Fault diagnosis method for pneumatic soot blowing system

Abstract

The invention relates to the technical field of power plant equipment operation and maintenance, and provides a fault diagnosis method for a pneumatic soot blowing system, which performs real-time fault diagnosis by acquiring system operation data, establishing a fault prediction model, matching a fault type knowledge base, processing prediction failure data and training a fault classification model. According to the method, faults can be automatically predicted, problems can be rapidly positioned, the diagnosis process is optimized, model robustness is enhanced, knowledge base updating is achieved, system stability and operation efficiency are improved, and the problems that a pneumatic soot blowing system is multiple in fault type and difficult to accurately recognize are effectively solved.

Description

Technical Field

[0001] The invention relates to the technical field of power plant equipment operation and maintenance, and in particular to a fault diagnosis method for a pneumatic soot blowing system. Background Art

[0002] As an important industrial dust removal equipment, pneumatic soot blowing system is widely used in the cleaning operations of various industrial boilers, dust collectors and other equipment. The system uses compressed air or other gases as a power source to blow dust out of the equipment to ensure the normal operation of the equipment and extend its service life. However, in the actual operation process, the pneumatic soot blowing system will encounter various faults, affecting the normal operation of the system and the dust removal effect. The pneumatic soot blowing system is usually composed of an air source system, a control system, an actuator (such as a feed valve, a discharge valve, an air intake valve, etc.) and a pipeline system. In terms of working principle, the system realizes the flow of gas and the blowing of dust by controlling the opening and closing of each valve.

[0003] The pneumatic soot blowing system has the following technical pain points during actual use, such as the feed valve and the discharge valve cannot be opened or closed tightly, resulting in gas leakage or dust cannot be effectively discharged. The lack of pressure rise in the silo pump or the system underpressure alarm is caused by insufficient gas source, valve blockage or control system failure. Blockage in the ash conveying pipeline affects the dust conveying efficiency. For example, false alarms in the alarm system and the inability to put the automatic mode into operation affect the automatic operation of the system. Since there are many types of faults in the pneumatic soot blowing system, it is necessary to identify the types of faults in the pneumatic soot blowing system to improve the stability and operation efficiency of the pneumatic soot blowing system. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides a fault diagnosis method for a pneumatic soot blowing system, which solves the problem that there are many types of faults in the pneumatic soot blowing system and the faults of the pneumatic soot blowing system cannot be accurately identified.

[0005] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows:

[0006] The present invention provides a fault diagnosis method for a pneumatic soot blowing system, comprising:

[0007] Step S101, obtaining the operating data of the pneumatic sootblowing system, which includes system alarm information, pressure data, temperature data, valve status, equipment historical trend, electrical control data and medium delivery data. According to the data type of the operating data of the pneumatic sootblowing system, a data detection group is established, and the operating data of the pneumatic sootblowing system is transmitted to the corresponding data detection group, and the data detection group outputs the detection qualified data information;

[0008] Step S102, substituting the qualified detection data information into a preset fault prediction model of the pneumatic soot blowing system to obtain a fault prediction result of the pneumatic soot blowing system, wherein the fault prediction result of the pneumatic soot blowing system includes no fault of the pneumatic soot blowing system, fault of the pneumatic soot blowing system and prediction failure of the pneumatic soot blowing system;

[0009] Step S103, if the pneumatic soot blowing system fault prediction result is that the pneumatic soot blowing system has a fault, the pneumatic soot blowing system fault prediction result is matched in a preset fault type knowledge base to obtain a pneumatic soot blowing system fault type matching result, the pneumatic soot blowing system fault type matching result includes a fault type matching success, a fault type partial matching success, and a fault type matching failure;

[0010] Step S104, if the pneumatic soot blowing system fault prediction result is that the pneumatic soot blowing system prediction fails, retrieve the data with data errors in the data processing process of the pneumatic soot blowing system fault prediction model, retrieve the corresponding pneumatic soot blowing system operation data according to the data attribute of the data error, obtain the data error source data set, detect the data in the data error source data set, obtain error data information, simulate the error data through a preset data simulation model according to the data generation time and data type of the error data information, generate error data simulation data, and substitute the error data simulation data into the pneumatic soot blowing system operation data as replacement data;

[0011] Step S105, use the pneumatic soot blowing system operation data and the data generated during the operation of the fault prediction model of the pneumatic soot blowing system to train the model, obtain the fault classification model of the pneumatic soot blowing system, collect the real-time data of the pneumatic soot blowing system operation data and the real-time pneumatic soot blowing system fault prediction results, and compare the real-time data of the pneumatic soot blowing system operation data and the real-time pneumatic soot blowing system fault prediction results. If the real-time data of the pneumatic soot blowing system operation data and the real-time pneumatic soot blowing system fault prediction results are consistent, then use the fault classification model of the pneumatic soot blowing system within a preset time range to classify the faults of the pneumatic soot blowing system fault prediction results to obtain the fault diagnosis results of the pneumatic soot blowing system.

