Combined cycle unit operation state monitoring sensor fault signal dynamic reconstruction method and system

Abstract

The invention discloses a combined cycle unit operation state monitoring sensor fault signal dynamic reconstruction method and system, and belongs to the technical field of power system digital intelligence. The method comprises the following steps: firstly, matching multi-source operation data of the combined cycle unit based on a data timestamp alignment rule to obtain basic operation data; secondly, according to an improved Pauta criterion method, analyzing a focused measuring point signal, and detecting a fault state of the sensor; then, correlation analysis is carried out on the basic operation data, a signal characteristic variable matrix is constructed, and a signal reconstruction model is input to generate a reconstruction signal to replace an abnormal data signal of the fault sensor; the system comprises a preprocessing module, a fault analysis module and a reconstruction module, and automatic operation of the method is achieved. According to the method, the fault signal problem of the sensor in an extreme environment is effectively solved through a dynamic reconstruction technology, the reliability and accuracy of a monitoring system are improved, and the method is suitable for running state monitoring and diagnosis of the combined cycle unit.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system digitization, in particular to a combined cycle unit operation state monitoring sensor fault signal dynamic reconstruction method. BACKGROUND

[0002] With the deep application of new generation information technologies such as big data, cloud computing, Internet of Things and artificial intelligence, the digitization construction of power systems is undergoing revolutionary changes. Under this background, gas power enterprises need to build a unit operation state monitoring and diagnosis system based on the forefront theory of power system digitization to achieve the strategic goal of "operation supervision, fault diagnosis, energy saving benchmarking and management consumption reduction". As the sensing front end of the monitoring system, sensors face multiple challenges in complex working conditions: in extreme environments such as high temperature, high pressure, high speed and strong vibration, the failure rate of combined cycle unit operation state monitoring sensors is 3-5 times that of conventional industrial environments. Such harsh working conditions lead to the risk of drift and distortion of measurement data, directly affecting the effective operation and accurate decision-making of subsequent diagnosis algorithms.

[0003] Existing sensor fault signal reconstruction methods are mostly based on physical models, which require accurate mathematical models. For complex systems such as gas turbines, model mismatch, nonlinear dynamics and environmental time variability can lead to error accumulation. Or based on redundant design, the reliability of monitoring signals is improved through multiple sensors, but hardware redundancy costs are high and are easily affected by common cause failures. Analyzing redundancy also faces the challenge of multi-source heterogeneous data fusion. Overall, existing sensor fault signal reconstruction methods still have significant shortcomings in terms of extreme environment adaptability, multi-physical field coupling, signal reconstruction accuracy and real-time requirements, which further affect the overall working effect of the combined cycle unit operation state monitoring and diagnosis system. SUMMARY

[0004] In view of the harsh working environment and complex fault signals of sensors in the combined cycle unit operation state monitoring system, to avoid the tedious work of sensor fault signal reconstruction, the present application proposes a combined cycle unit operation state monitoring sensor fault signal dynamic reconstruction method based on multi-feature recognition technology, which effectively reconstructs the sensor fault signal while ensuring the efficient operation and accurate decision-making of the combined cycle unit operation state monitoring and diagnosis system.

[0005] The present application is realized by the following technical solutions: A combined cycle unit operation state monitoring sensor fault signal dynamic reconstruction method, comprising the following steps: Step 1, match the multi-source operation data of the combined cycle unit based on the data timestamp alignment rule to obtain the basic operation data; Step 2, obtaining the key point signals of the basic operation data according to the working logic of the combined cycle unit operation state monitoring and diagnosis system, analyzing the signals according to the improved standard method, and detecting the faults of the sensors corresponding to the signals; Step 3, performing correlation analysis on the basic operation data to obtain the related variables of the fault sensors, constructing a signal feature variable matrix according to the related variables, inputting the signal feature variable matrix into a signal reconstruction model to obtain the reconstructed signals of the fault sensors, and replacing the abnormal data signals of the fault sensors with the reconstructed signals.

[0006] Preferably, step 1 includes matching the multi-source operation data of the combined cycle unit based on a data timestamp alignment rule, which comprises: adopting a timestamp interpolation matching method to match the multi-source operation data of the sensors to obtain the basic operation data.

[0007] Preferably, the timestamp interpolation matching method is as follows:

[0008] In the formula, is the preprocessed basic operation data, is the unit operation state monitoring parameter, is the timestamp corresponding to the data, is the parameter is the monitoring data at the timestamp .

