Heating surface tube panel leakage alarm on-line monitoring system

Abstract

The invention provides a heating surface tube panel leakage alarm online monitoring system, which relates to the technical field of boiler detection and comprises a multi-source data acquisition module, a multi-source data processing module, an intelligent detection module, a leakage trend prediction module and a maintenance decision management module. Collecting flue gas temperature data, sound wave sensing data and hearth pressure data, performing corresponding processing to obtain a temperature gradient matrix, a sound wave time-frequency feature vector and a hearth pressure parameter sequence, and performing fusion feature extraction processing to obtain a fusion feature vector matrix; and performing leakage identification detection on the fusion feature vector matrix to obtain a dynamic threshold compensation coefficient, further obtaining a leakage state judgment result, constructing a leakage trend prediction model, generating a boiler health prediction index, generating a leakage risk level, further executing an optimal maintenance scheme, and improving the safety of the boiler during operation.

Description

Technical Field

[0001] The present invention relates to the technical field of boiler detection, and specifically to an online monitoring system for leakage alarm of a heating surface tube panel. Background Art

[0002] The automatic boiler tube leakage alarm system is an auxiliary control system for power station boilers, which is widely used in power station boilers of thermal power plants and heating and steam supply boilers in the metallurgical and petrochemical systems. At present, there is no specific method to judge the leakage of the boiler heating surface tube panel. Usually, flow, pressure, and temperature measuring points are added to the header pipes where the working medium enters and exits the heating surface. By measuring the real-time (flow, temperature) parameters of the working medium entering and exiting the heating surface, it is indirectly judged whether there is a leakage in the heating surface.

[0003] During the actual operation process, the working medium parameters (flow, temperature) entering and exiting the heating surface will fluctuate due to the unit load. It is inaccurate to judge whether there is a leakage in the heating surface only through the change of the working medium parameters (flow, temperature) entering and exiting the heating surface. Secondly, there are certain errors in the measurement means of the working medium parameters itself, and the error will be amplified when superimposed for judgment. For the leakage situation with a small flow rate, the flowmeter on the pipeline has certain limitations and cannot judge whether there is a leakage.

[0004] To avoid the further spread of leakage, reduce the time of unplanned shutdown of the unit caused by leakage accidents and the maintenance cost after dealing with leakage, and accurately monitor in real time whether there is a leakage in the heating surface tube panel and give an early warning in the first time to protect the safe operation of the unit.

[0005] Therefore, an online monitoring system for leakage alarm of a heating surface tube panel is provided now. Summary of the Invention

[0006] In order to solve the above technical problems, the purpose of the present invention is to provide an online monitoring system for leakage alarm of a heating surface tube panel.

[0007] In order to achieve the above purpose, the present invention provides the following technical solution: An online monitoring system for leakage alarm of a heating surface tube panel, comprising: a multi-source data acquisition module, a multi-source data processing module, an intelligent detection module, a leakage trend prediction module, and a maintenance decision management module; The multi-source data acquisition module is used to acquire flue gas temperature data, acoustic wave sensing data, and furnace pressure data; The multi-source data processing module is used to perform matrix processing on the flue gas temperature data to obtain a temperature gradient matrix; perform feature vector quantization processing on the acoustic wave sensing data to obtain an acoustic wave time-frequency feature vector; perform serialization processing on the furnace pressure data to obtain a furnace pressure parameter sequence; perform fusion feature extraction processing on the temperature gradient matrix, the acoustic wave time-frequency feature vector, and the furnace pressure parameter sequence to obtain a fusion feature vector matrix; The intelligent detection module is used to perform leakage identification and detection on the fused feature vector matrix, obtain a dynamic threshold compensation coefficient, and based on the dynamic threshold compensation coefficient, obtain a leakage state determination result, and further obtain the location of the furnace tube leakage area; The leakage trend prediction module constructs a leakage trend prediction model based on the leakage state determination result and generates a boiler health prediction index; The maintenance decision management module generates a leakage risk level based on the boiler health prediction index and further executes an optimal maintenance plan.

[0008] According to one preferred embodiment of the present invention, the process of real-time collecting flue gas temperature data, acoustic wave sensing data, and furnace pressure data includes: Setting up a data collection device; The data collection device consists of several sensors, including a temperature sensor, an acoustic wave sensor, and a pressure sensor; the temperature sensor, the acoustic wave sensor, and the pressure sensor are respectively arranged at the corresponding measuring points of the boiler furnace tubes, and a data collection period is set, and the collection period includes several collection moments; The temperature sensor is used to collect the flue gas temperature data of the corresponding measuring point at the collection moment; the acoustic wave sensor is used to collect the acoustic wave sensing data of the corresponding measuring point according to the collection moment; the pressure sensor is used to collect the furnace pressure data of the corresponding measuring point according to the collection moment.