[0012] Furthermore, the fault diagnosis method for a pneumatic sootblowing system provided by the present invention, step S102, comprises:

[0013] Performing data preprocessing on the operation data of the pneumatic soot blowing system, and dividing the preprocessed operation data of the pneumatic soot blowing system into a training set, a validation set and a test set;

[0014] Extract features from the operating data of the pneumatic soot blowing system to obtain average features, standard deviation features, and peak features, and use a time series analysis model to capture dynamic change data;

[0015] The deep learning model is trained using the training set data, and the model is verified using the validation set data to obtain a fault prediction model for the pneumatic soot blowing system.

[0016] Furthermore, the fault diagnosis method for the pneumatic soot blowing system provided by the present invention, step S103 comprises:

[0017] Extract features from the fault prediction results of the pneumatic soot blowing system to obtain fault indication signal features, abnormal parameter value features and fault warning features;

[0018] The features extracted from the fault prediction results are compared one by one with the feature descriptions in the fault type knowledge base, and the fuzzy matching algorithm is used to evaluate the matching degree between the prediction results and the fault types in the knowledge base;

[0019] If the matching degree is higher than the preset threshold, it is determined that the fault type matching is successful; if the matching degree is lower than the preset threshold, it is determined that the fault type matching is failed or partially successful.

[0020] Furthermore, the fault diagnosis method for a pneumatic soot blowing system provided by the present invention, step S104, comprises:

[0021] Filter out the fault prediction data and data records known to have errors from the historical operation data of the pneumatic soot blowing system;

[0022] The model is trained using fault prediction data and data records known to have errors to obtain a data simulation model;

[0023] The constructed data simulation model is integrated into the fault prediction process of the pneumatic soot blowing system to generate error data simulation data.

[0024] Furthermore, the fault diagnosis method for a pneumatic soot blowing system provided by the present invention, step S105, comprises:

[0025] The real-time operation data collected by the pneumatic soot blowing system and the real-time prediction results output from the fault prediction model and the data generation time are used as matching criteria for data synchronization matching, so as to obtain the real-time operation data collected by the pneumatic soot blowing system and the real-time prediction results output from the fault prediction model corresponding to the same time period or the same operation cycle;

[0026] The real-time data of the pneumatic soot blowing system operation data is compared with each monitoring parameter value of the real-time pneumatic soot blowing system fault prediction result item by item. The error generated based on the comparison is compared with the preset error threshold to determine whether the actual value of the real-time operation data is consistent with the predicted value. If the errors of all key parameters are within the threshold range, the actual value of the real-time operation data is consistent with the predicted value. Otherwise, the actual value of the real-time operation data is inconsistent with the predicted value.

[0027] Furthermore, the fault diagnosis method for a pneumatic soot blowing system provided by the present invention, step S105, comprises:

[0028] Receiving task data of a pneumatic soot blowing system, substituting the task data of the pneumatic soot blowing system into a data simulation model, and generating task simulation data of the pneumatic soot blowing system;

[0029] The task simulation data of the pneumatic soot blowing system is classified through the fault classification model of the pneumatic soot blowing system to obtain the expected fault type of the pneumatic soot blowing system;

[0030] Match the predicted fault type of the pneumatic soot blowing system with a fault solution in a preset fault type knowledge base. If the match is successful, convert the successfully matched fault solution into a fault maintenance control instruction and a solution report information;

[0031] Transmitting fault maintenance control instructions to corresponding equipment in the pneumatic soot blowing system, and obtaining feedback information from corresponding equipment in the pneumatic soot blowing system;

[0032] The feedback information of the corresponding equipment in the pneumatic soot blowing system is substituted into the fault prediction model of the pneumatic soot blowing system. If there is a fault in the obtained prediction result, the preset fault type knowledge base is updated and optimized.

