Early warning diagnosis method for high-pressure main steam system of gas turbine
Through the method based on the normal model array, the data of high-pressure main steam system is analyzed in real time, and the problem of insufficient timeliness of gas turbine fault warning and diagnosis is solved, real-time fault warning and diagnosis is achieved, and system stability and operating efficiency are improved.
Patent Information
- Application Number
- CN202411546024.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-01
- Publication Date
- 2024-11-29
AI Technical Summary
The fault warning and diagnosis technology of the existing gas turbine high-pressure main steam system is poor in time and lacks real-time warning capabilities, resulting in delayed fault handling and affecting system stability and operating efficiency.
Using a method based on a normal model array, the data of high-pressure main steam system is analyzed in real time by determining measurement points, data acquisition, preprocessing, relevant factor screening and model training, and fault diagnosis and early warning are automatically performed to provide real-time abnormal judgment results.
Real-time automatic analysis and fault warning of high-pressure main steam system are realized, early warning of faults, reduce fault hazards, and improve unit stability and operating efficiency.
Smart Images

Figure CN119043431A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of power generation information technology, and more specifically, to an early warning diagnosis method for a high-pressure main steam system of a gas turbine. Background Art
[0002] As the core component of gas turbine operation, the high-pressure main steam system of a gas turbine is responsible for converting high-temperature and high-pressure steam into mechanical energy, which in turn drives the generator to generate electricity. The continuous and effective operation of the high-pressure main steam system plays a decisive role in ensuring the safe and stable operation of the gas turbine. At present, the high-pressure main steam system of a gas turbine is mostly analyzed after a fault, which has poor timeliness. There are few fault warning and diagnosis technologies, and there is a lack of effective real-time warning technologies. This method aims to provide a set of online real-time automatic analysis of the operation data of the high-pressure main steam system, automatic warning and diagnosis of high-pressure main steam system faults, provide real-time and accurate warning and diagnosis results for operating personnel, and warn of operation faults in advance, providing valuable lead time for fault handling, reducing the harm of faults, and improving the stability of the unit, thereby improving the operation efficiency of the unit. Summary of the invention
[0003] The purpose of the present invention is to address the deficiencies of the prior art and to propose an early warning diagnosis method for a high-pressure main steam system of a gas turbine.
[0004] In a first aspect, a method for early warning diagnosis of a high-pressure main steam system of a gas turbine is provided, comprising: Step 1: Determine the measuring point of the high-pressure main steam system; Step 2: Collect data from the measuring points of the high-pressure main steam system through sensors to obtain the original data set and system feature set; Step 3: preprocessing the original data set; Step 4: Screen the relevant factor features according to the preprocessed data set and the system feature set to obtain a screened relevant factor data set; Step 5: Obtain a correlation factor normal model array and a model threshold according to the correlation factor data set; Step 6: Based on the correlation factor normal model array and the model threshold, perform abnormal location and fault diagnosis on the real-time production data of the high-pressure main steam system, and output the abnormality judgment result and the fault diagnosis result.
[0005] Preferably, step 2 further includes: determining a data acquisition period of the measuring point in a non-fault state of the high-pressure main steam system. and acquisition frequency .
[0006] Preferably, step 3 comprises: Step 3.1, performing outlier processing on the original data set; Step 3.2: Smooth the data set after outlier processing.
[0007] Preferably, step 4 comprises: Step 4.1, perform correlation calculation based on the preprocessed data set to obtain a correlation matrix; Step 4.2: According to the correlation matrix, find the corresponding correlation factor for each feature of the system feature set to obtain a correlation factor data set.
[0008] Preferably, step 5 comprises: Step 5.1, normalizing the correlation factor data set; Step 5.2: Perform model training based on the normalized correlation factor data set and obtain a correlation factor normal model array; Step 5.3: According to the correlation factor normal model array and the correlation factor data set, a prediction value set with prediction features is obtained, and the model threshold is extracted using the actual value set to obtain the model threshold.
