Data-driven over-temperature monitoring method and device, storage medium, and terminal

Abstract

The present invention discloses a data-driven over-temperature monitoring method and device, storage medium, and terminal, which relates to the field of industrial process monitoring and data-driven technology, and its main purpose is to solve the problem of low accuracy of over-temperature monitoring of existing waste heat boilers. It mainly includes obtaining the operating data of the target monitoring point in the waste heat boiler within a preset time period, and the real-time temperature measurement value of the target monitoring point; using a pre-trained regression model to predict and process the operating data to obtain a health status prediction value, which is used to characterize the normal operating temperature of the waste heat boiler under the current operating conditions; based on the health status prediction value and the real-time temperature measurement value, the regression model residual is calculated, and the degradation degree of the target monitoring point is calculated based on the regression model residual; based on the matching result of the degradation degree and the preset monitoring conditions, the over-temperature monitoring result of the target monitoring point is determined. It is mainly used to improve the accuracy of over-temperature monitoring of waste heat boilers in waste incineration power generation.

Description

Technical Field

[0001] The present invention relates to the field of industrial process monitoring and data-driven technology, and in particular to a data-driven overtemperature monitoring method and device, storage medium, and terminal for a waste heat boiler for waste incineration power generation. Background Art

[0002] As an effective way to dispose of waste, waste incineration power generation has been widely used around the world. By incinerating domestic waste at high temperatures, not only can the volume of waste be reduced and environmental pollution be reduced, but the heat energy generated can also be used to generate electricity. As a key equipment in the waste incineration power generation system, the waste heat boiler is responsible for converting the high-temperature flue gas generated during the incineration process into steam to drive the steam turbine generator. However, due to the complex composition of the waste and the unstable combustion state, the waste heat boiler faces many challenges during operation, especially the over-temperature problem, which always threatens the operating safety and service life of the waste heat boiler equipment.

[0003] Existing over-temperature monitoring methods usually rely solely on temperature thresholds to identify over-temperature conditions, that is, when the temperature of the waste heat boiler exceeds the temperature threshold, it is confirmed to be in an over-temperature state. However, due to the complex actual operating conditions of the waste heat boiler, if the over-temperature state is identified solely based on the temperature threshold, the temperature detection is often not sensitive enough, which can easily cause false alarms or omissions in over-temperature monitoring, resulting in low accuracy of over-temperature monitoring. Summary of the invention

[0004] In view of this, the present invention provides a data-driven over-temperature monitoring method and device, a storage medium, and a terminal, the main purpose of which is to solve the problem of low accuracy of existing waste heat boiler over-temperature monitoring.

[0005] According to one aspect of the present invention, a data-driven over-temperature monitoring method is provided, comprising:

[0006] Obtaining the operating data of the target monitoring point in the waste heat boiler within a preset time period and the real-time temperature measurement value of the target monitoring point;

[0007] Using a pre-trained regression model to perform prediction processing on the operating data to obtain a health status prediction value, wherein the health status prediction value is used to characterize the normal operating temperature of the waste heat boiler under the current operating condition;

[0008] According to the health state prediction value and the real-time temperature measurement value, a regression model residual is calculated, and according to the regression model residual, a degradation degree of the target monitoring point is calculated;

[0009] The over-temperature monitoring result of the target monitoring point is determined according to the matching result of the degradation degree and the preset monitoring condition.

[0010] Furthermore, the degradation degree of the target monitoring point is obtained by calculating the residual error of the regression model, including:

[0011] Retrieve the residual influence weight and temperature sensitivity weight of the target monitoring point;

[0012] Performing an exponential operation on the regression model residual, and weighting the result of the exponential operation according to the residual influence weight to obtain an exponential residual weighted result;

[0013] Weighting the real-time temperature measurement value according to the temperature sensitivity weight to obtain a temperature weighted result;

[0014] Calculating the degradation degree of the target monitoring point according to the exponential residual weighted result and the temperature weighted result;

[0015] The calculation formula of degradation degree is: ,

[0016] in, represents the degradation degree of the target monitoring point, is the regression model residual, is the real-time temperature measurement value, is the residual influence weight, which is used to measure the impact of the regression model residual on the degradation degree. is the temperature sensitivity weight of the target monitoring point, which is used to reflect the sensitivity of the target monitoring point to the temperature value. 1, It is a fixed value parameter.

[0017] Furthermore, the preset monitoring condition includes a dynamic degradation degree threshold, and the over-temperature monitoring result of the target monitoring point is determined according to the matching result between the degradation degree and the preset monitoring condition, including:

[0018] Retrieving a dynamic degradation degree threshold of the target monitoring point, and if the degradation degree is greater than the dynamic degradation degree threshold, determining that the over-temperature monitoring result of the target monitoring point is abnormal, and generating over-temperature warning information to issue an over-temperature warning to the target monitoring point;

[0019] If the degradation degree is less than or equal to the dynamic degradation degree threshold, the over-temperature monitoring result of the target monitoring point is determined to be normal, and the step of obtaining the operating data of the target monitoring point in the waste heat boiler within a preset time period and the real-time temperature measurement value of the target monitoring point is returned.

[0020] Further, the step of retrieving the dynamic degradation degree threshold of the target monitoring point includes:

[0021] Acquiring environmental data around the target monitoring point;

[0022] From the degradation degree influence coefficient mapping relationship set, identifying the degradation degree influence coefficient matching the operating data and the environmental data, wherein the degradation degree influence coefficient is used to characterize the influence trend of different environmental data on the degradation degree under different operating conditions;

[0023] The dynamic degradation degree threshold is corrected according to the degradation degree influence coefficient, and the dynamic degradation degree threshold is updated according to the corrected dynamic degradation degree threshold.