[0033] Furthermore, the fault diagnosis method for a pneumatic soot blowing system provided by the present invention, step S105, comprises:

[0034] Collect data generated by the pneumatic soot blowing system during actual operation, including system alarm information, pressure data, temperature data, and valve status;

[0035] Clean and preprocess the newly collected data, integrate the preprocessed new data into the existing data set, use the deep learning model to perform fault analysis on the integrated data set, and identify new fault modes;

[0036] The identified new fault mode or feature is verified, and the verified new fault mode is added to the fault type knowledge base as a new fault type.

[0037] The beneficial effects of the present invention are mainly reflected in the following aspects: through the preset pneumatic soot blowing system fault prediction model, the present invention can automatically and accurately predict the fault conditions of the pneumatic soot blowing system, significantly improving the accuracy of fault identification. This helps to timely discover and handle faults, reducing downtime and maintenance costs caused by faults.

[0038] The present invention realizes the rapid location and analysis of pneumatic soot blowing system faults through detailed data acquisition, processing, prediction and diagnosis processes. Compared with the traditional manual diagnosis method, this process is more efficient and standardized, reduces the interference of human factors, and improves the reliability and consistency of fault diagnosis. In response to the situation of prediction failure, the present invention proposes an error data processing mechanism, which generates error data simulation data to replace the original error data, thereby eliminating the impact of error data on the prediction model. This mechanism enhances the robustness of the model, so that the model can still output relatively accurate prediction results when facing abnormal data.

[0039] The present invention can continuously discover new fault modes and characteristics by collecting and analyzing operating data in real time. These new discoveries will be added to the fault type knowledge base after verification, thereby realizing continuous updating and optimization of the knowledge base. This not only improves the system's ability to identify new faults, but also makes the entire fault diagnosis method more flexible and adaptable. Through timely and accurate fault prediction and diagnosis, the present invention can effectively reduce the downtime and maintenance costs of the pneumatic sootblowing system, thereby improving the stability and operating efficiency of the system. This is of great significance for ensuring the normal operation of industrial equipment and extending its service life.

[0040] In summary, the present invention provides an efficient and accurate fault diagnosis method for pneumatic sootblowing systems. By combining data-driven and model optimization, the efficiency and accuracy of fault identification and diagnosis are significantly improved, providing strong support for the operation and maintenance management of industrial equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0042] Figure 1 A schematic diagram of a production system for a fault diagnosis method for a pneumatic soot blowing system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. The technical solutions provided by the embodiments of the present invention are described in detail below in conjunction with the drawings.

[0044] In order to better understand the purpose of the present invention, the present invention is described in further detail below.

[0045] The present invention provides a fault diagnosis method for a pneumatic soot blowing system, comprising:

[0046] Step S101, obtaining the operating data of the pneumatic sootblowing system, which includes system alarm information, pressure data, temperature data, valve status, equipment historical trend, electrical control data and medium delivery data. According to the data type of the operating data of the pneumatic sootblowing system, a data detection group is established, and the operating data of the pneumatic sootblowing system is transmitted to the corresponding data detection group, and the data detection group outputs the detection qualified data information;

[0047] In step S101, the present invention describes in detail how to obtain and process the operating data of the pneumatic soot blowing system to improve the accuracy and completeness of the data. The specific steps are as follows:

[0048] Collect operating data from the pneumatic sootblowing system. These data are rich in variety, including system alarm information, pressure data, temperature data, valve status, equipment historical trends, electrical control data, and media delivery data. These data fully reflect the operating status and performance of the pneumatic sootblowing system.

[0049] The acquired operation data is classified according to the type and characteristics of the data. The purpose of this is to facilitate subsequent data processing and analysis. Specifically, different data detection groups are established according to the type of data (such as alarm information, pressure data, etc.).

[0050] The classified data is transmitted to the corresponding data detection groups. Each group is responsible for processing a specific type of data to improve the accuracy and effectiveness of the data.

[0051] In each data detection group, the transmitted data is detected. The purpose of this step is to identify and filter out invalid or erroneous data, so that the subsequent analysis uses high-quality data.

[0052] After data detection and grouping, the qualified data information is output. These data will serve as input for subsequent fault prediction and diagnosis, which is crucial to improving the accuracy of prediction and diagnosis.

[0053] Through the above steps, the operation data obtained from the pneumatic soot blowing system in step S101 is used in the subsequent fault prediction and diagnosis process in a high-quality form after strict classification, transmission and detection.