[0009] According to one preferred embodiment of the present invention, the process of matrix processing the flue gas temperature data includes: Obtain temperature sensors, and represent the collected flue gas temperature data as a two-dimensional array ; where ; represents the row index, ; represents the column index, represents the flue gas temperature data collected by the temperature sensor at the th row and the th column; Perform a moving average filtering process on the flue gas temperature data, including performing a moving average filtering process on the flue gas temperature data in the row direction and the flue gas temperature data in the column direction; And perform temperature gradient calculations on the flue gas temperature data in the column direction and the row direction , and decompose the flue gas temperature data in the column direction and the row direction into flue gas temperature data in the x direction and the y direction on a two-dimensional plane , obtain the temperature gradient component of the flue gas temperature data in the x direction and the flue gas temperature data in the y direction The temperature gradient component; Combine the temperature gradient components in the x - direction and y - direction into a two - dimensional matrix, denoted as the temperature gradient matrix.

[0010] According to one preferred embodiment of the present invention, the process of feature vectorizing acoustic sensing data includes: Obtain the acoustic sensing data; mark the acoustic sensing data as , where ; is a natural number; Based on statistical techniques, obtain the kurtosis, skewness, and variance of the acoustic sensing data ; based on time - domain statistics, obtain the root - mean - square value and the peak factor; based on the short - time Fourier transform, obtain the time - frequency matrix corresponding to the acoustic sensing data ; According to the time - frequency matrix, and based on band integration and low - frequency band integration, obtain the leakage energy ratio and the background noise energy; Based on frequency peak detection, obtain the main frequency and bandwidth in the time - frequency matrix; According to the kurtosis, skewness, variance, root - mean - square value, peak factor, leakage energy, background noise energy, main frequency, and bandwidth, perform combination to obtain the acoustic time - frequency feature vector.

[0011] According to one preferred embodiment of the present invention, the process of serializing furnace pressure data includes: Obtain the furnace pressure data; mark the furnace pressure data as , where ; is a natural number; Based on the moving average method, perform smoothing processing on the marked furnace pressure data to obtain the smoothed furnace pressure data ; normalize the smoothed furnace pressure data to the interval, and the normalized furnace pressure data is denoted as ; According to the normalized furnace pressure data denoted as , obtain the mean, standard deviation, maximum value, and minimum value of the normalized furnace pressure data ; Perform first - order difference processing on the normalized furnace pressure data to obtain the difference , obtain the number of greater than or equal to 0, and then obtain the positive difference ratio; obtain the number of less than 0, and then obtain the negative difference ratio; According to the mean value, standard deviation, maximum value, minimum value, positive difference ratio, and negative difference ratio, a furnace pressure parameter sequence is obtained.

[0012] According to one preferred embodiment of the present invention, the process of performing fusion feature extraction on the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence includes: Obtain the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence; Based on Z-score standardization, perform standardization processing on the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence; Convert the standardized temperature gradient matrix into a one-dimensional feature vector; For the one-dimensional feature vector, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence Perform weighted fusion processing to obtain a fusion feature vector matrix , the fusion feature vector matrix is: ; where is the one-dimensional feature vector, the weight of the one-dimensional feature vector ; is the acoustic wave time-frequency feature vector, is the weight of the acoustic wave time-frequency feature vector; is the furnace pressure parameter sequence, is the weight of the furnace pressure parameter sequence.

[0013] According to one preferred embodiment of the present invention, the process of performing leakage identification and detection on the fusion feature vector matrix to obtain a dynamic threshold compensation coefficient and obtaining a leakage state determination result based on the dynamic threshold compensation coefficient includes: Obtain the fusion feature vector matrix ; Based on a spatio-temporal fusion neural network, obtain the corresponding dynamic threshold coefficient in the fusion feature vector matrix ; The dynamic threshold coefficient includes a temperature threshold coefficient and an acoustic wave threshold coefficient, denoted as and respectively; Preset a standard flue gas temperature threshold, a standard acoustic wave sensing threshold, and a standard furnace pressure threshold; according to the standard flue gas temperature threshold, the standard acoustic wave sensing threshold, and the standard furnace pressure threshold, obtain a standard environmental parameter input, denoted as ; according to the fusion feature vector matrix The corresponding flue gas temperature data, acoustic wave sensing data, and furnace pressure data are obtained to get the actual environmental parameter input, denoted as ; According to the standard environmental parameter input and the actual environmental parameter input , the compensation formula is obtained as: ; where is the standard deviation of environmental fluctuations; According to the temperature threshold coefficient , the acoustic wave threshold coefficient and the compensation formula , the dynamic threshold compensation coefficient is obtained. The dynamic threshold compensation coefficient includes a temperature threshold compensation coefficient and an acoustic wave threshold compensation coefficient, denoted as and ; The temperature threshold compensation coefficient ; The acoustic wave threshold compensation coefficient ; Set the standard temperature compensation threshold and the standard acoustic wave compensation threshold ; According to the dynamic threshold compensation coefficient, a dual-channel verification mechanism is generated; The temperature channel verification is: ; The acoustic wave channel verification is: ; When the value is 1, it means the corresponding channel is triggered, and when the value is 0, it means the corresponding channel is not triggered; According to the dual-channel verification mechanism, the leakage state determination result is output. The leakage state determination result is: .