[0024] Furthermore, before identifying the degradation degree influence coefficient matching the operation data and the environmental data from the degradation degree influence coefficient mapping relationship set, the method further includes:

[0025] Acquiring historical degradation degree data, and historical operation data and historical environment data corresponding to the historical degradation degree;

[0026] Fitting the historical degradation degree data, the historical operation data and the historical environmental data to obtain the influence relationship between the environmental data and the operation data and the degradation degree;

[0027] A degradation degree influence coefficient mapping relationship set is constructed according to the influence relationship, and the degradation degree influence coefficient mapping relationship set includes degradation degree influence coefficients corresponding to different ranges of operating data and different ranges of environmental data.

[0028] Furthermore, before using the pre-trained regression model to perform prediction processing on the operating data, the method further includes:

[0029] Constructing an initial regression model and constructing a training sample set based on the historical operation data of the target monitoring point;

[0030] Using the training sample set to train the initial regression model to obtain a pre-trained regression model;

[0031] After the operation data is predicted using the pre-trained regression model, the method further includes:

[0032] Update the historical operation data at preset time intervals, and reconstruct the training sample set based on the updated historical operation data;

[0033] The pre-trained regression model is fine-tuned using the reconstructed training sample set, so as to perform subsequent prediction processing tasks using the fine-tuned regression model.

[0034] Furthermore, constructing a training sample set based on the historical operation data of the target monitoring point includes:

[0035] Preprocessing the historical operation data to obtain preprocessed historical operation data, wherein the preprocessing includes data completion processing, duplicate value elimination processing and filtering and denoising processing;

[0036] Taking the high-temperature superheater inlet flue gas temperature as the target variable, performing correlation feature screening on the preprocessed operation data to obtain correlation feature data;

[0037] Performing feature generation and feature dimension reduction processing on the associated feature data in sequence to obtain temperature feature data;

[0038] The operating condition labels corresponding to the temperature characteristic data are marked to obtain a training sample set.

[0039] According to another aspect of the present invention, there is provided a data-driven over-temperature monitoring device, comprising:

[0040] An acquisition module is used to acquire the operating data of a target monitoring point in the waste heat boiler within a preset time period, and the real-time temperature measurement value of the target monitoring point;

[0041] A prediction module, used to perform prediction processing on the operating data using a pre-trained regression model to obtain a health status prediction value, wherein the health status prediction value is used to characterize the normal operating temperature of the waste heat boiler under the current operating condition;

[0042] A calculation module, used to calculate the regression model residual according to the health state prediction value and the real-time temperature measurement value, and calculate the degradation degree of the target monitoring point according to the regression model residual;

[0043] The determination module is used to determine the over-temperature monitoring result of the target monitoring point according to the matching result of the degradation degree and the preset monitoring condition.

[0044] According to another aspect of the present invention, a storage medium is provided, in which at least one executable instruction is stored, and the executable instruction enables a processor to perform operations corresponding to the above-mentioned data-driven over-temperature monitoring method.

[0045] According to another aspect of the present invention, there is provided a terminal, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus;

[0046] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the above-mentioned data-driven over-temperature monitoring method.

[0047] By means of the above technical solution, the technical solution provided by the embodiment of the present invention has at least the following advantages:

[0048] The present invention provides a data-driven over-temperature monitoring method and device, storage medium, and terminal. The embodiment of the present invention obtains the operating data of a target monitoring point in a waste heat boiler within a preset time period, and the real-time temperature measurement value of the target monitoring point; uses a pre-trained regression model to predict and process the operating data to obtain a health status prediction value, wherein the health status prediction value is used to characterize the normal operating temperature of the waste heat boiler under the current operating conditions; calculates the regression model residual based on the health status prediction value and the real-time temperature measurement value, and calculates the degradation degree of the target monitoring point based on the regression model residual; determines the over-temperature monitoring result of the target monitoring point based on the matching result of the degradation degree and the preset monitoring conditions, and determines the over-temperature monitoring result based on the difference between the health temperature prediction value and the real-time temperature measurement value, thereby realizing dynamic monitoring of the target monitoring point, avoiding false alarms or missed alarms caused by a fixed temperature threshold, and improving the accuracy of over-temperature monitoring.

[0049] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0051] Figure 1 A flow chart of a data-driven over-temperature monitoring method provided by an embodiment of the present invention is shown;

[0052] Figure 2 Another data-driven over-temperature monitoring method flow chart provided by an embodiment of the present invention is shown;

[0053] Figure 3 A residual data distribution curve diagram corresponding to a test set provided by an embodiment of the present invention is shown;

[0054] Figure 4 A curve diagram showing a degradation degree variation over time provided by an embodiment of the present invention is shown;

[0055] Figure 5 A block diagram of a data-driven over-temperature monitoring device according to an embodiment of the present invention is shown;

[0056] Figure 6A schematic diagram of the structure of a terminal provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0057] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0058] Aiming at the problem of low accuracy of over-temperature monitoring of existing waste heat boilers, the embodiment of the present invention provides a data-driven over-temperature monitoring method, such as Figure 1 As shown, the method includes:

[0059] 101. Obtain operation data of a target monitoring point in a waste heat boiler within a preset time period and a real-time temperature measurement value of the target monitoring point.

[0060] In the embodiment of the present invention, the operating data and real-time temperature measurement values ​​collected by the target monitoring point in the waste heat boiler within the preset time period are used as the data basis for monitoring the over-temperature state of the waste heat boiler. Among them, the over-temperature monitoring object is the waste heat boiler in the waste incineration power generation system. The target monitoring point is a data collection test point for the temperature and operating data of the waste heat boiler. Multiple target monitoring points can be arranged in the waste heat boiler, and can be distributed at the flue gas outlet, high-temperature superheater inlet, heat exchanger inlet and outlet, etc. The real-time temperature measurement value is collected by thermocouples, thermal resistors and other temperature measuring equipment installed at the target monitoring point. The operating data is collected by high-precision sensors, including data related to the air volume of the primary fan, data related to the incinerator, data related to the waste heat boiler, data related to the induced draft fan, steam soot blowing related data, etc., which are not specifically limited in the embodiment of the present invention. The length of the preset time period corresponding to the operating data can be 1 minute or 2 minutes, which can be customized according to application requirements, and is not specifically limited in the embodiment of the present invention.