[0054] Step S102, substituting the qualified detection data information into a preset fault prediction model of the pneumatic soot blowing system to obtain a fault prediction result of the pneumatic soot blowing system, wherein the fault prediction result of the pneumatic soot blowing system includes no fault of the pneumatic soot blowing system, fault of the pneumatic soot blowing system and prediction failure of the pneumatic soot blowing system;

[0055] In step S102, a key step of the fault diagnosis method for a pneumatic soot blowing system proposed by the present invention is to substitute the qualified data information into a preset fault prediction model. This step is a bridge connecting data collection with subsequent fault type matching, error data processing and final fault diagnosis. The specific process is as follows:

[0056] The qualified data information outputted in step S101 is used as input and passed to a preset pneumatic soot blowing system fault prediction model. This model is trained based on previously collected system operation data and can identify and analyze the pattern of input data to predict whether the system has a fault.

[0057] After receiving the input data, the fault prediction model performs calculations and analysis based on its internal algorithm and the features learned during training, and then outputs the prediction results. There are three types of prediction results:

[0058] Pneumatic soot blowing system has no faults: It means that the current system is operating normally and no signs of faults are detected.

[0059] There is a fault in the pneumatic soot blowing system: It means that based on the input data, the model predicts that there is a certain fault in the system.

[0060] Pneumatic sootblowing system prediction failure: There are anomalies or errors in the input data that the model cannot handle or understand, or the model itself has insufficient prediction capabilities for certain specific situations.

[0061] Result processing: According to the output results of the fault prediction model, the system will perform subsequent processing. If the prediction result is "no system fault", the current monitoring state will continue to be maintained or regular maintenance inspections will be performed. If the prediction result is "there is a system fault", it will be further matched in the fault type knowledge base to determine the specific fault type (step S103). If the prediction result is "prediction failure", it is necessary to identify and solve the data problem through the data error handling mechanism (step S104).

[0062] The effectiveness of this step depends on the accuracy and robustness of the fault prediction model, and the performance of the model is directly affected by the quality and quantity of its training data. Therefore, the data collection, classification and detection process (step S101) before step S102 is crucial to the success of the entire fault diagnosis method.

[0063] Step S103, if the pneumatic soot blowing system fault prediction result is that the pneumatic soot blowing system has a fault, the pneumatic soot blowing system fault prediction result is matched in a preset fault type knowledge base to obtain a pneumatic soot blowing system fault type matching result, the pneumatic soot blowing system fault type matching result includes a fault type matching success, a fault type partial matching success, and a fault type matching failure;

[0064] The fault prediction results are matched in the preset fault type knowledge base. This knowledge base contains various known fault types and their corresponding feature descriptions, which are accumulated by the system through historical data and expert experience.

[0065] Extract key features from the fault prediction results, such as fault indication signal features, abnormal parameter value features, and fault warning features. Compare the features with the feature descriptions in the fault type knowledge base one by one, and use the fuzzy matching algorithm to evaluate the similarity or matching degree between the prediction results and each fault type in the knowledge base.

[0066] Based on the matching degree, the system will determine the result of the fault type matching. The results include: Successful fault type matching: If the characteristics of the prediction result are highly matched with a certain fault type in the knowledge base, that is, the matching degree is higher than the preset threshold, it is considered that the specific fault type is successfully matched. Partial successful fault type matching: In some cases, the prediction result has a certain degree of match with multiple fault types, but none of them is completely matched. At this time, the system will report the result of partial successful matching and give multiple fault types. Fault type matching failure: If the prediction result does not match all fault types in the knowledge base, that is, the matching degree is lower than the preset threshold, it is considered that the fault type matching has failed.

[0067] According to the result of fault type matching, the system performs corresponding follow-up processing. If the match is successful, a fault report will be directly generated and the corresponding maintenance process will be triggered. If the match is partially successful or failed, further manual inspection or data collection is required to confirm the fault type.

[0068] Through this step, the system can finely classify and identify fault prediction results based on the preset fault type knowledge base, thereby providing more accurate and specific guidance for subsequent repair and maintenance work.