[0014] According to one preferred embodiment of the present invention, the process of constructing a leakage trend prediction model includes: Obtain the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence corresponding to several groups of leakage state determination results; Group and label the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence corresponding to several groups of leakage state determination results, denoted as is a natural number; Take groups of the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence corresponding to several groups of leakage state determination results, the standard flue gas temperature threshold, the standard acoustic wave sensing threshold, and the standard furnace pressure threshold as sample data, and is a natural number less than , and using the sample data, the mean value of the sample data is obtained and denoted as the sample set; The temperature gradient matrix, the acoustic wave time-frequency feature vector, the furnace pressure parameter sequence corresponding to the determination results of the remaining several groups of leakage states, and the standard flue gas temperature threshold, the standard acoustic wave sensing threshold, and the standard furnace pressure threshold are used as the test set; According to the sample set and the test set, a training sample set is formed; Based on the convolutional neural network, a standard prediction model is constructed; And the training sample set is input into the standard prediction model, the standard prediction model is trained, and the trained standard prediction model is obtained, and the trained standard prediction model is denoted as the leakage trend prediction model; According to the leakage trend prediction model, a boiler health prediction index is generated under the current environmental factor conditions , and the boiler health prediction index is: ; where represents the flue gas temperature data, represents the standard flue gas temperature threshold, represents the proportionality coefficient of the temperature influence degree, represents the acoustic wave sensing data, represents the standard acoustic wave sensing threshold, represents the furnace pressure data, represents the standard furnace pressure threshold.

[0015] According to one preferred embodiment of the present invention, the process of generating the leakage risk level according to the boiler health prediction index includes: Preset the standard boiler health prediction index ; If the boiler health prediction index , there is no leakage risk in the water wall tubes, superheater tubes, reheater tubes, and economizer tubes in the boiler under the current environmental factor conditions; If the boiler health prediction index , there is a leakage risk in the water wall tubes, superheater tubes, reheater tubes, and economizer tubes in the boiler under the current environmental factor conditions, and a leakage risk level is generated; it should be further noted that the leakage risk level includes a first-level leakage alarm and a second-level leakage alarm; Preset the minimum boiler health prediction index , if , there is a risk of sand hole leakage in the water wall tubes, superheater tubes, reheater tubes, and economizer tubes in the boiler, a first-level leakage alarm is generated, and a maintenance plan for the sand hole leakage risk is executed; if When there is a risk of crack leakage in the water-cooled wall tubes, superheater tubes, reheater tubes, and economizer tubes in the boiler, a secondary leakage alarm is generated, and a maintenance plan for crack leakage risk is executed.

[0016] The present invention further provides a computer-readable storage medium storing a computer program executable by a processor to implement the above-mentioned on-line monitoring system for leakage alarm of heating surface tube screens.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. Smoke temperature data, acoustic wave sensing data, and furnace pressure data are collected and corresponding processing is performed to obtain a temperature gradient matrix, an acoustic wave time-frequency feature vector, and a furnace pressure parameter sequence, and fusion feature extraction processing is performed to obtain a fusion feature vector matrix; leakage identification and detection are performed on the fusion feature vector matrix to obtain a dynamic threshold compensation coefficient, and then a leakage state determination result is obtained, reducing the influence of unit load fluctuations and the interference of measurement errors themselves, being able to monitor more real-time and accurately whether there is leakage in the heating surface tube screen, avoiding misjudgment caused by error amplification, solving the problem of difficult judgment of small-flow leakage, and improving the accuracy of detection.

[0018] 2. A leakage trend prediction model is constructed, a boiler health prediction index is generated, a leakage risk level is generated, and then an optimal maintenance plan is executed, which can not only give early warnings in time to avoid leakage spread, but also reduce the unplanned outage time of the unit, lower the maintenance cost, provide a strong guarantee for the safe and stable operation of the unit, and at the same time improve the economic benefits and management efficiency of the unit operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings.

[0020] Figure 1 It is a step schematic diagram of an on-line monitoring system for leakage alarm of heating surface tube screens.

[0021] Figure 2 It is a flow schematic diagram of an on-line monitoring system for leakage alarm of heating surface tube screens.