[0061] 102. Use the pre-trained regression model to perform prediction processing on the operating data to obtain a health status prediction value.

[0062] In an embodiment of the present invention, a pre-trained regression model is used to predict the health status of the target monitoring point under the current operating data conditions to obtain a health status prediction value. The health status prediction value is used to characterize the normal operating temperature of the target monitoring point under the current operating conditions. The current operating conditions are the operating conditions characterized by the operating data of the current target monitoring point. The pre-trained regression model is constructed based on the historical operating data of the target monitoring point and the optimal feature data extracted through data preprocessing and feature engineering. The regression model can predict the expected temperature value corresponding to the health status of the target monitoring point under the current operating data based on the correlation between the health status and temperature learned from the historical operating data, as a reference for determining the health status of the target monitoring point reflected by the real-time temperature measurement value.

[0063] It should be noted that the target monitoring point can be any point distributed in different positions in the waste heat boiler. The characteristics of the relationship between temperature and health status of the target monitoring points at different positions in the waste heat boiler are also different. A regression model is constructed based on the historical operating data of the target monitoring point, so that the regression model can learn the specificity of the relationship between the temperature and health status of the target monitoring point. Based on this regression model, the operating data of the current target monitoring point is predicted, which can make the prediction result closer to the actual value under the healthy state, thereby ensuring the accuracy of the predicted value of the health status. In addition, constructing a regression model based on the operating data of the target monitoring point can also achieve a comprehensive assessment of the health status of the target monitoring point, and the accuracy of the assessment is higher.

[0064] 103. Calculate a regression model residual based on the health status prediction value and the real-time temperature measurement value, and calculate a degradation degree of the target monitoring point based on the regression model residual.

[0065] In the embodiment of the present invention, the real-time temperature measurement value is used as the target output of the regression model to calculate the regression model residual. The regression model residual is used to characterize the difference between the real-time temperature measurement value and the health status prediction value. The temperature value indicated by the health status prediction value is the temperature under a healthy state. If the difference (residual) between the real-time temperature measurement value and the health status prediction value is large, it indicates that the health status of the current target monitoring point is not good. If the difference between the real-time temperature measurement value and the health status prediction value is small, it indicates that the health status of the current target monitoring point is good.

[0066] In one embodiment of the present invention, for further explanation and limitation, the step of calculating the degradation degree of the target monitoring point according to the regression model residual includes:

[0067] Retrieve the residual influence weight and temperature sensitivity weight of the target monitoring point.

[0068] An exponential operation is performed on the regression model residual, and the result of the exponential operation is weighted according to the residual influence weight to obtain an exponential residual weighted result.

[0069] The real-time temperature measurement value is weighted according to the temperature sensitivity weight to obtain a temperature weighted result.

[0070] The degradation degree of the target monitoring point is calculated based on the exponential residual weighted result and the temperature weighted result.

[0071] In the embodiment of the present invention, by calculating the degradation degree, the health status of the target monitoring point can be dynamically tracked. The calculation of the degradation degree combines the comprehensive weight of the residual and the real-time temperature measurement value, so it can reflect the actual operating status of the target monitoring point and the severity of the temperature anomaly. The calculation formula of the degradation degree is:

[0072] (1);

[0073] in, represents the degradation degree of the target monitoring point, The larger the value is, the greater the degree of degradation is and the more serious the abnormality of the target monitoring point is, and vice versa. is the regression model residual, - (2); among which, is the real-time temperature measurement value, Predicted value for health status. is the weight of the residual part, that is, the residual influence weight, which is used to measure the impact of the residual on the degradation degree. It is the temperature sensitivity weight of the target monitoring point, which is used to reflect the sensitivity of the target monitoring point to the temperature value. 1. It is a fixed value parameter. , , The overall design goal is to make The value of is generally between 0 and 1. The specific value can be customized and adjusted according to specific application requirements to meet the needs of different application scenarios, and is not specifically limited in the embodiments of the present invention.

[0074] It should be noted that by using the exponential residual weighting method in the calculation of the degradation degree, the exponential residual weighting can achieve that when the residual is small, its impact on the degradation degree is small, ensuring the robustness of the system and tolerance to small fluctuations. However, when the residual is greater than a certain threshold, the degradation degree will increase exponentially, rapidly amplifying the impact of abnormal conditions, indicating potential temperature anomalies or equipment failure risks. In addition to the regression model residual, the real-time temperature measurement value of the target monitoring point is also an important factor affecting the degradation degree. Higher temperature values ​​often mean that the equipment is operating in a higher risk state, so the higher the temperature, the more sensitive the system is to degradation. By appropriately weighting the temperature value, the degradation degree can accurately reflect the degree of abnormality in different temperature ranges. This design enables the degradation degree to capture potential risks more sensitively when the temperature value is high.

[0075] 104. Determine an over-temperature monitoring result of the target monitoring point according to a matching result between the degradation degree and a preset monitoring condition.

[0076] In an embodiment of the present invention, the degradation degree of the target monitoring point is calculated based on the residual of the regression model, the preset monitoring condition is the degradation degree threshold, and the over-temperature monitoring state is determined based on the comparison result between the degradation degree and the dynamic degradation degree threshold. Based on the healthy operating state of the target monitoring point as a premise assumption, the temperature value under the healthy state is predicted based on the operating data of the target monitoring point, and the abnormal trend of the target monitoring point can be identified. The over-temperature monitoring result is determined based on the difference between the healthy temperature prediction value and the real-time temperature measurement value. Since the healthy temperature prediction value changes dynamically with the operating data, it can avoid false alarms or omissions caused by the fixed temperature threshold, thereby improving the accuracy of over-temperature monitoring.

[0077] In an embodiment of the present invention, for further explanation and limitation, as Figure 2 As shown, the step of determining the over-temperature monitoring result of the target monitoring point according to the matching result of the degradation degree and the preset monitoring condition includes:

[0078] 201. Retrieve the dynamic degradation degree threshold of the target monitoring point. If the degradation degree is greater than the dynamic degradation degree threshold, determine that the over-temperature monitoring result of the target monitoring point is abnormal, and generate over-temperature warning information to issue an over-temperature warning to the target monitoring point.