[0069] Step S104, if the pneumatic soot blowing system fault prediction result is that the pneumatic soot blowing system prediction fails, retrieve the data with data errors in the data processing process of the pneumatic soot blowing system fault prediction model, retrieve the corresponding pneumatic soot blowing system operation data according to the data attribute of the data error, obtain the data error source data set, detect the data in the data error source data set, obtain error data information, simulate the error data through a preset data simulation model according to the data generation time and data type of the error data information, generate error data simulation data, and substitute the error data simulation data into the pneumatic soot blowing system operation data as replacement data;

[0070] In step S104, for the case where the fault prediction result of the pneumatic soot blowing system is "prediction failure", the present invention proposes a series of specific data error processing mechanisms to improve the continuity and accuracy of the fault diagnosis process. The following is a detailed step description:

[0071] When a fault prediction model returns a "failed prediction" result, the system first identifies and retrieves data errors that occurred during the model data processing process. These errors are caused by incomplete data, outliers, sensor failures, etc.

[0072] According to the data attributes of the data error (such as data type, source, timestamp, etc.), the system will further retrieve the corresponding pneumatic soot blowing system operation data to construct a data set containing the source of the error.

[0073] Conduct in-depth inspection of the data in the data set at the source of the data error to extract specific error data information. This information will help identify the specific nature and cause of the error.

[0074] Using the preset data simulation model, simulation is performed based on the data generation time and data type of the erroneous data information. This step is intended to generate simulated data that is similar to the actual erroneous data but does not affect the accuracy of the model prediction.

[0075] The generated error data simulation data is used as replacement data and re-substituted into the operating data of the pneumatic soot blowing system. The purpose of this is to eliminate the impact of the original error data on the fault prediction model, so that the model can make predictions on a data set that is closer to the actual situation.

[0076] The fault prediction model is rerun using the updated data set (containing simulated error data replacement) in the hope of obtaining more accurate and reliable prediction results.

[0077] Through this step, the system can automatically correct and compensate for data errors, thereby avoiding prediction failures caused by data problems. This not only improves the robustness of the fault prediction model, but also improves the continuity and effectiveness of the entire fault diagnosis process. At the same time, by continuously accumulating error handling experience and optimizing the data simulation model, the system can gradually improve its ability to respond to future error situations.

[0078] Step S105, use the pneumatic soot blowing system operation data and the data generated during the operation of the fault prediction model of the pneumatic soot blowing system to train the model, obtain the fault classification model of the pneumatic soot blowing system, collect the real-time data of the pneumatic soot blowing system operation data and the real-time pneumatic soot blowing system fault prediction results, and compare the real-time data of the pneumatic soot blowing system operation data and the real-time pneumatic soot blowing system fault prediction results. If the real-time data of the pneumatic soot blowing system operation data and the real-time pneumatic soot blowing system fault prediction results are consistent, then use the fault classification model of the pneumatic soot blowing system within a preset time range to classify the faults of the pneumatic soot blowing system fault prediction results to obtain the fault diagnosis results of the pneumatic soot blowing system.

[0079] In step S105, the present invention proposes a method for training a fault classification model using the operating data of the pneumatic sootblowing system and the data generated during the operation of the fault prediction model, and further using the model for real-time fault diagnosis. The following is a detailed explanation of this step:

[0080] The operating data of the pneumatic sootblowing system (including system alarm information, pressure data, temperature data, valve status, etc.) and the data generated by the fault prediction model during operation are used as training materials.

[0081] A fault classification model is trained using these data, which can automatically identify and classify the types of faults in the pneumatic sootblowing system based on the input data.

[0082] Real-time collection of operating data of the pneumatic soot blowing system and real-time prediction results output by the fault prediction model. Real-time data reflects the current operating status of the system and existing fault conditions.

[0083] Compare the real-time operation data with the real-time prediction results to verify the accuracy of the prediction results. During the comparison process, the system calculates the difference between the actual value of the operation data and the predicted value, and determines whether the two are consistent based on the preset error threshold.

[0084] If the real-time operating data is consistent with the real-time prediction results (i.e. the errors of all key parameters are within the preset threshold range), the prediction results are considered reliable. In this case, the system will use the fault classification model to classify the predicted faults within the preset time range to determine the specific fault type and cause. The results of the fault classification will serve as the final fault diagnosis results of the pneumatic soot blowing system to guide subsequent repair and maintenance work.

[0085] Throughout the process, the system continuously optimizes the fault classification model and fault prediction model based on real-time feedback and fault diagnosis results. If a new fault mode or feature is found, the system will add it to the fault type knowledge base and retrain the model to improve its ability to identify unknown faults.

