulfur dioxide concentration, and absorbent flow rate are obtained. These characteristic data reflect the key operation parameters of the desulfurization system and directly affect the desulfurization efficiency. For example, the inlet flue gas temperature may fluctuate between 120°C and 180°C, the sulfur dioxide concentration may vary between 2000 mg / m³ and 5000 mg / m³, and the absorbent flow rate may be adjusted between 50 m³ / h and 100 m³ / h. To eliminate the dimensional differences between different characteristics, a data normalization method is used for normalization. This step ensures that characteristics with different dimensions can be compared and analyzed on the same scale. For example, through z-score normalization, each characteristic can be transformed into a standard normal distribution with a mean of 0 and a standard deviation of 1. Next, the principal component analysis (PCA) algorithm is used to reduce the dimension of the normalized multi-dimensional characteristic data. PCA can extract the principal components that best represent the characteristics of the original data, effectively reducing the data dimension while retaining key information. Suppose the first three principal components with a cumulative variance contribution rate reaching 95% are selected. These principal components may respectively represent potential factors such as the comprehensive influence of temperature-concentration, absorbent utilization efficiency, and system stability. After dimensionality reduction, the K-means clustering algorithm is used to perform clustering analysis on the feature matrix. This step can divide the data into different operating conditions, such as high-load, medium-load, and low-load conditions. Under each operating condition, there may be significant differences in the operating characteristics and parameter relationships of the desulfurization system. For each operating condition, a decision tree algorithm is used to establish a non-linear relationship model between the inlet flue gas parameters and the desulfurization efficiency.

[0081] S105. According to the feature matrix, use Bayesian optimization to determine the optimal network structure of the deep neural network, and obtain the optimized network structure.

[0082] According to the feature matrix, the principal component analysis method is used to reduce the dimension of the features, and a reduced-dimensional feature subset is obtained; for the feature subset, the K-means clustering algorithm is used for data clustering to obtain different sample categories; according to the clustered sample categories, a decision tree algorithm is used to construct a classification model, and the optimal decision tree parameters are determined through cross-validation; the classification result of the decision tree classification model is used as the prior probability, and the Bayesian optimization algorithm is combined to search for the optimal network structure parameters of the deep neural network; during the Bayesian optimization search process, the performance of the neural network under the current network structure parameters is used as the optimization objective function, and the optimal network structure is obtained by maximizing the objective function; according to the optimal network structure parameters obtained by Bayesian optimization, a corresponding deep neural network model is constructed and trained on the training data to obtain the final network model; the performance of the trained deep neural network model is evaluated using the test data, and indicators such as the accuracy rate and recall rate of the model are calculated to verify the generalization ability of the model.

[0083] S106. Use the optimized network structure to construct a deep neural network model, and train the model through the backpropagation algorithm. If the training error is lower than the preset threshold, stop the training to obtain the trained model.

[0084] According to the pre-established business knowledge base, data and rules related to the current business scenario are obtained, feature extraction and data preprocessing are performed on them to obtain a standardized training data set. For the training data set, an optimized network structure is used to construct a multi-layer deep neural network model, and appropriate hyperparameters and activation functions are set. The training data set is input into the constructed deep neural network model, and the backpropagation algorithm is used to adaptively adjust the weights and thresholds of the neurons in each layer, and continuously optimize the model parameters. During each iteration of training, calculate the training error of the current model. If the training error is lower than the preset threshold, it is determined that the model training is completed and the training process is stopped. Obtain the trained deep neural network model. For the newly input business data, through the forward propagation calculation of the model, the corresponding output result is obtained. Compare the output result of the deep neural network model with the expected target, calculate its deviation degree, and fine-tune and optimize the parameters of the model according to the deviation degree.

[0085] S107. Introduce a long short-term memory layer into the trained model, determine the number of memory units according to the length of historical data. If the length of historical data is greater than the set value, increase the number of memory units, otherwise decrease the number of memory units to obtain a dynamically adjusted model.