[0022] Figure 3 It is a module schematic diagram of an on-line monitoring system for leakage alarm of heating surface tube screens. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be described in detail below. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0024] As Figure 1 shown, a on-line monitoring system for leakage alarm of a heating surface tube screen includes: a multi-source data acquisition module, a multi-source data processing module, an intelligent detection module, a leakage trend prediction module, and a maintenance decision management module; The multi-source data acquisition module is used to acquire flue gas temperature data, acoustic wave sensing data, and furnace pressure data; The multi-source data processing module is used to perform matrix processing on the flue gas temperature data to obtain a temperature gradient matrix; perform feature vector processing on the acoustic wave sensing data to obtain an acoustic wave time-frequency feature vector; perform serialization processing on the furnace pressure data to obtain a furnace pressure parameter sequence; and perform fusion feature extraction processing on the temperature gradient matrix, the acoustic wave time-frequency feature vector, and the furnace pressure parameter sequence to obtain a fusion feature vector matrix; The intelligent detection module is used to perform leakage identification detection on the fusion feature vector matrix to obtain a dynamic threshold compensation coefficient, and based on the dynamic threshold compensation coefficient, obtain a leakage state determination result, and further obtain the position of the furnace tube leakage area; The leakage trend prediction module is used to construct a leakage trend prediction model based on the leakage state determination result and generate a boiler health prediction index; The maintenance decision management module is used to generate a leakage risk level based on the boiler health prediction index, and then execute an optimal maintenance plan.

[0025] It should be further noted that in the specific implementation process, the specific process of real-time acquisition of flue gas temperature data, acoustic wave sensing data, and furnace pressure data includes: Set up a data acquisition device; The data acquisition device is composed of several sensors, including a temperature sensor, an acoustic wave sensor, and a pressure sensor; the temperature sensor, the acoustic wave sensor, and the pressure sensor are respectively arranged at the corresponding measuring points of the boiler furnace tubes, and a data acquisition period is set, and the acquisition period includes several acquisition times; The temperature sensor is used to acquire the flue gas temperature data of the corresponding measuring point according to the acquisition time; the acoustic wave sensor is used to acquire the acoustic wave sensing data of the corresponding measuring point according to the acquisition time; the pressure sensor is used to acquire the furnace pressure data of the corresponding measuring point according to the acquisition time.

[0026] It should be further noted that in the specific implementation process, the specific process of matrix processing the flue gas temperature data includes: Obtain temperature sensors, and represent the collected flue gas temperature data as a two-dimensional array ; where ; represents the row index, ; represents the column index, represents the th row and the th column of the flue gas temperature data collected by the temperature sensor; Perform a moving average filtering process on the flue gas temperature data to remove high-frequency noise. The formula for the moving average filtering is: ; It should be further noted that this formula performs a moving average filtering process on the flue gas temperature data in the row direction; ; It should be further noted that this formula performs a moving average filtering process on the flue gas temperature data in the column direction; where is the flue gas temperature data after the moving average filtering process; Perform a temperature gradient calculation on the flue gas temperature data in the column direction and the row direction , and the specific process includes: Decompose the flue gas temperature data in the column direction and the row direction into the flue gas temperature data in the x direction and the y direction on a two-dimensional plane ; The flue gas temperature data in the x direction is represented as the flue gas temperature data in the column direction ; The flue gas temperature data in the y direction is represented as the flue gas temperature data in the row direction ; The temperature gradient component of the flue gas temperature data in the x direction is: ; where is the distance between adjacent sensors in the x direction; The temperature gradient component of the flue gas temperature data in the y direction is: ; where is the distance between adjacent sensors in the y direction; It should be further noted that forward difference or backward difference can be calculated; For example, when , , perform a forward difference calculation along the x direction; when When ; backward difference calculation in the x direction; when When , forward difference calculation in the y direction; when When ; backward difference calculation in the y direction; Combine the temperature gradient components in the x direction and the y direction into a two-dimensional matrix, denoted as the temperature gradient matrix , the temperature gradient matrix is: .

[0027] It should be further noted that in the specific implementation process, the specific process of feature vectorization of acoustic wave sensing data includes: Obtain acoustic wave sensing data; Mark the acoustic wave sensing data, denoted as , where ; is a natural number; Based on statistical techniques, obtain the kurtosis, skewness, and variance of the acoustic wave sensing data , denoted as , and ; Based on time-domain statistics, obtain the root mean square value and the peak factor, denoted as and ; Based on the short-time Fourier transform, obtain the time-frequency matrix corresponding to the acoustic wave sensing data ; According to the time-frequency matrix , and based on frequency band integration and low-frequency band integration, obtain the leakage energy ratio and the background noise energy, denoted as and ; it should be further noted that the leakage energy ; the background noise energy , where is set according to the actual situation; Based on frequency peak detection, obtain the main frequency and the bandwidth in the time-frequency matrix , denoted as and ; According to the kurtosis , skewness , variance , root mean square value , peak factor , leakage energy Background noise energy Main frequency And bandwidth are combined to obtain a time-frequency feature vector of the acoustic wave The time-frequency feature vector of the acoustic wave is as follows .