[0079] 202. If the degradation degree is less than or equal to the dynamic degradation degree threshold, it is determined that the over-temperature monitoring result of the target monitoring point is normal, and the step of obtaining the operating data of the target monitoring point in the waste heat boiler within a preset time period and the real-time temperature measurement value of the target monitoring point is returned.

[0080] In an embodiment of the present invention, the preset monitoring condition includes a dynamic degradation degree threshold, that is, the over-temperature monitoring result is determined based on the comparison result of the degradation degree of the target monitoring point and the dynamic degradation degree threshold. When the degradation degree is greater than the dynamic degradation degree threshold, it indicates that the target monitoring point has an over-temperature anomaly, then an over-temperature warning message is generated and sent to the monitoring terminal to prompt the existence of the over-temperature situation and to maintain the abnormal position in time. Specifically, the warning information can be delivered to relevant personnel through the front-end interface, SMS notification or email. Among them, the over-temperature warning information may include the location of the target monitoring point indicated by the over-temperature warning, the specific data of the over-temperature anomaly, the over-temperature warning level and other contents, so as to help the on-site personnel take timely countermeasures. When the degradation degree is less than or equal to the dynamic degradation degree threshold, it indicates that the target monitoring point does not have an over-temperature anomaly, then the operating data of the target monitoring point in the waste heat boiler within the preset time period and the real-time temperature measurement value of the target monitoring point can be re-acquired to realize the cyclic monitoring of the target monitoring point.

[0081] It should be noted that the degree of degradation of the target monitoring point is determined based on the exponential weighting result of the regression model residual. By performing exponential weighting on the regression model residual, the difference between the real-time temperature measurement value and the health status prediction value of the target monitoring point can be appropriately amplified, especially in a high temperature environment, so that the degree of degradation can show higher sensitivity to the abnormality of the target monitoring point, thereby improving the timeliness of monitoring. At the same time, by fully exploring the potential classification of historical operating data and perceiving temperature anomalies in advance, the early warning capability of over-temperature failures can be improved, reducing the probability of equipment failure. In addition, the present application introduces a dynamic threshold adjustment mechanism, that is, the dynamic degradation threshold is adaptively adjusted according to the historical degradation degree of the target monitoring point, environmental factors (such as external temperature, load changes, etc.), and operating conditions, so as to ensure that the alarm mechanism can capture anomalies in a timely manner and avoid unnecessary false alarms.

[0082] In one embodiment of the present invention, for further explanation and limitation, before using the pre-trained regression model to perform prediction processing on the operating data, the method further includes:

[0083] Constructing an initial regression model and constructing a training sample set based on the historical operation data of the target monitoring point;

[0084] Using the training sample set to train the initial regression model to obtain a pre-trained regression model;

[0085] After the operation data is predicted using the pre-trained regression model, the method further includes:

[0086] Update the historical operation data at preset time intervals, and reconstruct the training sample set based on the updated historical operation data;

[0087] The pre-trained regression model is fine-tuned using the reconstructed training sample set, so as to perform subsequent prediction processing tasks using the fine-tuned regression model.

[0088] In an embodiment of the present invention, in order to ensure the accuracy and robustness of the regression model, the regression model is trained based on the operating data within a period of historical time, so that the trained regression model is applied to the prediction task of the next time period, and the training sample set of the regression model is updated according to the preset time interval, and the regression model is fine-tuned. Among them, the preset time interval can be 10 days, one month, and it is also customized according to the application requirements, and the embodiment of the present invention does not make specific limitations. The algorithm model of the regression model can be any one of a linear regression model, a least absolute shrinkage and selection operator (Lasso: Least Absolute Shrinkage and Selection Operator) regression model, a support vector machine, a random forest, a gradient boosting decision tree (XGBoost: Gradient Boosting Decision Tree), LightGBM (LightGradient Boosting Machine) and a neural network model. Specifically, a suitable algorithm model can be selected according to the specific application scenario and requirements. For example, linear regression is selected for scenarios that process simple linear relationships to improve computational efficiency and reduce the difficulty of interpretation. For application scenarios with high data dimensions, the Lasso regression model is selected, which can process high-dimensional data based on its added L1 regularization, which helps to automatically select features and reduce the risk of overfitting. For scenarios with complex nonlinear relationships, especially in scenarios with small samples and high-dimensional data, the selection of support vector machines can make the model have better generalization capabilities. For scenarios with more noisy data, the use of random forests can use multiple decision trees for modeling, so that the regression model has better noise resistance and robustness to data distribution. For scenarios with large data scales, the XGBoost model or the LightGBM model can be used. For complex and highly nonlinear temperature prediction scenarios, the selection of neural network models can better mine implicit patterns in the data and ensure the prediction accuracy of the model.

[0089] In one embodiment of the present invention, for further explanation and limitation, the step of constructing a training sample set based on the historical operation data of the target monitoring point includes:

[0090] Preprocessing the historical operation data to obtain preprocessed historical operation data;

[0091] Taking the high-temperature superheater inlet flue gas temperature as the target variable, performing correlation feature screening on the preprocessed operation data to obtain correlation feature data;

[0092] Performing feature generation and feature dimension reduction processing on the associated feature data in sequence to obtain temperature feature data;

[0093] The operating condition labels corresponding to the temperature characteristic data are marked to obtain a training sample set.

[0094] In an embodiment of the present invention, the originally collected operation data may contain missing values, duplicate values ​​and noise due to sensor failure, environmental interference or communication delay. In order to ensure the integrity and consistency of the data, the operation data is preprocessed. The preprocessing includes data completion processing, duplicate value elimination processing and filtering and denoising processing. Among them, the completion processing is the missing value filling, and the missing data is reasonably supplemented by interpolation, mean filling or inference algorithm based on historical data, so as to ensure the continuity of the operation data. Duplicate value elimination is the analysis and processing of duplicate data appearing in the collection process through timestamps or data values. It is used to remove redundant records and maintain data uniqueness. Filtering and denoising processing is to use filtering algorithms (such as Kalman filtering, wavelet denoising, etc.) or outlier detection based on statistical methods to eliminate abnormal data interfered by environmental noise to improve the accuracy of the data.