[0009] Preferably, the key point signals of the basic operation data in step 2 include the gas turbine operation parameters, the heat recovery boiler operation parameters, the steam turbine operation parameters, and the cold end operation parameters.

[0010] The gas turbine operation parameters include the gas turbine power, the gas turbine speed, the inlet pressure and temperature of the compressor, the exhaust pressure and temperature of the compressor, the pressure, temperature and flow of the fuel, the exhaust pressure and temperature of the turbine; The heat recovery boiler operation parameters include the high-pressure feed water pressure, temperature and flow, the medium-pressure feed water pressure, temperature and flow, and the condensate water pressure, temperature and flow; The steam turbine operation parameters include the pressure and temperature before the high-pressure main valve, the exhaust pressure and temperature of the high-pressure cylinder, the pressure and temperature before the medium-pressure main valve, the pressure and temperature before the low-pressure supplementary valve, the inlet pressure and temperature of the low-pressure cylinder, and the exhaust pressure of the low-pressure cylinder; The cold end operation parameters include the condenser pressure, the inlet temperature of the condenser circulating water, and the outlet temperature of the condenser circulating water.

[0011] Preferably, step 2 includes analyzing the signals according to the improved The criteria method analyzes the measurement point signals and performs fault detection on the corresponding sensors. Determine the data change of the measurement point signal at adjacent times, determine the mean value of the signal data change at adjacent times within the window period based on the signal data change, determine the root mean square error of the signal data change at adjacent times within the window period based on the mean value of the change, and determine the fault state of the sensor based on the absolute value of the signal data change and the root mean square error within the window period.

[0012] Preferably, step 3, which involves performing correlation analysis on the basic operating data to obtain relevant variables of the fault sensor and constructing a signal feature variable matrix based on these relevant variables, includes: Normalize the basic operational data; Constructed based on normalized basic operational data The correlation coefficient matrix is ​​as follows:

[0013] In the formula, Basic data Correlation coefficient matrix For parameters Signals and parameters signal Correlation coefficient Parameters after normalization Mean value of signal data Parameters after normalization Mean value of signal data; Based on basic data The correlation coefficient matrix is ​​used to construct a signal feature variable matrix by selecting variables with strong correlations.

[0014] Preferably, the method for constructing the signal reconstruction model is as follows: Acquire historical monitoring data from the sensors, divide the historical monitoring data into training and testing sets according to a certain ratio, and use the training set to... The multilayer perceptron is trained, and the test set is used to evaluate the trained data. The predictive performance of multilayer sensing is tested based on the comprehensive mean square error. and coefficient of determination Determine signal The prediction accuracy of a multilayer sensing mechanism can be improved by refreshing the data within the most recent window period. The multi-layer sensing mechanism is dynamically updated until... The prediction results of the multilayer sensing mechanism meet the requirements, and the signal remodel is obtained.

[0015] Preferably, the mean square error The calculation method is as follows:

[0016] wherein, is a parameter signal the first data actual value, is a parameter signal the first data reconstruction value. The determination coefficient The calculation method is as follows:

[0017] wherein, is a parameter signal data actual value mean.

[0018] A combined cycle unit operation state monitoring sensor fault signal dynamic reconstruction system, comprising: A preprocessing module is used for matching the multi-source operation data of the combined cycle unit based on the data timestamp alignment rule to obtain basic operation data; A fault analysis module is used for acquiring the key attention measuring point signal in the basic operation data according to the working logic of the combined cycle unit operation state monitoring and diagnosis system, and analyzing the measuring point signal according to the improved rule method, and detecting the fault of the sensor corresponding to the measuring point signal; A reconstruction module is used for performing correlation analysis on the basic operation data, acquiring the related variables of the fault sensor, constructing a signal characteristic variable matrix according to the related variables, inputting the signal characteristic variable matrix into a signal reconstruction model to obtain the reconstruction signal of the fault sensor, and replacing the abnormal data signal of the fault sensor with the reconstruction signal.

[0019] An electronic device, comprising: A memory is used for storing a computer program; A processor is used for executing the computer program to realize the steps of the combined cycle unit operation state monitoring sensor fault signal dynamic reconstruction method.