[0009] Preferably, step 6 comprises: Step 6.1, according to the correlation factor data set, obtain the real-time data of the corresponding correlation factor and perform normalization processing to obtain the real-time data set and the real-time value of the prediction feature; Step 6.2, obtaining the predicted value of the corresponding feature according to the real-time data set and the normal model array of the correlation factor; Step 6.3, calculating the error between the real-time value of the prediction feature and the prediction value, and comparing the error with the model threshold to determine whether a fault exists, and obtaining a fault identification code set of the system feature according to the fault determination result; Step 6.4: Obtain a fault diagnosis tree, and obtain a high-pressure main steam system fault diagnosis result based on the fault identification code set of the system characteristics and the fault diagnosis tree.
[0010] In a second aspect, an early warning diagnosis system for a high-pressure main steam system of a gas turbine is provided, comprising: A determination module is used to determine the measuring points of the high-pressure main steam system; The acquisition module is used to collect data from the measuring points of the high-pressure main steam system through sensors to obtain the original data set and system feature set; A preprocessing module, used for preprocessing the original data set; A screening module is used to screen the relevant factor characteristics according to the preprocessed data set and the system feature set, and obtain a screened relevant factor data set; An acquisition module, used for acquiring a correlation factor normal model array and a model threshold according to the correlation factor data set; The diagnosis module is used to locate abnormalities and diagnose faults in the real-time production data of the high-pressure main steam system according to the normal model array of the relevant factors and the model threshold, and output abnormality judgment results and fault diagnosis results.
[0011] According to a third aspect, a computer storage medium is provided, wherein a computer program is stored in the computer storage medium; when the computer program is executed on a computer, the computer executes any method described in the first aspect.
[0012] In a fourth aspect, an electronic device is provided, including: Memory, used to store computer programs; A processor is used to execute the computer program to implement any method as described in the first aspect.
[0013] The beneficial effects of the present invention are as follows: the present invention adopts a modeling method based on a normal model array, which can automatically analyze the operating data of the high-pressure main steam system online in real time, automatically warn and diagnose high-pressure main steam system faults, provide real-time and accurate early warning diagnosis results for operating personnel, and warn of operating faults in advance, providing valuable lead time for fault handling, reducing the hazards of faults, and improving the stability of the unit, thereby improving the operating efficiency of the unit. BRIEF DESCRIPTION OF THE DRAWINGS
[0014] Figure 1 A flow chart of a method for early warning diagnosis of a high-pressure main steam system of a gas turbine provided by an embodiment of the present invention; Figure 2 A warning judgment effect diagram provided by an embodiment of the present invention; Figure 3 Another early warning judgment effect diagram provided by an embodiment of the present invention; Figure 4 A line chart of actual data provided by an embodiment of the present invention; Figure 5 A high-pressure main steam system fault diagnosis tree and a diagnosis result diagram are provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0015] The present invention is further described below in conjunction with embodiments. The description of the following embodiments is only used to help understand the present invention. It should be noted that for ordinary persons in the art, without departing from the principle of the present invention, the present invention can also be modified in some ways, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.
[0016] Embodiment 1: like Figure 1As shown, Example 1 of the present application provides an early warning diagnosis method for a high-pressure main steam system of a gas turbine, comprising: Step 1: Determine the measuring points of the high-pressure main steam system.
[0017] Specifically, the embodiment of the present application first needs to determine the relevant measuring points of the required high-pressure main steam system, such as multiple measuring points including high-pressure main steam temperature, high-pressure drum pressure, high-pressure superheater outlet pressure and other characteristics, and then clearly needs to collect relevant data of the high-pressure main steam system. Step 2: Collect data from the measuring points of the high-pressure main steam system through sensors to obtain the original data set and system feature set.