[0095] In order to improve the predictive performance and generalization ability of the regression model, it is necessary to perform feature engineering on the preprocessed operating data. Specifically, it includes: correlation feature screening, i.e. feature selection, feature generation and feature dimensionality reduction. Among them, through statistical analysis or machine learning algorithms, key features that are highly correlated with the target variable are screened out to reduce the data dimension and avoid the interference of noise features on the regression model. Preferably, the flue gas temperature at the inlet of the high-temperature superheater in the waste heat boiler is selected as the target variable. Feature generation generates more predictive features through mathematical transformation or construction of combined features to mine implicit information in the data. Feature dimensionality reduction is the use of dimensionality reduction technology to compress high-dimensional data into low-dimensional features to reduce computational complexity, thereby increasing the training speed of the model while retaining the main information of the data. By performing feature engineering on historical operating data and constructing a regression model based on the key features in the operating data, the prediction accuracy of the regression model and the robustness of the model can be ensured.

[0096] After obtaining the temperature characteristic data, the corresponding operating condition label is marked for the temperature characteristic data. The operating condition can be extracted from the operating data. The regression model is trained by training samples marked with operating conditions, so that the regression model can predict the health status value in combination with the operating conditions, thereby accurately predicting the normal operating temperature under the operating condition corresponding to the current operating data.

[0097] It should be noted that before using the pre-trained regression model to predict the operation data, the operation data also needs to be pre-processed and feature-engineered. The specific processing process is the same as the processing process of historical operation data, and the embodiment of the present invention will not be repeated here. Through the cooperation of data pre-processing and feature engineering, structured and efficient input data is provided for the subsequent evaluation of degradation degree and over-temperature monitoring results, so that the regression model can maintain good prediction performance and stability in a complex operating environment.

[0098] In one embodiment of the present invention, for further explanation and limitation, the step of retrieving the dynamic degradation degree threshold of the target monitoring point includes:

[0099] Acquiring environmental data around the target monitoring point;

[0100] From the degradation degree influence coefficient mapping relationship set, identifying the degradation degree influence coefficient matching the operation data and the environmental data;

[0101] The dynamic degradation degree threshold is corrected according to the degradation degree influence coefficient, and the dynamic degradation degree threshold is updated according to the corrected dynamic degradation degree threshold.

[0102] In the embodiment of the present invention, the dynamic degradation degree threshold is a threshold value configured corresponding to the target monitoring point, and its initial value can be set according to the operation history and engineering experience of the waste heat boiler, and dynamically optimized with the update of the operation data and the change of the environmental data. Specifically, the corresponding degradation degree influence coefficient can be matched from the degradation degree influence coefficient mapping relationship set according to the current operation data and environmental data. Among them, the degradation degree influence coefficient mapping relationship set includes the degradation degree influence coefficients corresponding to the operation data of different ranges and the environmental data of different ranges. The degradation degree influence coefficient is used to characterize the influence trend of different environmental data on the degradation degree under different operating conditions. The degradation degree influence coefficient can correct the dynamic degradation degree threshold based on the product operation, or it can correct the dynamic degradation degree threshold based on the exponential relationship, and the embodiment of the present invention does not make specific restrictions. For example, when the operating conditions represented by the operation data are good and the environment has no effect on the degradation, the degradation degree influence coefficient is a value that can make the corrected dynamic degradation degree threshold greater than the previous dynamic degradation degree threshold. When the operating conditions represented by the operating data are poor and the environment has a negative impact on degradation, the degradation degree influence coefficient is a value that can make the corrected dynamic degradation degree threshold value smaller than the previous dynamic degradation degree threshold value, thereby achieving dynamic optimization of the dynamic degradation degree threshold value.

[0103] In one embodiment of the present invention, for further explanation and limitation, before identifying the degradation degree influence coefficient matching the operation data and the environmental data from the degradation degree influence coefficient mapping relationship set, the method further includes:

[0104] Acquiring historical degradation degree data, and historical operation data and historical environment data corresponding to the historical degradation degree;

[0105] Fitting the historical degradation degree data, the historical operation data and the historical environmental data to obtain the influence relationship between the environmental data and the operation data and the degradation degree;

[0106] A degradation degree influence coefficient mapping relationship set is constructed according to the influence relationship.

[0107] In an embodiment of the present invention, a linear or nonlinear method can be selected for data fitting based on the relationship between historical operation data, historical environmental data, and historical degradation data to extract the correlation between different environmental data and different operation data on the change in degradation. The operation data and environmental data are divided into multiple range intervals according to the correlation, and a corresponding degradation influence coefficient is configured for each environmental data range and operation data range to obtain a degradation influence coefficient mapping relationship set. By fitting the historical data, the influence relationship between the operation data and the environmental data on the degradation degree is extracted, which can fully explore the potential laws of the historical operation data and perceive the influence trend of the environment and the operation status on the degradation degree in advance. A mapping relationship set is constructed based on this influence relationship, so that during the monitoring process, the dynamic degradation threshold is dynamically adjusted according to the current operation data and environmental data, thereby improving the adaptability of the dynamic degradation threshold to the operating conditions and the environment.

[0108] In one example, a waste incineration power plant is taken as an example to illustrate the specific implementation method of over-temperature monitoring of the waste heat boiler with the over-temperature monitoring system as the execution body. Among them, the over-temperature monitoring system includes a data processing unit and an intelligent sensing module. The over-temperature monitoring system interacts with the existing distributed control system (DCS: Distributed Control System) of the power plant through the OPC (OLE for Process Control) communication protocol to obtain 1 minute of operating data of the incinerator and waste heat boiler from June 1, 2024 to July 10, 2024, including but not limited to primary fan air volume related data, incinerator related data, waste heat boiler related data, induced draft fan related data, steam soot blowing related data, etc. The collected data is transmitted to the data processing unit through the database for processing.