[0020] Compared with the prior art, the present application has the following beneficial technical effects: The application provides a kind of combined cycle unit operating state monitoring sensor fault signal dynamic reconstruction method, first utilize data time stamp alignment rule to the multiple-source operating data of combined cycle unit is matched and integrated, ensure the accuracy and consistency of basic data, then, through the improved rule method, the signal of key attention is analyzed in depth, effectively identify sensor fault state, improve the sensitivity and accuracy of fault detection.The method further analyzes the correlation of the basic operating data after detecting the fault, constructs a signal feature variable matrix, and generates a reconstructed signal by means of a signal reconstruction model to replace the abnormal data signal of the faulty sensor, realizing the fault-tolerant control of sensor failure;The method first reduces the false positive and false negative rate by improving the rule method and time stamp alignment technology;Second, the signal reconstruction model built by multilayer perception mechanism can be dynamically adjusted according to real-time data, ensuring the accuracy and reliability of the reconstructed signal, and ensuring the continuous and stable operation of the system in the case of sensor failure.

[0021] The application also provides a combined cycle unit operating state monitoring sensor fault signal dynamic reconstruction system, an electronic device and a computer storage medium, which have all the advantages of the above-mentioned combined cycle unit operating state monitoring sensor fault signal dynamic reconstruction method. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed in the embodiments. It should be understood that the following drawings only show some embodiments of the application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0023] Figure 1 The flowchart of the combined cycle unit operating state monitoring sensor fault signal dynamic reconstruction method of the application.

[0024] Figure 2 The result of step 2 of the application for sensor fault detection of the fuel flow signal of the #3 combined cycle unit of a certain gas turbine power plant.

[0025] Figure 3 The result of step 3 of the application for correlation analysis of the basic data signal of the #3 combined cycle unit of a certain gas turbine power plant.

[0026] Figure 4 The result of step 4 of the application for dynamic reconstruction of the fuel flow signal of the #3 combined cycle unit of a certain gas turbine power plant. DETAILED DESCRIPTION

[0027] ​To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0028] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0029] This invention provides a method for dynamic reconstruction of fault signals from sensors used in monitoring the operating status of combined cycle power units, comprising the following steps: Step 1: Match the multi-source operating data of the combined cycle unit based on the data timestamp alignment rules to obtain the basic operating data; Step 2: Based on the working logic of the combined cycle unit operation status monitoring and diagnostic system, obtain the key measurement point signals from the basic operation data, and then improve accordingly. The criteria method analyzes the measurement point signals and performs fault detection on the corresponding sensors. Step 3: Perform correlation analysis on the basic operating data to obtain the relevant variables of the fault sensor. Construct a signal feature variable matrix based on the relevant variables, input the signal feature variable matrix into the signal reconstruction model to obtain the reconstructed signal of the fault sensor, and use the reconstructed signal to replace the abnormal data signal of the fault sensor.

[0030] This method first matches and integrates the raw data collected by the sensor to form preprocessed basic data; secondly, it utilizes improved... The criterion method is used to detect anomalies in the data signal of the sensor; then, the following is introduced A multilayer sensing mechanism is used to construct a data signal reconstruction model; finally, the normal data signal of the sensor is reconstructed based on the dynamic data of feature variables. This invention can effectively detect signal changes caused by sensor faults, and can also use the model to dynamically reconstruct simulated signals to achieve fault-tolerant control of sensor faults, thereby ensuring the efficient operation and accurate decision-making of the combined cycle unit operation status monitoring and diagnosis system.

[0031] like Figure 1 As shown, a method for dynamic reconstruction of fault signals from sensors used for monitoring the operating status of a combined cycle power unit includes the following steps: Step 1: Given the complex nature of combined cycle unit on-site operating data, including different sources, frequencies, structures, and types, the raw data collected by sensors from all monitoring systems of the unit is first integrated by timestamp interpolation according to the "data timestamp alignment" rule, thus obtaining preprocessed basic operating data. Specifically:

[0032] In the formula, For the preprocessed base data, These are parameters for monitoring the operating status of the generating unit. The timestamp corresponding to the data. For parameters timestamp The monitoring data.

[0033] Considering that, from the gas turbine control system ( Distributed control system for generating units ( ), plant-level monitoring information system ( The operational data collected by sensors in monitoring systems such as [list of systems] often suffers from inconsistencies and misalignments in timestamps due to factors such as scattered data storage, varying data acquisition system configurations, and harsh operating conditions of monitoring instruments. Therefore, to facilitate subsequent diagnostics of the unit's operational status, [the following text is missing]. As needed, parameters can be calculated using linear interpolation. At the same timestamp The value.