[0028] It should be further noted that in the specific implementation process, the specific process of serializing the furnace pressure data includes Obtain furnace pressure data Mark the furnace pressure data, denoted as , where ; is a natural number Based on the moving average method, smooth the marked furnace pressure data to reduce noise interference and obtain the smoothed furnace pressure data The smoothed furnace pressure data is as follows ; where is the window size Normalize the smoothed furnace pressure data to the interval for convenient subsequent processing and analysis. The normalized furnace pressure data is denoted as ; Obtain the mean value of the normalized furnace pressure data The mean value is as follows ; Obtain the standard deviation of the normalized furnace pressure data The mean value is as follows ; Denote the maximum value in the normalized furnace pressure data as , and the minimum value as ; Perform a first-order difference operation on the normalized furnace pressure data to obtain the difference The difference is ; where is less than natural numbers; Obtain the number greater than or equal to 0, and then obtain the positive difference ratio, denoted as ; Obtain the number less than 0, and then obtain the negative difference ratio, denoted as ; According to the mean value , standard deviation , , , positive difference ratio and negative difference ratio , obtain the furnace pressure parameter sequence , the furnace pressure parameter sequence is: .

[0029] It should be further noted that in the specific implementation process, the specific process of performing fusion feature extraction on the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence includes: Obtain the temperature gradient matrix , acoustic wave time-frequency feature vector and furnace pressure parameter sequence ; Based on Z-score standardization, perform standardization processing on the temperature gradient matrix , acoustic wave time-frequency feature vector and furnace pressure parameter sequence to eliminate the influence of dimension and scale between different data; Convert the standardized temperature gradient matrix into a one-dimensional feature vector, denoted as ; It should be further noted that is the value range of the one-dimensional feature vector ; Perform weighted fusion processing on the one-dimensional feature vector , acoustic wave time-frequency feature vector and furnace pressure parameter sequence to obtain the fusion feature vector matrix , the fusion feature vector matrix is: ; where is the weight of the one-dimensional feature vector ; is the weight of the acoustic wave time-frequency feature vector; is the weight of the furnace pressure parameter sequence; and .

[0030] It should be further noted that in the specific implementation process, the process of performing leakage identification and detection on the fused feature vector matrix to obtain the dynamic threshold compensation coefficient and obtaining the leakage state determination result according to the dynamic threshold compensation coefficient includes: Obtain the fused feature vector matrix ; Based on the spatio-temporal fusion neural network, obtain the dynamic threshold coefficient corresponding to that in ; The dynamic threshold coefficient includes a temperature threshold coefficient and an acoustic wave threshold coefficient, which are respectively denoted as and ; Preset the standard flue gas temperature threshold, the standard acoustic wave sensing threshold, and the standard furnace pressure threshold; according to the standard flue gas temperature threshold, the standard acoustic wave sensing threshold, and the standard furnace pressure threshold, obtain the standard environmental parameter input, denoted as ; according to the flue gas temperature data, the acoustic wave sensing data, and the furnace pressure data corresponding to that in the fused feature vector matrix obtain the actual environmental parameter input, denoted as ; According to the standard environmental parameter input and the actual environmental parameter input , the compensation formula is obtained as: ; where is the standard deviation of environmental fluctuations; According to the temperature threshold coefficient , the acoustic wave threshold coefficient , and the compensation formula , obtain the dynamic threshold compensation coefficient, and the dynamic threshold compensation coefficient includes a temperature threshold compensation coefficient and an acoustic wave threshold compensation coefficient, which are respectively denoted as and ; It should be further noted that the temperature threshold compensation coefficient ; the acoustic wave threshold compensation coefficient ; Set the standard temperature compensation threshold and the standard acoustic wave compensation threshold ; Generate a dual-channel verification mechanism according to the dynamic threshold compensation coefficient; it should be further noted that the dual-channel verification mechanism includes temperature channel verification and acoustic wave channel verification, and it is judged that the leakage point is near a certain measurement point, between two certain measurement points or multiple measurement points, and the preset optimal isolation range is a spherical radius of 8m; The temperature channel verification is: ; The verification of the acoustic wave channel is as follows: ; It should be further noted that when the value is 1, it represents that the corresponding channel is triggered, and when the value is 0, it represents that the corresponding channel is not triggered; According to the dual-channel verification mechanism, the determination result of the leakage state is output, and the determination result of the leakage state is as follows: ; Obtain the flue gas temperature data, acoustic wave sensor data, and furnace pressure data with the determination results of leakage and suspicion of leakage status, and obtain the flue gas temperature data, acoustic wave sensor data, and furnace pressure data at other acquisition moments during the corresponding historical acquisition period, denoted as historical flue gas temperature data, historical acoustic wave sensor data, and historical furnace pressure data; Based on the multiple linear regression model, convert the historical flue gas temperature data, historical acoustic wave sensor data, and historical furnace pressure data during the corresponding historical acquisition period of the multiple linear regression model into a real-time leakage trend curve that changes with time. The real-time leakage trend curve includes a flue gas temperature data trend curve, an acoustic wave sensor data trend curve, and a furnace pressure data trend curve; And display the flue gas temperature data trend curve, the acoustic wave sensor data trend curve, and the furnace pressure data trend curve to facilitate the analysis and judgment of the staff; It should be further noted that in the specific implementation process, the hypothesis function adopted by the multiple linear regression model is: ; where is the operating state value of the boiler furnace tube; is the historical flue gas temperature data; is the historical acoustic wave sensor data; is the historical furnace pressure data; is the weight value of the historical flue gas temperature data; is the weight value of the historical acoustic wave sensor data; is the weight value of the historical furnace pressure data.