[0018] Step 2 also includes: determining the data acquisition period of the measuring point in the non-fault state of the high-pressure main steam system and acquisition frequency . Data acquisition period at the measuring point According to the preset collection frequency Collect data from relevant measuring points. For example, The original data set of the measurement points And the predictions that need to be made System characteristic set of measurement points For example, determine the collection duration For 1 day and frequency is 10 seconds, forming a The original data set of 12 measurement points And the predictions that need to be made is the system feature set of 5 measurement points .
[0019] Among them, the system feature set that needs to be predicted Represents a data set of characteristic quantities that can reflect the current system failure, the original data set Represents a data set that can affect the characteristics of the system or other systems that need to be predicted.
[0020] Step 3: Perform preprocessing.
[0021] Step 3 includes: Step 3.1: Use the corresponding outlier processing method to process the original data set Perform outlier processing to obtain the data set after outlier processing .
[0022] Specifically, the outlier processing method is to fill in null values and use moving smoothing filling.
[0023] Step 3.2: Dataset after outlier processing Perform smoothing to obtain a data set after data smoothing .
[0024] Specifically, the data smoothing method is an SG filtering method.
[0025] Step 4: Based on the preprocessed data set and the system feature set , perform feature screening of relevant factors and obtain the screened relevant factor data set .
[0026] Step 4 includes: Step 4.1: Based on the preprocessed data set , perform correlation calculation and obtain the correlation matrix.
[0027] Specifically, the Pearson correlation coefficient is used to calculate the correlation coefficient, and the correlation matrix is expressed as:
[0028] in, is the correlation matrix, Indicates the first The same characteristics The correlation coefficient between the features of the original data set.
[0029] Step 4.2: According to the correlation matrix, the system feature set Find the corresponding correlation factor for each feature and obtain the correlation factor data set .
[0030] Specifically, the correlation matrix obtained in step 4.1 is used for the system feature set to be predicted determined in step 2. Filter the relevant features of the system feature set one by one Each feature of the Correlation factors. It is an adaptive value and can be set freely according to the actual situation. For example, In addition, during the screening process, if the absolute value of the correlation coefficient is greater than It is considered relevant when , otherwise it is irrelevant. It is an adaptive value and can be freely set according to the actual situation, such as setting is 0.75.
[0031] Step 5: According to the correlation factor data set, obtain the correlation factor normal model array and model threshold.
[0032] Step 6: Based on the correlation factor normal model array and the model threshold, perform abnormal location and fault diagnosis on the real-time production data of the high-pressure main steam system, and output the abnormality judgment result and the fault diagnosis result.
[0033] Embodiment 2: Based on Example 1, Example 2 of the present application provides a more specific early warning diagnosis method for a high-pressure main steam system of a gas turbine, including: Step 1: Determine the measuring points of the high-pressure main steam system.
[0034] Step 2: Collect data from the measuring points of the high-pressure main steam system through sensors to obtain the original data set and system feature set.
[0035] Step 3: preprocess the original data set.
[0036] Step 4: Screen the relevant factor features based on the preprocessed data set and the system feature set to obtain the screened relevant factor data set. .
[0037] Step 5: Based on the correlation factor data set , get the correlation factor normal model array and model threshold.
[0038] Step 5 includes: Step 5.1: For the correlation factor dataset Perform normalization processing to obtain the processed correlation factor data set .
[0039] Step 5.2: Based on the normalized correlation factor data set , perform model training, and obtain the normal model array of related factors.
[0040] In step 5.2, the model building algorithm uses a bidirectional long short-term memory neural network Bi-LSTM.
[0041] Specifically, we get a model array with the model number , , ..., .
[0042] Step 5.3: Based on the correlation factor normal model array and the correlation factor data set , obtain a set of predicted values with predictive features, and use the actual value set, i.e., the system feature set, to extract the model threshold and obtain the model threshold.
[0043] Specifically, the model array obtained in step 5.2 , , ..., , bring in the correlation factor data set , get the predicted value set for each prediction feature , , ..., , and then use the actual value set , , ..., Perform model threshold extraction to obtain the corresponding model threshold , , ..., Among them, the threshold construction method is Law determination method.