[0034] Step 2: Based on the working logic of the combined cycle unit operation status monitoring and diagnosis system, select the key measurement point signals from the basic operation data that require special attention, and then, based on the measurement point signals and in conjunction with improvements... The criterion-based method performs fault detection on the sensors corresponding to the measurement point signals. It detects sensor faults by determining whether the data signals are abnormal. Specifically: Based on the working logic of the combined cycle unit operation status monitoring and diagnostic system, the key measurement point signals selected from the basic operation data include: The main points include the operating parameters of the gas turbine (gas turbine power, gas turbine speed, compressor inlet pressure / temperature, compressor discharge pressure / temperature, fuel pressure / temperature / flow, turbine discharge pressure / temperature, etc.), the operating parameters of the waste heat boiler (high-pressure feed water pressure / temperature / flow, medium-pressure feed water pressure / temperature / flow, condensate water pressure / temperature / flow, etc.), the operating parameters of the steam turbine (pressure / temperature before high-pressure main valve, exhaust pressure / temperature of high-pressure cylinder, pressure / temperature before medium-pressure main valve, pressure / temperature before low-pressure supplementary valve, inlet pressure / temperature of low-pressure cylinder, exhaust pressure of low-pressure cylinder, etc.), and the operating parameters of the cold end (condenser pressure, inlet temperature of condenser circulating water, outlet temperature of condenser circulating water, hot well water temperature, etc.).

[0035] For example, assuming that is the current stable operation time series, the known parameter is the current stable operation signal data , and the data change amount of the current time adjacent time is .

[0036] According to the signal data change amount, the average value of the signal data change amount of adjacent time in the window period is determined as .

[0037] In the formula, is the sliding window period of data signal anomaly detection, .

[0038] According to the change amount average value, the root mean square error of the signal data change amount of adjacent time in the window period is determined as .

[0039] The absolute value of the signal data change amount in the window period is determined as . If , it is confirmed that the signal data of the parameter is a normal value, otherwise it is an abnormal value.

[0040] Similarly, data signal anomaly analysis is performed on all the key points of interest, so as to detect whether the sensor has failed.

[0041] Step 3: Signal reconstruction is performed for a sensor that has failed.

[0042] First, all the basic operation data are normalized, and Correlation analysis is performed to select several variables having high correlation with the fault sensor, and a signal characteristic variable matrix is constructed according to the selected several variables.

[0043] Then, according to The multilayer perception mechanism constructs a signal reconstruction model in the following manner: The historical monitoring data of the sensor are acquired, and the historical monitoring data are divided into a training set and a test set according to a proportion. The multilayer perception mechanism is trained, and the test set is used to test the prediction performance of the trained The multilayer perception mechanism is trained, and the test set is used to test the prediction performance of the trained The prediction result of the multilayer perception mechanism meets the requirement, and a signal reconstruction model is obtained.

[0044] In this embodiment, the comprehensive mean square error , the coefficient of determination and other indexes are used to analyze the prediction result of the model. The multilayer perception mechanism is dynamically updated until The prediction result of the multilayer perception mechanism meets the requirement, and a signal reconstruction model is obtained.

[0045] The normalized basic operation data are as follows:

[0046] In the formula, min (x) is the minimum value of the signal data x, is the normalized basic data x, is a parameter, is the minimum value of the signal data x, is a parameter, is the maximum value of the signal data x. The correlation coefficient matrix of the basic operation data is as follows:

[0047] In the formula, the correlation coefficient matrix of the basic data x is is a parameter, is the correlation coefficient of the signal and the parameter x,

[0048] is the normalized mean value of the signal data x, is the normalized mean value of the signal data x.

[0049] ​​​​​​​​​The range of correlation coefficient is , the symbol "+" represents positive correlation, the symbol "-" represents negative correlation, and the value represents the strength of correlation. The larger the value, the stronger the correlation.

[0050] According to the correlation coefficient matrix, select variables with strong correlation with the parameter signal, and construct a signal feature variable matrix:

[0051] The multi-layer perception includes an input layer, a hidden layer, and an output layer. The weight values between the nodes of each layer of the model can be continuously adjusted according to the training samples until the test results of the model reach a suitable accuracy.