[0086] Obtain the initial model that has completed training, introduce a long short-term memory layer, and construct a dynamic optimization model. Obtain the historical data to be processed and determine the historical length of the data. Determine whether the historical data length exceeds a preset threshold. If it exceeds the threshold, dynamically increase the number of memory units; if it does not exceed the threshold, dynamically reduce the number of memory units. According to the adjusted number of memory units, reconfigure the structure and parameters of the long short-term memory layer. Embed the reconfigured long short-term memory layer into the initial model to obtain an optimized model with a dynamic memory mechanism. Use the optimized model to process and predict new input data, dynamically adjust the memory ability of the model according to the data characteristics, and improve the adaptability and accuracy of the model. Continuously monitor the change in the historical length of the input data, dynamically adjust the number of memory units according to the change trend of the length, realize the adaptive optimization of the model, and maintain the stable performance and improvement of the model.

[0087] S108. Use the dynamically adjusted model to monitor the operating status of the flue gas desulfurization equipment in real time. If the monitoring result deviates from the preset status by more than the threshold, trigger the early warning mechanism to obtain the monitoring result of the equipment health status.

[0088] Based on the historical operation data of the flue gas desulfurization equipment, establish an equipment operation status prediction model using a machine learning algorithm to obtain the dynamically adjusted model parameters; use the dynamically adjusted model to analyze the real-time operation data of the flue gas desulfurization equipment to obtain the monitoring result of the current operating status of the equipment; compare the monitoring result with the preset normal operating status of the equipment, calculate the deviation value, and determine whether the deviation value exceeds the preset threshold; if the deviation value exceeds the preset threshold, trigger the early warning mechanism and send an equipment abnormality early warning signal to the system; according to the early warning signal, obtain the equipment fault diagnosis knowledge base, use the fault diagnosis algorithm to analyze the current operation data of the equipment, and determine the potential fault types and causes of the equipment; according to the diagnosed equipment fault types and causes, match the corresponding treatment measures from the preset fault solution knowledge base to generate an equipment maintenance plan; output and display the equipment health status monitoring result, fault diagnosis result, and maintenance plan to provide a decision-making basis for equipment management for operation and maintenance personnel.

[0089] As Figure 2 shown, this embodiment also provides a flue gas desulfurization equipment health status monitoring system based on a deep neural network, including:

[0090] A data acquisition module for acquiring the operation data of the flue gas desulfurization equipment;

[0091] A denoising processing module for performing wavelet transform denoising processing on the noise data of the operation data of the flue gas desulfurization equipment to obtain the denoised operation data;

[0092] Anomaly detection module, which is used to detect the denoised operation data through the Isolation Forest algorithm. If a data point in the operation data is determined to be abnormal, the abnormal data point is removed to obtain the preliminarily cleaned operation data;

[0093] Data cleaning module, which is used to calculate the feature similarity between the normal working condition fluctuations and equipment anomalies in the preliminarily cleaned operation data by using the Dynamic Time Warping algorithm to obtain the precisely cleaned operation data;

[0094] Feature extraction module, which is used to extract multi-dimensional features of the inlet flue gas temperature, sulfur dioxide concentration, and absorbent flow rate based on the precisely cleaned operation data and construct a feature matrix;

[0095] Network optimization module, which is used to determine the optimal network structure of the deep neural network by using Bayesian optimization according to the feature matrix to obtain the optimized network structure;

[0096] Model training module, which is used to construct a deep neural network model by using the optimized network structure and train the model through the backpropagation algorithm. If the training error is lower than the preset threshold, the training is stopped to obtain the trained model;

[0097] State monitoring module, which is used to introduce a long short-term memory layer into the trained model and determine the number of memory units according to the length of historical data. If the length of historical data is greater than the set value, the number of memory units is increased, otherwise it is decreased to obtain the dynamically adjusted model; based on the dynamically adjusted model, the operation state of the flue gas desulfurization equipment is monitored in real time. If the monitoring result deviates from the preset state by more than the threshold, the early warning mechanism is triggered to obtain the equipment health state monitoring result.