[0086] Through this step, the present invention realizes real-time fault diagnosis of the pneumatic soot blowing system and can improve the accuracy and efficiency of fault diagnosis by continuously optimizing the model, which is of great significance for ensuring the stable operation of the pneumatic soot blowing system and extending its service life.

[0087] Specifically, the fault diagnosis method for a pneumatic soot blowing system provided by the present invention, step S102, comprises:

[0088] Performing data preprocessing on the operation data of the pneumatic soot blowing system, and dividing the preprocessed operation data of the pneumatic soot blowing system into a training set, a validation set and a test set;

[0089] Extract features from the operating data of the pneumatic soot blowing system to obtain average features, standard deviation features, and peak features, and use a time series analysis model to capture dynamic change data;

[0090] The deep learning model is trained using the training set data, and the model is verified using the validation set data to obtain a fault prediction model for the pneumatic soot blowing system.

[0091] Specifically, the fault diagnosis method for a pneumatic soot blowing system provided by the present invention, step S103, comprises:

[0092] Extract features from the fault prediction results of the pneumatic soot blowing system to obtain fault indication signal features, abnormal parameter value features and fault warning features;

[0093] The features extracted from the fault prediction results are compared one by one with the feature descriptions in the fault type knowledge base, and the fuzzy matching algorithm is used to evaluate the matching degree between the prediction results and the fault types in the knowledge base;

[0094] If the matching degree is higher than the preset threshold, it is determined that the fault type matching is successful; if the matching degree is lower than the preset threshold, it is determined that the fault type matching is failed or partially successful.

[0095] Specifically, the fault diagnosis method for a pneumatic soot blowing system provided by the present invention, step S104, comprises:

[0096] Filter out the fault prediction data and data records known to have errors from the historical operation data of the pneumatic soot blowing system;

[0097] The model is trained using fault prediction data and data records known to have errors to obtain a data simulation model;

[0098] The constructed data simulation model is integrated into the fault prediction process of the pneumatic soot blowing system to generate error data simulation data.

[0099] Specifically, the fault diagnosis method for a pneumatic soot blowing system provided by the present invention, step S105, comprises:

[0100] The real-time operation data collected by the pneumatic soot blowing system and the real-time prediction results output from the fault prediction model and the data generation time are used as matching criteria for data synchronization matching, so as to obtain the real-time operation data collected by the pneumatic soot blowing system and the real-time prediction results output from the fault prediction model corresponding to the same time period or the same operation cycle;

[0101] The real-time data of the pneumatic soot blowing system operation data is compared with each monitoring parameter value of the real-time pneumatic soot blowing system fault prediction result item by item. The error generated based on the comparison is compared with the preset error threshold to determine whether the actual value of the real-time operation data is consistent with the predicted value. If the errors of all key parameters are within the threshold range, the actual value of the real-time operation data is consistent with the predicted value. Otherwise, the actual value of the real-time operation data is inconsistent with the predicted value.

[0102] Specifically, the fault diagnosis method for a pneumatic soot blowing system provided by the present invention, step S105, comprises:

[0103] Receiving task data of a pneumatic soot blowing system, substituting the task data of the pneumatic soot blowing system into a data simulation model, and generating task simulation data of the pneumatic soot blowing system;

[0104] The task simulation data of the pneumatic soot blowing system is classified through the fault classification model of the pneumatic soot blowing system to obtain the expected fault type of the pneumatic soot blowing system;

[0105] Match the predicted fault type of the pneumatic soot blowing system with a fault solution in a preset fault type knowledge base. If the match is successful, convert the successfully matched fault solution into a fault maintenance control instruction and a solution report information;

[0106] Transmitting fault maintenance control instructions to corresponding equipment in the pneumatic soot blowing system, and obtaining feedback information from corresponding equipment in the pneumatic soot blowing system;

[0107] The feedback information of the corresponding equipment in the pneumatic soot blowing system is substituted into the fault prediction model of the pneumatic soot blowing system. If there is a fault in the obtained prediction result, the preset fault type knowledge base is updated and optimized.

[0108] Specifically, the fault diagnosis method for a pneumatic soot blowing system provided by the present invention, step S105, comprises:

[0109] Collect data generated by the pneumatic soot blowing system during actual operation, including system alarm information, pressure data, temperature data, and valve status;

[0110] Clean and preprocess the newly collected data, integrate the preprocessed new data into the existing data set, use the deep learning model to perform fault analysis on the integrated data set, and identify new fault modes;

[0111] The identified new fault mode or feature is verified, and the verified new fault mode is added to the fault type knowledge base as a new fault type.