[0044] Step 6: Based on the correlation factor normal model array and the model threshold, perform abnormal location and fault diagnosis on the real-time production data of the high-pressure main steam system, and output the abnormality judgment result and the fault diagnosis result.
[0045] Specifically, step 6 includes: Step 6.1: Based on the correlation factor data set and the system feature set F that needs to be predicted, obtain the real-time data of the corresponding related factors and perform normalization processing to obtain the real-time data set and predicting feature real-time values .
[0046] Specifically, first according to the relevant factor data set Get real-time dataset , and then obtain the real-time value of the predicted feature according to the system feature set F that needs to be predicted. .
[0047] Step 6.2: Based on the real-time data set and the normal model array of the related factors to obtain the predicted values of the corresponding features .
[0048] Specifically, the real-time data of the correlation factor obtained in step 6.1 is brought into the model array obtained in step 5 , , ..., , get the predicted value of the corresponding feature .
[0049] Step 6.3: Calculate the real-time value of the predicted feature and the predicted value and compare the error with the model threshold , , ..., A comparison is performed to determine whether a fault exists, and a set of fault identification codes of system characteristics is obtained based on the fault determination result.
[0050] Specifically, for features greater than the threshold, the returned identification code 1 indicates a fault, otherwise it returns 0 to indicate no fault, and the fault identification code set of the system features is obtained. .
[0051] The high-pressure main steam temperature warning judgment effect diagram is as follows Figure 2 As shown in the figure, the early warning judgment effect diagram of the high-pressure drum pressure is as follows Figure 3 As shown in the figure, the model prediction value and the actual value of the feature are accurately fitted, and the model threshold band falls on the upper and lower sides of the actual value of the feature, giving the precise boundary of the model warning; Step 6.4: Figure 5 As shown, a fault diagnosis tree is obtained, wherein the green and white areas indicate no corresponding faults, and the orange area is the fault diagnosis result. Figure 4 The real-time data graph shown, combined with Figure 5 The fault diagnosis tree is obtained as follows Figure 5 The high-pressure main steam system fault diagnosis result of "the furnace side high-pressure steam injection electric valve is not opened" is shown.
[0052] It should be noted that the parts in this embodiment that are the same or similar to those in Embodiment 1 can be referenced to each other and will not be described in detail in this application.
[0053] Embodiment 3: Based on Example 1, Example 3 of the present application provides an early warning diagnosis system for a high-pressure main steam system of a gas turbine, comprising: A determination module is used to determine the measuring points of the high-pressure main steam system; The acquisition module is used to collect data from the measuring points of the high-pressure main steam system through sensors to obtain the original data set and system feature set; A preprocessing module, used for preprocessing the original data set; A screening module is used to screen the relevant factor characteristics according to the preprocessed data set and the system feature set, and obtain a screened relevant factor data set; An acquisition module, used for acquiring a correlation factor normal model array and a model threshold according to the correlation factor data set; The diagnosis module is used to locate abnormalities and diagnose faults in the real-time production data of the high-pressure main steam system according to the normal model array of the relevant factors and the model threshold, and output abnormality judgment results and fault diagnosis results.
[0054] Specifically, the system provided in this embodiment is a system corresponding to the method provided in Embodiments 1 and 2. Therefore, the parts in this embodiment that are the same or similar to Embodiments 1 and 2 can be referenced to each other and will not be repeated in this application.
Claims
1. A method for early warning diagnosis of a high-pressure main steam system of a gas turbine, characterized in that: include: Step 1: Determine the measuring point of the high-pressure main steam system; Step 2: Collect data from the measuring points of the high-pressure main steam system through sensors to obtain the original data set and system feature set; Step 3: preprocessing the original data set; Step 4: Screen the relevant factor features according to the preprocessed data set and the system feature set to obtain a screened relevant factor data set; Step 5: Obtain a correlation factor normal model array and a model threshold according to the correlation factor data set; Step 6: Based on the correlation factor normal model array and the model threshold, perform abnormal location and fault diagnosis on the real-time production data of the high-pressure main steam system, and output the abnormality judgment result and the fault diagnosis result.