[0052] For a certain parameter , the constructed signal feature variable can be used as input, and the dynamic data of the parameter signal and the dynamic data of the signal feature variable can be used as training and test samples to establish a signal reconstruction model of the parameter .

[0053] At the same time, the mean square error is calculated:

[0054] In the formula, is the actual value of the parameter signal data, is the reconstructed value of the parameter signal data. The smaller the mean square error , the better the model reconstruction effect.

[0055] The determination coefficient is calculated:

[0056]

[0057] In the formula, is the mean value of the actual value of the parameter signal data.

[0058] The closer the determination coefficient is to 1, the better the model reconstruction effect.

[0059] Step 4, based on the real-time data of the signal feature variable, the above signal reconstruction model is used to reconstruct the normal data signal of the sensor in real time, so as to replace the abnormal data signal of the sensor and realize fault tolerance processing. Specifically as follows: For a certain moment​​ Faulty parameter Signal, actual abnormal data of signal Dynamically build signal feature variable data As a signal reconstruction model input, reconstruct the simulated signal when the sensor is normal That is

[0060] Wherein, The multi-layer perception machine can be a linear regression, a polynomial, an exponential function, a logarithmic function, a power function, etc.

[0061] Then, the simulated signal data is used to replace the abnormal data after the sensor fails , to complete the signal reconstruction of the faulty sensor and realize fault-tolerant processing. Embodiment 1

[0062] This embodiment takes the historical operation data of a gas turbine power plant #3 combined cycle unit on March 25, 2025 as an example to explain in detail the dynamic reconstruction method of a combined cycle unit operation state monitoring sensor fault signal according to the present application, including the following steps: Step 1, in view of the complex conditions of different sources, different frequencies, different structures, different types, etc. of the field combined cycle unit field operation data, according to the rule of "data timestamp alignment", the original data collected by all monitoring system sensors of the unit are matched and integrated by timestamp interpolation, so as to obtain the preprocessed basic data. As shown in Table 1, part of the preprocessed basic data after matching and integration is shown.

[0063] Table 1 Part of the preprocessed basic data of a gas turbine power plant #3 combined cycle unit

[0064] Step 2, according to the working logic of the combined cycle unit operation state monitoring and diagnosis system, the measurement point signals that need to be focused on are selected, and the improved rule method is used to carry out fault detection on the sensors corresponding to the measurement point signals one by one, and the sensor fault detection is carried out by judging whether the data signal is abnormal.

[0065] Taking the fuel flow signal fault of #3 combined cycle unit as an example, the Python programming calculation detection obtains the fault result of #3 combined cycle unit fuel flow signal sensor, as shown in Figure 2 . It is detected that the #3 combined cycle unit fuel flow signal sensor fails at 03 / 25 / 2025 15:33:00.

[0066] ​Step 3, for a certain malfunctioning sensor, signal reconstruction is suggested. All basic data can be normalized first, and correlation analysis can be carried out to select several variables with strong correlation from them to form a signal characteristic variable matrix, and then introduce a multilayer perception to build a signal reconstruction model in the latest window period, and analyze the calculation effect of the model by comprehensively considering mean square error , coefficient of determination and other indicators.

[0067] Take the fuel flow signal of the #3 combined cycle unit as an example, carry out correlation analysis through Python programming calculation, and the results are shown in Figure 3 , build a signal characteristic variable matrix, establish a signal reconstruction model, the mean square error is 0.002, the coefficient of determination is 0.958, and the model has good reconstruction effect, as shown in Figure 4 .

[0068] Step 4, based on the real-time data of signal characteristic variables, the above signal reconstruction model is used to reconstruct the normal data signal of the sensor in real time, so as to replace the abnormal data signal of the sensor and realize fault-tolerant processing. Take the fuel flow signal fault of the #3 combined cycle unit as an example, and the reconstructed fuel flow signal of the #3 combined cycle unit is obtained through Python programming calculation, as shown in Figure 4 . The sensor reconstruction signal is used to replace the abnormal signal of the sensor for fault-tolerant control, so as to calibrate or replace the sensor in time.