[0112] The technical solution of the present invention aims at the problem that there are many types of faults in the pneumatic soot blowing system and it is difficult to accurately identify them, and proposes a fault diagnosis method for the pneumatic soot blowing system. The problem is solved by the following steps:

[0113] Data acquisition and detection (step S101):

[0114] Obtain pneumatic sootblowing system operation data, including system alarm information, pressure data, temperature data, valve status, equipment historical trends, electrical control data, and media delivery data.

[0115] Establish data detection groups according to data types, transfer data to corresponding groups for detection, and improve data quality.

[0116] Fault prediction (step S102):

[0117] The qualified test data is substituted into the preset fault prediction model, which is trained through deep learning and other technologies and can output fault prediction results, including no fault, fault and prediction failure.

[0118] Fault type matching (step S103):

[0119] If the prediction result is that there is a fault, the prediction result is matched in the preset fault type knowledge base to determine the fault type. The matching results include success, partial success and failure.

[0120] Error data processing (step S104):

[0121] If the prediction result is a prediction failure, simulated data of the erroneous data is generated through the data simulation model to replace the erroneous data, thereby improving the data quality of the model input and improving the prediction accuracy.

[0122] Model training and fault diagnosis (step S105):

[0123] Use the operating data and model operating data to train the fault classification model, predict and compare the real-time data, and if they are consistent, classify the fault within the preset time to obtain the final diagnosis result.

[0124] Knowledge Base Updates:

[0125] In the process of processing real-time data and predicting results, if new fault modes or characteristics are found, they will be added to the fault type knowledge base after verification to achieve update and optimization of the knowledge base.

[0126] Through these steps, the technical solution of the present invention can effectively solve the problem that there are many types of faults in the pneumatic soot blowing system and they cannot be accurately identified. It realizes the rapid and accurate prediction and diagnosis of faults through a data-driven approach, combined with a deep learning model and a data simulation model, and improves the stability and operating efficiency of the pneumatic soot blowing system. At the same time, by continuously updating the knowledge base, the system can also ensure its ability to identify new faults.

[0127] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations, and the above-described embodiments of the present invention do not constitute a limitation on the scope of protection of the present invention.

Claims

1. A fault diagnosis method for a pneumatic sootblowing system, characterized in that: include: Step S101, obtaining the operating data of the pneumatic sootblowing system, which includes system alarm information, pressure data, temperature data, valve status, equipment historical trend, electrical control data and medium delivery data. According to the data type of the operating data of the pneumatic sootblowing system, a data detection group is established, and the operating data of the pneumatic sootblowing system is transmitted to the corresponding data detection group, and the data detection group outputs the detection qualified data information; Step S102, substituting the qualified detection data information into a preset fault prediction model of the pneumatic soot blowing system to obtain a fault prediction result of the pneumatic soot blowing system, wherein the fault prediction result of the pneumatic soot blowing system includes no fault of the pneumatic soot blowing system, fault of the pneumatic soot blowing system and prediction failure of the pneumatic soot blowing system; Step S103, if the pneumatic soot blowing system fault prediction result is that the pneumatic soot blowing system has a fault, the pneumatic soot blowing system fault prediction result is matched in a preset fault type knowledge base to obtain a pneumatic soot blowing system fault type matching result, the pneumatic soot blowing system fault type matching result includes a fault type matching success, a fault type partial matching success, and a fault type matching failure; Step S104, if the pneumatic soot blowing system fault prediction result is that the pneumatic soot blowing system prediction fails, retrieve the data with data errors in the data processing process of the pneumatic soot blowing system fault prediction model, retrieve the corresponding pneumatic soot blowing system operation data according to the data attribute of the data error, obtain the data error source data set, detect the data in the data error source data set, obtain error data information, simulate the error data through a preset data simulation model according to the data generation time and data type of the error data information, generate error data simulation data, and substitute the error data simulation data into the pneumatic soot blowing system operation data as replacement data; Step S105, use the pneumatic soot blowing system operation data and the data generated during the operation of the fault prediction model of the pneumatic soot blowing system to train the model, obtain the fault classification model of the pneumatic soot blowing system, collect the real-time data of the pneumatic soot blowing system operation data and the real-time pneumatic soot blowing system fault prediction results, and compare the real-time data of the pneumatic soot blowing system operation data and the real-time pneumatic soot blowing system fault prediction results. If the real-time data of the pneumatic soot blowing system operation data and the real-time pneumatic soot blowing system fault prediction results are consistent, then use the fault classification model of the pneumatic soot blowing system within a preset time range to classify the faults of the pneumatic soot blowing system fault prediction results to obtain the fault diagnosis results of the pneumatic soot blowing system.