2. The early warning diagnosis method for a gas turbine high-pressure main steam system according to claim 1, characterized in that: Step 2 also includes: determining the data acquisition period of the measuring point in the non-fault state of the high-pressure main steam system and acquisition frequency .
3. The early warning diagnosis method for a gas turbine high-pressure main steam system according to claim 2, characterized in that: Step 3 includes: Step 3.1, performing outlier processing on the original data set; Step 3.2: Smooth the data set after outlier processing.
4. The early warning diagnosis method for a gas turbine high-pressure main steam system according to claim 3, characterized in that: Step 4 includes: Step 4.1, perform correlation calculation based on the preprocessed data set to obtain a correlation matrix; Step 4.2: According to the correlation matrix, find the corresponding correlation factor for each feature of the system feature set to obtain a correlation factor data set.
5. The early warning diagnosis method for a gas turbine high-pressure main steam system according to claim 4, characterized in that: Step 5 includes: Step 5.1, normalizing the correlation factor data set; Step 5.2: Perform model training based on the normalized correlation factor data set and obtain a correlation factor normal model array; Step 5.3: According to the correlation factor normal model array and the correlation factor data set, a prediction value set with prediction features is obtained, and the model threshold is extracted using the actual value set to obtain the model threshold.
6. The early warning diagnosis method for a gas turbine high-pressure main steam system according to claim 5, characterized in that: Step 6 includes: Step 6.1, according to the correlation factor data set and the system feature set to be predicted, obtain the real-time data of the corresponding correlation factor and perform normalization processing to obtain the real-time data set and the real-time value of the predicted feature; Step 6.2, obtaining the predicted value of the corresponding feature according to the real-time data set and the normal model array of the correlation factor; Step 6.3, calculating the error between the real-time value of the prediction feature and the prediction value, and comparing the error with the model threshold to determine whether a fault exists, and obtaining a fault identification code set of the system feature according to the fault determination result; Step 6.4: Obtain a fault diagnosis tree, and obtain a high-pressure main steam system fault diagnosis result based on the fault identification code set of the system characteristics and the fault diagnosis tree.
7. An early warning diagnosis system for a high-pressure main steam system of a gas turbine, used to execute the method according to any one of claims 1 to 6, characterized in that: include: A determination module is used to determine the measuring points of the high-pressure main steam system; The acquisition module is used to collect data from the measuring points of the high-pressure main steam system through sensors to obtain the original data set and system feature set; A preprocessing module, used for preprocessing the original data set; A screening module is used to screen the relevant factor characteristics according to the preprocessed data set and the system feature set, and obtain a screened relevant factor data set; An acquisition module, used for acquiring a correlation factor normal model array and a model threshold according to the correlation factor data set; The diagnosis module is used to locate abnormalities and diagnose faults in the real-time production data of the high-pressure main steam system according to the normal model array of the relevant factors and the model threshold, and output abnormality judgment results and fault diagnosis results.
8. A computer storage medium, characterized in that: The computer storage medium stores a computer program; when the computer program is executed on a computer, the computer executes any one of the methods described in claims 1 to 6.
9. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the method according to any one of claims 1 to 6.
Citation Information
Patent Citations
Monitoring control method and device for main steam system of high-temperature gas cooled reactor
CN113345613A
Fault diagnosis method and system of turboset, computer storage medium and equipment
CN117574322A
Intelligent operation and maintenance method and system for turboset based on digital twinning
CN117974108A
Cited By
Intelligent fault prediction method and system for air separation equipment
CN121955084A
Intelligent fault prediction method and system for air separation equipment
CN121955084B