[0069] The combined cycle unit operation state monitoring sensor fault signal dynamic reconstruction method. By matching and integrating basic data, detecting abnormal data signals, building a signal reconstruction model, and reconstructing signal fault-tolerant control, the dynamic reconstruction of sensor fault signals can be realized. The present application can effectively detect the signal change caused by sensor failure, and can also use the model to dynamically reconstruct the simulated signal to realize the fault-tolerant control of the sensor, so as to ensure the efficient operation and accurate decision of the combined cycle unit operation state monitoring and diagnosis system.

[0070] It should be noted that in the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other manners. For example, the division of the above-described device embodiments is merely an example, and each module can be integrated into another module, or some features can be ignored, or not executed. The division of the modules is merely logical function division, and actual implementation can be other division manners. The modules described as separated components can be or can not be physical separated, and can be or can not be physical components. The modules shown as modules can be one physical unit or multiple physical units, which can be located in one place, or distributed to multiple places. According to actual needs, some or all of the modules can be selected to implement the purposes of the embodiments.

[0071] In addition, each module in the various embodiments of the present application can be integrated into a processing unit, or each module can exist physically, or two or more modules can be integrated into one unit. The above-mentioned integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0072] The electronic device provided in the embodiments of the present application includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the joint cycle unit operation state monitoring sensor fault signal dynamic reconstruction method described in any of the above embodiments when executing the computer program.

[0073] The electronic device provided in another embodiment of the present application can further include: an input port connected to the processor, configured to transmit the multi-modal data collected by an external collection device to the processor; a display unit connected to the processor, configured to display the processing result of the processor to the outside world; and a communication module connected to the processor, configured to realize the communication between the electronic device and the outside world. The display unit can be a display panel, a laser scanning display, etc. The communication mode adopted by the communication module includes but is not limited to mobile high-definition link technology (HML), universal serial bus (USB), high-definition multimedia interface (HDMI), wireless connection (including wireless fidelity technology (WiFi), Bluetooth communication technology, low-power Bluetooth communication technology, and IEEE 802.11s-based communication technology).

[0074] The computer readable storage medium provided in the embodiments of the present application stores a computer program, and the computer program is executed by the processor to implement the steps of the joint cycle unit operation state monitoring sensor fault signal dynamic reconstruction method described in any of the above embodiments.

[0075] The related parts in the computer readable storage medium provided by the embodiments of the present application will be described in detail in the corresponding part of the method for dynamically reconstructing the fault signal of the sensor for monitoring the operating state of the combined cycle unit. Here, no further description is given. In addition, the parts of the above technical solutions provided by the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail, so as not to be too redundant.

[0076] The above is only an illustration of the technical idea of the present application, and cannot limit the protection scope of the present application. Any modification made on the basis of the technical solutions according to the technical idea of the present application falls within the protection scope of the claims of the present application.

Claims

1. A method for dynamic reconstruction of a fault signal of a sensor for monitoring an operating state of a combined cycle unit, characterized in that, The method comprises the following steps: Step 1, matching the multi-source operation data of the combined cycle unit based on a data timestamp alignment rule to obtain basic operation data; Step 2, obtain the key point signals of the basic operation data according to the working logic of the combined cycle unit operation state monitoring and diagnosis system, analyze the point signals according to the improved criterion method, and detect faults of the sensors corresponding to the point signals; Step 3, performing correlation analysis on the basic operation data to obtain a fault sensor related variable, constructing a signal feature variable matrix according to the related variable, inputting the signal feature variable matrix into a signal reconstruction model to obtain a reconstructed signal of the fault sensor, and replacing the abnormal data signal of the fault sensor with the reconstructed signal.

2. The method for dynamic reconstruction of fault signal of a combined cycle unit operation state monitoring sensor according to claim 1, characterized in that, Step 1 of matching the multi-source operation data of the combined cycle unit based on a data timestamp alignment rule comprises: The timestamp interpolation matching method is used to match the multi-source operation data of the sensor to obtain the basic operation data.

3. The method of claim 2, wherein the method further comprises: The timestamp interpolation matching method is as follows: In the formula, is the pre-processed basic operation data, is the unit operation state monitoring parameter, is the time stamp corresponding to the data, is the parameter The time stamp The monitoring data.