[0109] The operation data processing process includes preprocessing the collected real-time operation data of the boiler, mainly including missing value filling, outlier removal and noise filtering. After the data preprocessing is completed, the system extracts the key parameters in the operation of the boiler through the feature engineering module, including the temperature gradient of each level of heat exchange equipment, the temperature change rate of the incinerator, etc. Finally, 109 features related to the high-temperature superheater inlet flue gas temperature were selected to provide feature data support for the training and prediction of the subsequent regression model.

[0110] Based on the processed operating data, a regression model of the high-temperature superheater inlet flue gas temperature was established. The model uses historical temperature data and related operating conditions as input variables. After multiple iterations and optimization, the root mean square error of the regression model test set is 2.4°C and the determination coefficient is 0.94, indicating that the better the model fits the data, the more it can be used for subsequent predictions.

[0111] The regression model was used to obtain a more accurate prediction value of the high temperature superheater inlet flue gas temperature. , compare the real-time measured temperature value with the predicted value of the regression model, and calculate the residual between the two , - During the regression model training process, the residual data distribution curve corresponding to the test set is as follows: Figure 3 shown.

[0112] The trained regression model is used to support the intelligent perception module to monitor over-temperature conditions from July 10, 2024 to July 23, 2024. During the actual operation of the system, the degradation threshold is set to 0.6 to distinguish between the normal and abnormal states of the equipment. Figure 4 As shown in the figure, on July 13, the system detected that the degradation degree of the boiler temperature measurement point began to rise significantly, and quickly approached the set threshold of 0.6 in a short period of time. At this time, the system did not trigger an alarm. On July 19, the degradation index exceeded 0.6. When the degradation degree exceeded 0.6, the system immediately issued an over-temperature alarm signal and displayed the temperature change trend and degradation data of the relevant measurement points to the on-site operator. After receiving the alarm, the on-site operator immediately checked the relevant equipment. After investigation, it was found that the abnormal cause was the failure of the cooling tower fan, which indirectly affected the overall thermal balance of the boiler and the system operation efficiency, resulting in an over-temperature alarm. After the initial failure of the cooling tower fan, the system's intelligent perception module continued to monitor the operating status of the equipment and recorded the change trend of the degradation degree. As time went on, the operating condition of the cooling tower fan further deteriorated. The fan finally stopped operating completely on July 29. The on-site operator immediately took emergency measures to reduce the load and reduce the combustion intensity of the boiler. At the same time, the cooling tower fan was repaired urgently. After the repair was completed, the degradation degree returned to less than the threshold. According to the degradation degree threshold defined in this example, the equipment abnormality can be detected 10 days in advance.

[0113] The present invention provides an over-temperature monitoring method based on data-driven. An embodiment of the present invention obtains operating data of a target monitoring point in a waste heat boiler within a preset time period and a real-time temperature measurement value of the target monitoring point; uses a pre-trained regression model to predict and process the operating data to obtain a health status prediction value, wherein the health status prediction value is used to characterize the normal operating temperature of the waste heat boiler under the current operating conditions; calculates a regression model residual based on the health status prediction value and the real-time temperature measurement value; determines an over-temperature monitoring result of the target monitoring point based on a matching result of the regression model residual and a preset monitoring condition, and determines the over-temperature monitoring result based on the difference between the health temperature prediction value and the real-time temperature measurement value, thereby realizing dynamic monitoring of the target monitoring point and avoiding false alarms or missed alarms caused by a fixed temperature threshold, thereby improving the accuracy of over-temperature monitoring.

[0114] Furthermore, as a response to the above Figure 1 The embodiment of the present invention provides a data-driven over-temperature monitoring device, such as Figure 5 As shown, the device comprises:

[0115] The acquisition module 31 is used to acquire the operation data of the target monitoring point in the waste heat boiler within a preset time period, and the real-time temperature measurement value of the target monitoring point;

[0116] A prediction module 32, used for predicting the operation data using a pre-trained regression model to obtain a health status prediction value, wherein the health status prediction value is used to characterize the normal operating temperature of the waste heat boiler under the current operating condition;

[0117] A calculation module 33, used to calculate the regression model residual according to the health state prediction value and the real-time temperature measurement value, and calculate the degradation degree of the target monitoring point according to the regression model residual;

[0118] The determination module 34 is used to determine the over-temperature monitoring result of the target monitoring point according to the matching result between the degradation degree and the preset monitoring condition.

[0119] Furthermore, the calculation module includes:

[0120] A calling unit, used to call the residual influence weight and temperature sensitivity weight of the target monitoring point;

[0121] A first calculation unit is used to perform an exponential operation on the regression model residual, and weight the result of the exponential operation according to the residual influence weight to obtain an exponential residual weighted result;

[0122] A second calculation unit, configured to weight the real-time temperature measurement value according to the temperature sensitivity weight to obtain a temperature weighted result;

[0123] A third calculation unit, configured to calculate the degradation degree of the target monitoring point according to the exponential residual weighted result and the temperature weighted result;

[0124] The calculation formula of degradation degree is: ;

[0125] in, represents the degradation degree of the target monitoring point, is the regression model residual, is the real-time temperature measurement value, is the residual influence weight, which is used to measure the impact of the regression model residual on the degradation degree; is the temperature sensitivity weight of the target monitoring point, which is used to reflect the sensitivity of the target monitoring point to the temperature value. 1, It is a fixed value parameter.

[0126] Furthermore, the determination module 34 includes:

[0127] an early warning unit, configured to retrieve a dynamic degradation degree threshold of the target monitoring point, and if the degradation degree is greater than the dynamic degradation degree threshold, determine that the over-temperature monitoring result of the target monitoring point is abnormal, and generate over-temperature early warning information to issue an over-temperature early warning to the target monitoring point;

[0128] A determination unit is used to determine that the over-temperature monitoring result of the target monitoring point is normal if the degradation degree is less than or equal to the dynamic degradation degree threshold, and return to the step of obtaining the operating data of the target monitoring point in the waste heat boiler within a preset time period and the real-time temperature measurement value of the target monitoring point.