[0098] The above is only the preferred specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for monitoring the health status of a flue gas desulfurization device based on a deep neural network, characterized in that, include: Acquire flue gas desulfurization equipment operation data, and perform wavelet transform denoising on noise data of the flue gas desulfurization equipment operation data to obtain denoised operation data; The denoised operation data is tested by an isolation forest algorithm, and if a data point in the operation data is determined to be abnormal, the abnormal data point is removed to obtain the operation data after preliminary cleaning; A dynamic time warping algorithm is used to calculate the characteristic similarity between normal operating condition fluctuations and equipment abnormalities in the operation data after the preliminary cleaning, so as to obtain accurate operation data after cleaning; Extracting multi-dimensional features of inlet flue gas temperature, sulfur dioxide concentration, and absorbent flow rate based on the operation data after precise cleaning, and constructing a feature matrix; According to the feature matrix, using Bayesian optimization to determine the optimal network structure of the deep neural network to obtain an optimized network structure; A deep neural network model is constructed using the optimized network structure, and the model is trained using a back propagation algorithm. If the training error is lower than a preset threshold, the training is stopped to obtain a trained model. Introducing a long short-term memory layer into the trained model, determining the number of memory units according to the length of historical data, and increasing the number of memory units if the length of historical data is greater than a set value, otherwise reducing the number of memory units to obtain a dynamically adjusted model; Based on the dynamically adjusted model, the operating status of the flue gas desulfurization equipment is monitored in real time. If the deviation between the monitoring result and the preset status exceeds a threshold, an early warning mechanism is triggered to obtain the equipment health status monitoring result; The process of obtaining the dynamically adjusted model includes: Get the trained model, introduce the long short-term memory layer, and build a dynamic optimization model; Obtain the historical data to be processed and determine the historical length of the data; Determine whether the length of historical data exceeds a preset threshold. If so, dynamically increase the number of memory units. If the threshold is not exceeded, the number of memory cells is dynamically reduced; Reconfigure the structure and parameters of the long short-term memory layer according to the adjusted number of memory units; The reconfigured long and short-term memory layer is embedded into the initial model to obtain an optimized model introducing a dynamic memory mechanism; the trained initial model is obtained, the long and short-term memory layer is introduced, a dynamic optimization model is constructed, the historical data to be processed is obtained, the historical length of the data is determined, and whether the length of the historical data exceeds a preset threshold is judged. If it exceeds the threshold, the number of memory units is dynamically increased; if it does not exceed the threshold, the number of memory units is dynamically reduced. According to the adjusted number of memory units, the structure and parameters of the long and short-term memory layer are reconfigured, and the reconfigured long and short-term memory layer is embedded into the initial model to obtain an optimized model introducing a dynamic memory mechanism. The optimized model is used to process and predict new input data, and the memory capacity of the model is dynamically adjusted according to the data characteristics to improve the adaptability and accuracy of the model. The historical length changes of the input data are continuously monitored, and the number of memory units is dynamically adjusted according to the length change trend to achieve adaptive optimization of the model and maintain stable and improved performance of the model.

2. The method according to claim 1, wherein The process of obtaining the denoised operating data includes: Acquire flue gas desulfurization equipment operation data, and perform wavelet transform denoising on noise data of the flue gas desulfurization equipment operation data to obtain denoised data; According to the characteristics of the denoised data, determine whether the data quality meets the preset standards. If the data quality meets the standards, use the support vector machine algorithm to classify the data. If the data quality does not meet the standards, use the principal component analysis algorithm to reduce the dimension of the data. According to the data characteristics after dimensionality reduction, the random forest algorithm is used to perform regression analysis on the data. The regression analysis results are used to determine whether the data accuracy reaches the preset threshold. If the data accuracy reaches the threshold, the final processing result is output.

3. The method according to claim 1, characterized in that, The process of obtaining the operation data after preliminary cleaning includes: Constructing an isolation forest model, training the isolation forest model, and obtaining a trained isolation forest model; Based on the trained isolation forest model, the denoised operating data is tested, and if a data point in the operating data is determined to be an abnormal data point, the data point is marked to obtain marked operating data; According to the marks of the data points in the operation data, the marked data points are removed to obtain the operation data after the abnormal data points are removed, and according to the operation data after the abnormal data points are removed, the number of the data points is counted, and if the number is lower than a preset threshold, the data points are supplemented to obtain the supplemented operation data; According to the supplemented operation data, the data point density is calculated. If the data point density is lower than a preset threshold, data point sampling is performed to obtain the sampled operation data. According to the sampled operating data, the variance of the data point is calculated. If the variance is greater than a preset threshold, the data point is smoothed to obtain the smoothed operating data. According to the smoothed operating data, a data feature vector is constructed, and a principal component analysis algorithm is used to perform dimensionality reduction processing to obtain the operating data after preliminary cleaning.