[0031] It should be further noted that in the specific implementation process, the specific process of constructing the leakage trend prediction model includes: Obtain a number of temperature gradient matrices, acoustic wave time-frequency feature vectors, and furnace pressure parameter sequences corresponding to the determination results of the leakage state; Group and label a number of temperature gradient matrices, acoustic wave time-frequency feature vectors, and furnace pressure parameter sequences corresponding to the determination results of the leakage state, denoted as where is a natural number; Group the temperature gradient matrices, acoustic wave time-frequency eigenvectors, furnace pressure parameter sequences corresponding to several groups of leakage state determination results, and standard flue gas temperature thresholds, standard acoustic wave sensing thresholds, and standard furnace pressure thresholds as sample data, and is a natural number less than , and using the sample data, obtain the sample data mean, denoted as the sample set; Take the temperature gradient matrices, acoustic wave time-frequency eigenvectors, furnace pressure parameter sequences corresponding to the remaining several groups of leakage state determination results, and standard flue gas temperature thresholds, standard acoustic wave sensing thresholds, and standard furnace pressure thresholds as the test set; According to the sample set and the test set, form a training sample set; Based on the convolutional neural network, construct a standard prediction model; And input the training sample set into the standard prediction model, train the standard prediction model, obtain the trained standard prediction model, and denote the trained standard prediction model as the leakage trend prediction model; According to the leakage trend prediction model, generate the boiler health prediction index under the current environmental factor conditions , the boiler health prediction index is: ; where represents the flue gas temperature data, represents the standard flue gas temperature threshold, represents the proportionality coefficient of the temperature influence degree, represents the acoustic wave sensing data, represents the standard acoustic wave sensing threshold, represents the furnace pressure data, represents the standard furnace pressure threshold.

[0032] It should be further noted that in the specific implementation process, the specific process of generating the leakage risk level according to the boiler health prediction index includes: Preset the standard boiler health prediction index ; If the boiler health prediction index , then there is no leakage risk in the inner water-cooled wall tubes, superheater tubes, reheater tubes, and economizer tubes of the boiler under the current environmental factor conditions; If the boiler health prediction index , then there is a leakage risk in the inner water-cooled wall tubes, superheater tubes, reheater tubes, and economizer tubes of the boiler under the current environmental factor conditions, and generate the leakage risk level; it should be further noted that the leakage risk level includes a first-level leakage alarm and a second-level leakage alarm; Preset the minimum boiler health prediction index , if When there is a risk of leakage due to sand holes in the water wall tubes, superheater tubes, reheater tubes, and economizer tubes in the boiler, a first-level leakage alarm is generated, and a maintenance plan for the risk of leakage due to sand holes is executed; if When there is a risk of leakage due to cracks in the water wall tubes, superheater tubes, reheater tubes, and economizer tubes in the boiler, a second-level leakage alarm is generated, and a maintenance plan for the risk of leakage due to cracks is executed.

[0033] The above embodiments are only used to illustrate the technical method of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An online monitoring system for leakage alarm of a heating surface tube screen, characterized in that, Including: A multi-source data acquisition module for acquiring flue gas temperature data, acoustic wave sensing data, and furnace pressure data; A multi-source data processing module for performing matrix processing on the flue gas temperature data to obtain a temperature gradient matrix; performing feature vectorization processing on the acoustic wave sensing data to obtain an acoustic wave time-frequency feature vector; performing serialization processing on the furnace pressure data to obtain a furnace pressure parameter sequence; and performing fusion feature extraction processing on the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence to obtain a fusion feature vector matrix; An intelligent detection module for performing leakage identification detection on the fusion feature vector matrix to obtain a dynamic threshold compensation coefficient, and obtaining a leakage state determination result based on the dynamic threshold compensation coefficient, and further obtaining the position of the furnace tube leakage area; A leakage trend prediction module for constructing a leakage trend prediction model based on the leakage state determination result and generating a boiler health prediction index; A maintenance decision management module for generating a leakage risk level based on the boiler health prediction index and further executing an optimal maintenance plan.