4. The method of claim 1, wherein, The key point signals in the basic operation data in step 2 include: gas turbine operation parameters, waste heat boiler operation parameters, steam turbine operation parameters and cold end operation parameters; The gas turbine operation parameters include gas turbine power, gas turbine speed, compressor inlet pressure and temperature, compressor exhaust pressure and temperature, fuel pressure, temperature and flow, turbine exhaust pressure and temperature; The waste heat boiler operation parameters include high-pressure feedwater pressure, temperature and flow, medium-pressure feedwater pressure, temperature and flow, condensate water pressure, temperature and flow; The steam turbine operation parameters include high-pressure main valve front pressure and temperature, high-pressure cylinder exhaust pressure and temperature, medium-pressure main valve front pressure and temperature, low-pressure supplementary valve front pressure and temperature, low-pressure cylinder inlet steam pressure and temperature, low-pressure cylinder exhaust pressure; The cold end operation parameters include condenser pressure, condenser circulating water inlet temperature, condenser circulating water outlet temperature, hot well water temperature.

5. The method for dynamic reconstruction of fault signal of a combined cycle unit operating state monitoring sensor according to claim 1, characterized in that, In step 2, according to the improved The criterion method analyzes the measuring point signal and detects faults of the sensor corresponding to the measuring point signal. The data change amount of the adjacent time of the point signal is determined, the mean value of the signal data change amount of the adjacent time in the window period is determined according to the signal data change amount, the root mean square error of the signal data change amount of the adjacent time in the window period is determined according to the change amount mean value, and the fault state of the sensor is determined according to the absolute value of the signal data change amount in the window period and the root mean square error.

6. The method for dynamic reconstruction of fault signal of a combined cycle unit operating state monitoring sensor according to claim 1, characterized in that, Step 3 of performing correlation analysis on the basic operation data to obtain a fault sensor related variable, and constructing a signal feature variable matrix according to the related variable comprises: The basic operation data is normalized; The base operation data is normalized according to the following formula: The correlation coefficient matrix is as follows: In the formula, is the basic data correlation coefficient matrix, is the parameter signal and parameter signal of correlation coefficient, is the normalized parameter signal data mean value, is the normalized parameter signal data mean value; Based on basic data The correlation coefficient matrix is ​​used to construct a signal feature variable matrix by selecting variables with strong correlations.

7. The method for dynamic reconstruction of fault signal of a combined cycle unit operating state monitoring sensor according to claim 1, characterized in that, The construction method of the signal reconstruction model is as follows: Acquire historical monitoring data from the sensors, divide the historical monitoring data into training and testing sets according to a certain ratio, and use the training set to... The multilayer perceptron is trained, and the test set is used to evaluate the trained data. The predictive performance of multilayer sensing is tested based on the comprehensive mean square error. and coefficient of determination Determine signal The prediction accuracy of a multilayer sensing mechanism can be improved by refreshing the data within the most recent window period. The multi-layer sensing mechanism is dynamically updated until... The prediction results of the multilayer sensing mechanism meet the requirements, and the signal remodel is obtained.

8. The method of claim 7, wherein the method further comprises: The mean square error The calculation method is as follows: wherein is a parameter signal the first data actual value, is a parameter signal the first data reconstructed value; The decision coefficient The calculation method is as follows: In the formula, is a parameter signal data actual value mean. 9.A system for dynamic reconstruction of sensor fault signals for operation state monitoring of a combined cycle unit, characterized in that, It comprises: A preprocessing module is configured to match the multi-source operation data of the combined cycle unit based on a data timestamp alignment rule to obtain basic operation data; The fault analysis module is used for obtaining the key attention measuring point signals in the basic operation data according to the working logic of the combined cycle unit operation state monitoring and diagnosis system, and detecting the faults of the sensors corresponding to the measuring point signals according to the improved criterion method. The fault analysis module is used for obtaining the key attention measuring point signals in the basic operation data according to the working logic of the combined cycle unit operation state monitoring and diagnosis system, and detecting the faults of the sensors corresponding to the measuring point signals according to the improved criterion method. A reconstruction module is configured to perform correlation analysis on the basic operation data to obtain a fault sensor related variable, construct a signal feature variable matrix according to the related variable, input the signal feature variable matrix into a signal reconstruction model to obtain a reconstructed signal of the fault sensor, and replace the abnormal data signal of the fault sensor with the reconstructed signal.

10. An electronic device, comprising: It comprises: A memory is configured to store a computer program; A processor is configured to execute the computer program to implement the steps of the combined cycle unit operation state monitoring sensor fault signal dynamic reconstruction method according to any one of claims 1-8.

Images