4. The method according to claim 1, wherein The process of obtaining the operation data after precise cleaning includes: Constructing a time series matrix based on the running data of the initial state; wherein each row of the time series matrix represents all data point values at a time point; The sliding window method is used to segment the time series matrix to obtain multiple time series sub-windows; Based on each time series sub-window, statistical features are extracted to obtain the statistical feature vector of each sub-window; According to the statistical eigenvector, the normal operating condition fluctuation range of each time series sub-window is calculated; Obtain the abnormal operation data of the equipment, extract its statistical features, and obtain the feature vector of the equipment abnormality; According to the feature vector of equipment anomaly and the fluctuation range of normal working condition, the dynamic time warping algorithm is used to calculate the feature similarity between each time series sub-window and the equipment anomaly, and the feature similarity value of each time series sub-window is obtained; According to the feature similarity value of each time series sub-window, it is judged whether it is greater than the set threshold. If it is greater than the set threshold, the data points corresponding to the time series sub-window are retained, otherwise the data points corresponding to the time series sub-window are removed to obtain the accurately cleaned running data.

5. The method according to claim 1, characterized in that, The process of constructing the feature matrix includes: Obtain multi-dimensional characteristic data of the inlet flue gas temperature, sulfur dioxide concentration, and absorbent flow rate based on the operation data after precise cleaning; Based on the obtained multi-dimensional characteristic data, use the data standardization method to normalize the characteristic data to obtain the normalized multi-dimensional characteristic data; Perform dimensionality reduction on the normalized multi-dimensional characteristic data through the principal component analysis algorithm, and extract the principal components that can best represent the characteristics of the original data as new features; According to the results of the principal component analysis, select the first k principal components whose cumulative variance contribution rate reaches the preset threshold to construct a dimensionality-reduced characteristic matrix.

6. The method according to claim 1, characterized in that, According to the said characteristic matrix, the process of using Bayesian optimization to determine the optimal network structure of the deep neural network and obtaining the optimized network structure includes: Obtain different sample categories based on the said characteristic matrix; Construct a classification model based on the decision tree algorithm, and determine the optimal decision tree parameters through cross-validation; Use the classification results of the decision tree classification model as the prior probability, and combine the Bayesian optimization algorithm to search for the optimal network structure parameters of the deep neural network.

7. The method according to claim 1, characterized in that, The process of obtaining the trained model includes: Obtain the training data set. For the training data set, use the optimized network structure to construct a multi-layer deep neural network model, and set appropriate hyperparameters and activation functions; Input the training data set into the constructed deep neural network model, and use the backpropagation algorithm to adaptively adjust the weights and thresholds of each layer of neurons, and continuously optimize the model parameters; In each iteration training process, calculate the training error of the current model. If the training error is lower than the preset threshold, it is determined that the model training is completed, and the training process is stopped to obtain the trained model.

8. A health status monitoring system for a flue gas desulfurization device based on a deep neural network, characterized in that, For implementing the method described in any one of claims 1-7, the system includes: A data acquisition module for acquiring the operation data of the flue gas desulfurization equipment; A denoising processing module for performing denoising processing on the noise data of the operation data of the flue gas desulfurization equipment by using wavelet transform to obtain the denoised operation data; An anomaly detection module for detecting the denoised operation data through the isolation forest algorithm. If a data point in the operation data is determined to be abnormal, the abnormal data point is removed to obtain the preliminarily cleaned operation data; A data cleaning module for calculating the characteristic similarity between the normal working condition fluctuations and equipment anomalies in the preliminarily cleaned operation data by using the dynamic time warping algorithm to obtain the precisely cleaned operation data; A feature extraction module for extracting multi-dimensional features of the inlet flue gas temperature, sulfur dioxide concentration, and absorbent flow rate based on the precisely cleaned operation data, and constructing a feature matrix; A network optimization module for using Bayesian optimization to determine the optimal network structure of the deep neural network according to the said characteristic matrix to obtain the optimized network structure; A model training module for using the optimized network structure to construct a deep neural network model, training the model through the backpropagation algorithm, and stopping training if the training error is lower than the preset threshold to obtain the trained model; The status monitoring module is used to introduce a long short-term memory layer into the trained model, determine the number of memory cells according to the length of historical data. If the length of historical data is greater than the set value, the number of memory cells is increased; otherwise, the number of memory cells is decreased, and a dynamically adjusted model is obtained. Based on the dynamically adjusted model, the operating status of the flue gas desulfurization equipment is monitored in real time. If the deviation between the monitoring result and the preset status exceeds the threshold, the early warning mechanism is triggered, and the monitoring result of the equipment health status is obtained.

Images