2. The online monitoring system for leakage alarm of a heating surface tube screen according to claim 1, characterized in that, The process of real-time acquiring flue gas temperature data, acoustic wave sensing data, and furnace pressure data includes: Setting up a data acquisition device; The data acquisition device is composed of a number of sensors, including a temperature sensor, an acoustic wave sensor, and a pressure sensor; the temperature sensor, acoustic wave sensor, and pressure sensor are respectively arranged at the corresponding measuring points of the boiler furnace tubes, and a data acquisition period is set, and the acquisition period includes a number of acquisition moments; The temperature sensor is used to acquire the flue gas temperature data of the corresponding measuring point at the acquisition moment; the acoustic wave sensor is used to acquire the acoustic wave sensing data of the corresponding measuring point according to the acquisition moment; the pressure sensor is used to acquire the furnace pressure data of the corresponding measuring point according to the acquisition moment.

3. An online monitoring system for leakage alarm of a heating surface tube screen according to claim 2, characterized in that The process of performing matrix processing on the flue gas temperature data includes: Obtain temperature sensors, and represent the collected flue gas temperature data as a two-dimensional array ; where ; represents the row index, ; represents the column index, represents the th row and the th column of the flue gas temperature data collected by the temperature sensor; Performing moving average filtering processing on the flue gas temperature data, including performing moving average filtering processing on the flue gas temperature data in the row direction and the column direction; And calculate the temperature gradients of the flue gas temperature data in the column direction and the row direction Perform temperature gradient calculations on the flue gas temperature data in the column direction and the row direction In a two-dimensional plane, decompose the flue gas temperature data in the column direction and the row direction into the flue gas temperature data in the x direction and the y direction to obtain the temperature gradient components of the flue gas temperature data in the x direction and the temperature gradient components of the flue gas temperature data in the y direction; Combining the temperature gradient components in the x direction and the y direction into a two-dimensional matrix, denoted as the temperature gradient matrix.

4. An on-line monitoring system for leakage alarm of a heating surface tube screen according to claim 3, characterized in that, The process of performing feature vectorization processing on the acoustic wave sensing data includes: Obtain acoustic wave sensing data; mark the acoustic wave sensing data as , where ; is a natural number; Obtain acoustic wave sensing data The kurtosis, skewness, and variance; Based on time-domain statistics, obtain the root mean square value and the peak factor; Based on the short-time Fourier transform, obtain the acoustic wave sensing data The corresponding time-frequency matrix; Based on the time-frequency matrix, and based on frequency band integration and low-frequency band integration, obtaining the leakage energy ratio and the background noise energy; Based on frequency peak detection, obtaining the main frequency and bandwidth in the time-frequency matrix; Combining the kurtosis, skewness, variance, root mean square value, peak factor, leakage energy, background noise energy, main frequency, and bandwidth to obtain an acoustic wave time-frequency feature vector.

5. An on-line monitoring system for detecting leakage of a heating surface tube screen according to claim 4, characterized in that, The process of performing serialization processing on the furnace pressure data includes: Obtain furnace pressure data; mark the furnace pressure data, denoted as , where ; is a natural number; Based on the moving average method, the marked furnace pressure data is smoothed to obtain the smoothed furnace pressure data ; The smoothed furnace pressure data is normalized to the interval, and the normalized furnace pressure data is denoted as ; According to the normalized furnace pressure data denoted as , obtain the mean, standard deviation, maximum value, and minimum value of the normalized furnace pressure data ; For the normalized furnace pressure data Perform first-order difference processing to obtain the difference , and obtain The number greater than or equal to 0, and then obtain the positive difference ratio; obtain The number less than 0, and then obtain the negative difference ratio; Obtaining a furnace pressure parameter sequence according to the mean value, standard deviation, maximum value, minimum value, positive difference ratio, and negative difference ratio.

6. An on-line monitoring system for leakage alarm of a heating surface tube screen according to claim 5, characterized in that, The process of performing fusion feature extraction processing on the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence includes: Obtaining the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence; Based on Z-score standardization, perform standardization processing on the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence; Convert the standardized temperature gradient matrix into a one-dimensional feature vector; For the one-dimensional feature vector, the acoustic wave time-frequency feature vector, and the furnace pressure parameter sequence perform weighted fusion processing to obtain a fused feature vector matrix , and the fused feature vector matrix is: ; wherein, is a one-dimensional feature vector, the one-dimensional feature vector weight; is a time-frequency feature vector of sound wave, is the weight of the time-frequency feature vector of sound wave; is a sequence of furnace pressure parameters, is the weight of the sequence of furnace pressure parameters.