[0129] Furthermore, in a specific application scenario, the early warning unit is specifically used for:

[0130] Acquiring environmental data around the target monitoring point;

[0131] From the degradation degree influence coefficient mapping relationship set, identifying the degradation degree influence coefficient matching the operating data and the environmental data, wherein the degradation degree influence coefficient is used to characterize the influence trend of different environmental data on the degradation degree under different operating conditions;

[0132] The dynamic degradation degree threshold is corrected according to the degradation degree influence coefficient, and the dynamic degradation degree threshold is updated according to the corrected dynamic degradation degree threshold.

[0133] Furthermore, the device also includes:

[0134] The acquisition module 31 is further used to acquire historical degradation degree data, and historical operation data and historical environment data corresponding to the historical degradation degree;

[0135] A fitting module, used to fit the historical degradation degree data, the historical operation data and the historical environment data to obtain the influence relationship between the environment data and the operation data and the degradation degree;

[0136] The first construction module is used to construct a degradation degree influence coefficient mapping relationship set according to the influence relationship, and the degradation degree influence coefficient mapping relationship set includes degradation degree influence coefficients corresponding to different ranges of operating data and different ranges of environmental data.

[0137] Furthermore, the device also includes:

[0138] The second construction module is used to construct an initial regression model and construct a training sample set based on the historical operation data of the target monitoring point;

[0139] A training module, used to train the initial regression model using the training sample set to obtain a pre-trained regression model;

[0140] The sample update module is used to update the historical operation data at a preset time interval and reconstruct the training sample set based on the updated historical operation data;

[0141] The retraining module is used to retrain the pre-trained regression model using the reconstructed training sample set, so as to perform subsequent prediction processing tasks using the retrained regression model.

[0142] Furthermore, the second building block includes:

[0143] A preprocessing unit, used to preprocess the historical operation data to obtain preprocessed historical operation data, wherein the preprocessing includes data completion processing, duplicate value elimination processing and filtering and denoising processing;

[0144] A feature screening unit is used to perform correlation feature screening on the pre-processed operation data with the high-temperature superheater inlet flue gas temperature as the target variable to obtain correlation feature data;

[0145] A feature processing unit, used for sequentially performing feature generation and feature dimension reduction processing on the associated feature data to obtain temperature feature data;

[0146] The marking unit is used to mark the corresponding operating condition labels for the temperature characteristic data to obtain a training sample set.

[0147] According to one embodiment of the present invention, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the computer executable instruction can execute the data-driven over-temperature monitoring method in any of the above method embodiments.

[0148] The present invention provides an over-temperature monitoring device based on data-driven. An embodiment of the present invention obtains operating data of a target monitoring point in a waste heat boiler within a preset time period and a real-time temperature measurement value of the target monitoring point; uses a pre-trained regression model to predict and process the operating data to obtain a health status prediction value, wherein the health status prediction value is used to characterize the normal operating temperature of the waste heat boiler under the current operating conditions; calculates the regression model residual based on the health status prediction value and the real-time temperature measurement value, and calculates the degradation degree of the target monitoring point based on the regression model residual; determines the over-temperature monitoring result of the target monitoring point based on the matching result of the degradation degree and the preset monitoring condition, and determines the over-temperature monitoring result based on the difference between the health temperature prediction value and the real-time temperature measurement value, thereby realizing dynamic monitoring of the target monitoring point and avoiding false alarms or missed alarms caused by a fixed temperature threshold, thereby improving the accuracy of over-temperature monitoring.

[0149] Figure 6 A schematic diagram of the structure of a terminal provided according to an embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the terminal.

[0150] like Figure 6 As shown, the terminal may include: a processor (processor) 402 , a communication interface (Communications Interface) 404 , a memory (memory) 406 , and a communication bus 408 .

[0151] The processor 402 , the communication interface 404 , and the memory 406 communicate with each other via a communication bus 408 .

[0152] The communication interface 404 is used for network communication with other devices such as a client or other servers.

[0153] The processor 402 is used to execute the program 410, and specifically can execute the relevant steps in the above-mentioned data-driven over-temperature monitoring method embodiment.

[0154] Specifically, the program 410 may include program codes, which include computer operation instructions.

[0155] The processor 402 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiment of the present invention. The one or more processors included in the terminal may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0156] The memory 406 is used to store the program 410. The memory 406 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0157] The program 410 may be specifically configured to enable the processor 402 to perform the following operations:

[0158] Obtaining the operating data of the target monitoring point in the waste heat boiler within a preset time period and the real-time temperature measurement value of the target monitoring point;

[0159] Using a pre-trained regression model to perform prediction processing on the operating data to obtain a health status prediction value, wherein the health status prediction value is used to characterize the normal operating temperature of the waste heat boiler under the current operating condition;

[0160] According to the health state prediction value and the real-time temperature measurement value, a regression model residual is calculated, and according to the regression model residual, a degradation degree of the target monitoring point is calculated;

[0161] The over-temperature monitoring result of the target monitoring point is determined according to the matching result between the degradation degree and the preset monitoring condition.