2. The fault diagnosis method for a pneumatic sootblowing system according to claim 1, characterized in that: The step S102 includes: Performing data preprocessing on the operation data of the pneumatic soot blowing system, and dividing the preprocessed operation data of the pneumatic soot blowing system into a training set, a validation set and a test set; Extract features from the operating data of the pneumatic soot blowing system to obtain average features, standard deviation features, and peak features, and use a time series analysis model to capture dynamic change data; The deep learning model is trained using the training set data, and the model is verified using the validation set data to obtain a fault prediction model for the pneumatic soot blowing system.

3. The fault diagnosis method for a pneumatic sootblowing system according to claim 1, characterized in that: The step S103 includes: Extract features from the fault prediction results of the pneumatic soot blowing system to obtain fault indication signal features, abnormal parameter value features and fault warning features; The features extracted from the fault prediction results are compared one by one with the feature descriptions in the fault type knowledge base, and the fuzzy matching algorithm is used to evaluate the matching degree between the prediction results and the fault types in the knowledge base; If the matching degree is higher than the preset threshold, it is determined that the fault type matching is successful; if the matching degree is lower than the preset threshold, it is determined that the fault type matching is failed or partially successful.

4. The fault diagnosis method for a pneumatic sootblowing system according to claim 1, characterized in that: The step S104 includes: Filter out the fault prediction data and data records known to have errors from the historical operation data of the pneumatic soot blowing system; The model is trained using fault prediction data and data records known to have errors to obtain a data simulation model; The constructed data simulation model is integrated into the fault prediction process of the pneumatic soot blowing system to generate error data simulation data.

5. The fault diagnosis method for a pneumatic sootblowing system according to claim 1, characterized in that: The step S105 includes: The real-time operation data collected by the pneumatic soot blowing system and the real-time prediction results output from the fault prediction model and the data generation time are used as matching criteria for data synchronization matching, so as to obtain the real-time operation data collected by the pneumatic soot blowing system and the real-time prediction results output from the fault prediction model corresponding to the same time period or the same operation cycle; The real-time data of the pneumatic soot blowing system operation data is compared with each monitoring parameter value of the real-time pneumatic soot blowing system fault prediction result item by item. The error generated based on the comparison is compared with the preset error threshold to determine whether the actual value of the real-time operation data is consistent with the predicted value. If the errors of all key parameters are within the threshold range, the actual value of the real-time operation data is consistent with the predicted value. Otherwise, the actual value of the real-time operation data is inconsistent with the predicted value.

6. The fault diagnosis method for a pneumatic sootblowing system according to claim 5, characterized in that: The step S105 includes: Receiving task data of a pneumatic soot blowing system, substituting the task data of the pneumatic soot blowing system into a data simulation model, and generating task simulation data of the pneumatic soot blowing system; The task simulation data of the pneumatic soot blowing system is classified through the fault classification model of the pneumatic soot blowing system to obtain the expected fault type of the pneumatic soot blowing system; Match the predicted fault type of the pneumatic soot blowing system with a fault solution in a preset fault type knowledge base. If the match is successful, convert the successfully matched fault solution into a fault maintenance control instruction and a solution report information; Transmitting fault maintenance control instructions to corresponding equipment in the pneumatic soot blowing system, and obtaining feedback information from corresponding equipment in the pneumatic soot blowing system; The feedback information of the corresponding equipment in the pneumatic soot blowing system is substituted into the fault prediction model of the pneumatic soot blowing system. If there is a fault in the obtained prediction result, the preset fault type knowledge base is updated and optimized.

7. The fault diagnosis method for a pneumatic sootblowing system according to claim 6, characterized in that: The step S105 includes: Collect data generated by the pneumatic soot blowing system during actual operation, including system alarm information, pressure data, temperature data, and valve status; Clean and preprocess the newly collected data, integrate the preprocessed new data into the existing data set, use the deep learning model to perform fault analysis on the integrated data set, and identify new fault modes; The identified new fault mode or feature is verified, and the verified new fault mode is added to the fault type knowledge base as a new fault type.

Images