[0162] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0163] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A data-driven over-temperature monitoring method, characterized in that: include: Obtaining the operating data of the target monitoring point in the waste heat boiler within a preset time period and the real-time temperature measurement value of the target monitoring point; Using a pre-trained regression model to perform prediction processing on the operating data to obtain a health status prediction value, wherein the health status prediction value is used to characterize the normal operating temperature of the waste heat boiler under the current operating condition; According to the health state prediction value and the real-time temperature measurement value, a regression model residual is calculated, and according to the regression model residual, a degradation degree of the target monitoring point is calculated; Determining the over-temperature monitoring result of the target monitoring point according to the matching result of the degradation degree and the preset monitoring condition; The step of calculating the degradation degree of the target monitoring point according to the residual of the regression model includes: Retrieve the residual influence weight and temperature sensitivity weight of the target monitoring point; Performing an exponential operation on the regression model residual, and weighting the result of the exponential operation according to the residual influence weight to obtain an exponential residual weighted result; Weighting the real-time temperature measurement value according to the temperature sensitivity weight to obtain a temperature weighted result; Calculating the degradation degree of the target monitoring point according to the exponential residual weighted result and the temperature weighted result; The calculation formula of degradation degree is: , in, represents the degradation degree of the target monitoring point, is the regression model residual, is the real-time temperature measurement value, is the residual influence weight, which is used to measure the impact of the regression model residual on the degradation degree. is the temperature sensitivity weight of the target monitoring point, which is used to reflect the sensitivity of the target monitoring point to the temperature value. 1, It is a fixed value parameter.

2. The method according to claim 1, characterized in that The preset monitoring condition includes a dynamic degradation degree threshold, and the over-temperature monitoring result of the target monitoring point is determined according to the matching result of the degradation degree and the preset monitoring condition, including: Retrieving a dynamic degradation degree threshold of the target monitoring point, and if the degradation degree is greater than the dynamic degradation degree threshold, determining that the over-temperature monitoring result of the target monitoring point is abnormal, and generating over-temperature warning information to issue an over-temperature warning to the target monitoring point; If the degradation degree is less than or equal to the dynamic degradation degree threshold, the over-temperature monitoring result of the target monitoring point is determined to be normal, and the step of obtaining the operating data of the target monitoring point in the waste heat boiler within a preset time period and the real-time temperature measurement value of the target monitoring point is returned.

3. The method according to claim 1, characterized in that Before using the pre-trained regression model to perform prediction processing on the operation data, the method further includes: Constructing an initial regression model and constructing a training sample set based on the historical operation data of the target monitoring point; Using the training sample set to train the initial regression model to obtain a pre-trained regression model; After the operation data is predicted using the pre-trained regression model, the method further includes: Update the historical operation data at preset time intervals, and reconstruct the training sample set based on the updated historical operation data; The pre-trained regression model is retrained using the reconstructed training sample set, so as to perform subsequent prediction processing tasks using the retrained regression model.

4. The method according to claim 3, characterized in that The step of constructing a training sample set based on the historical operation data of the target monitoring point includes: Preprocessing the historical operation data to obtain preprocessed historical operation data, wherein the preprocessing includes data completion processing, duplicate value elimination processing and filtering and denoising processing; Taking the high-temperature superheater inlet flue gas temperature as the target variable, performing correlation feature screening on the preprocessed operation data to obtain correlation feature data; Performing feature generation and feature dimension reduction processing on the associated feature data in sequence to obtain temperature feature data; The operating condition labels corresponding to the temperature characteristic data are marked to obtain a training sample set.

5. The method according to claim 2, characterized in that: The step of retrieving the dynamic degradation degree threshold of the target monitoring point includes: Acquiring environmental data around the target monitoring point; From the degradation degree influence coefficient mapping relationship set, identifying the degradation degree influence coefficient matching the operating data and the environmental data, wherein the degradation degree influence coefficient is used to characterize the influence trend of different environmental data on the degradation degree under different operating conditions; The dynamic degradation degree threshold is corrected according to the degradation degree influence coefficient, and the dynamic degradation degree threshold is updated according to the corrected dynamic degradation degree threshold.

6. The method according to claim 5, characterized in that Before identifying the degradation degree influence coefficient matching the operation data and the environmental data from the degradation degree influence coefficient mapping relationship set, the method further includes: Acquiring historical degradation degree data, and historical operation data and historical environment data corresponding to the historical degradation degree; Fitting the historical degradation degree data, the historical operation data and the historical environmental data to obtain the influence relationship between the environmental data and the operation data and the degradation degree; A degradation degree influence coefficient mapping relationship set is constructed according to the influence relationship, and the degradation degree influence coefficient mapping relationship set includes degradation degree influence coefficients corresponding to different ranges of operating data and different ranges of environmental data.

7. A data-driven over-temperature monitoring device, characterized in that: include: An acquisition module is used to acquire the operating data of a target monitoring point in the waste heat boiler within a preset time period, and a real-time temperature measurement value of the target monitoring point; A prediction module, used to perform prediction processing on the operating data using a pre-trained regression model to obtain a health status prediction value, wherein the health status prediction value is used to characterize the normal operating temperature of the waste heat boiler under the current operating condition; A calculation module, used to calculate the regression model residual according to the health state prediction value and the real-time temperature measurement value, and calculate the degradation degree of the target monitoring point according to the regression model residual; A determination module, used to determine the over-temperature monitoring result of the target monitoring point according to the matching result of the degradation degree and the preset monitoring condition; The computing module comprises: A calling unit, used to call the residual influence weight and temperature sensitivity weight of the target monitoring point; A first calculation unit is used to perform an exponential operation on the regression model residual, and weight the result of the exponential operation according to the residual influence weight to obtain an exponential residual weighted result; A second calculation unit, configured to weight the real-time temperature measurement value according to the temperature sensitivity weight to obtain a temperature weighted result; A third calculation unit, configured to calculate the degradation degree of the target monitoring point according to the exponential residual weighted result and the temperature weighted result; The calculation formula of degradation degree is: ; in, represents the degradation degree of the target monitoring point, is the regression model residual, is the real-time temperature measurement value, is the residual influence weight, which is used to measure the impact of the regression model residual on the degradation degree; is the temperature sensitivity weight of the target monitoring point, which is used to reflect the sensitivity of the target monitoring point to the temperature value. 1, It is a fixed value parameter.

8. A storage medium storing at least one executable instruction, wherein the executable instruction enables a processor to execute operations corresponding to the data-driven over-temperature monitoring method according to any one of claims 1 to 6.

9. A terminal, comprising: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the data-driven over-temperature monitoring method as described in any one of claims 1-6.

Images