7. An online monitoring system for leakage alarm of a heating surface tube screen according to claim 6, characterized in that, The process of performing leakage identification and detection on the fused feature vector matrix to obtain a dynamic threshold compensation coefficient and obtaining a leakage state determination result based on the dynamic threshold compensation coefficient includes: Obtain the fused feature vector matrix ; Based on the spatio-temporal fusion neural network, obtain the fused feature vector matrix The corresponding dynamic threshold coefficient in The dynamic threshold coefficients include a temperature threshold coefficient and an acoustic wave threshold coefficient, denoted respectively as and ; Preset the standard flue gas temperature threshold, the standard acoustic wave sensing threshold, and the standard furnace pressure threshold; according to the standard flue gas temperature threshold, the standard acoustic wave sensing threshold, and the standard furnace pressure threshold, obtain the standard environmental parameter input, denoted as ; according to the flue gas temperature data, the acoustic wave sensing data, and the furnace pressure data corresponding in the fusion feature vector matrix, obtain the actual environmental parameter input, denoted as ; According to the input of the standard environmental parameters and the input of the actual environmental parameters , the compensation formula obtained is: ; wherein, is the standard deviation of environmental fluctuations; According to the temperature threshold coefficient , the acoustic wave threshold coefficient , and the compensation formula , a dynamic threshold compensation coefficient is obtained. The dynamic threshold compensation coefficient includes a temperature threshold compensation coefficient and an acoustic wave threshold compensation coefficient, which are respectively denoted as and ; The temperature threshold compensation coefficient ; The acoustic wave threshold compensation coefficient ; Set the standard temperature compensation threshold and the standard acoustic wave compensation threshold ; Generate a dual-channel verification mechanism according to the dynamic threshold compensation coefficient; The temperature channel verification is: ； The acoustic wave channel verification is: ； When the value is 1, it represents that the corresponding channel is triggered, and when the value is 0, it represents that the corresponding channel is not triggered; According to the dual-channel verification mechanism, output the leakage state determination result, and the leakage state determination result is: 。 8. An on-line monitoring system for leakage alarm of a heating surface tube screen according to claim 7, characterized in that, The process of constructing a leakage trend prediction model includes: Obtain the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence corresponding to several groups of leakage state determination results; Group numbers are assigned to the temperature gradient matrices, acoustic wave time-frequency feature vectors, and furnace pressure parameter sequences corresponding to several groups of leakage state determination results, denoted as is a natural number; Take a number of temperature gradient matrices corresponding to the leakage state determination results, acoustic wave time-frequency feature vectors, furnace pressure parameter sequences, standard flue gas temperature thresholds, standard acoustic wave sensing thresholds, and standard furnace pressure thresholds as sample data, and is a natural number less than , and use the sample data to obtain the sample data mean, denoted as the sample set; Use the temperature gradient matrix, acoustic wave time-frequency feature vector, and furnace pressure parameter sequence corresponding to the remaining several groups of leakage state determination results, as well as the standard flue gas temperature threshold, standard acoustic wave sensing threshold, and standard furnace pressure threshold, as the test set; According to the sample set and the test set, form a training sample set; Based on a convolutional neural network, construct a standard prediction model; And input the training sample set into the standard prediction model to train the standard prediction model, obtain the trained standard prediction model, and denote the trained standard prediction model as the leakage trend prediction model; Generate a boiler health prediction index under the current environmental factor conditions according to the leakage trend prediction model , the boiler health prediction index is as follows: ; wherein, represents the flue gas temperature data, represents the standard flue gas temperature threshold, represents the proportionality coefficient of the temperature influence degree, represents the acoustic wave sensing data, represents the standard acoustic wave sensing threshold, represents the furnace pressure data, represents the standard furnace pressure threshold.

9. The on-line monitoring system for leakage alarm of a heating surface tube screen according to claim 8, characterized in that The process of generating a leakage risk level according to the boiler health prediction index includes: Preset standard boiler health prediction index ; If the boiler health prediction index , there is no leakage risk in the water-cooled wall tubes, superheater tubes, reheater tubes, and economizer tubes in the boiler under the current environmental factor conditions; If the boiler health prediction index , there is a risk of leakage in the water-cooled wall tubes, superheater tubes, reheater tubes, and economizer tubes in the boiler under the current environmental factor conditions, and a leakage risk level is generated; it should be further noted that the leakage risk level includes a first-level leakage alarm and a second-level leakage alarm; Preset minimum boiler health prediction index , if , there is a risk of sand hole leakage in the water wall tubes, superheater tubes, reheater tubes and economizer tubes in the boiler, a first-level leakage alarm is generated, and a maintenance plan for the risk of sand hole leakage is executed; if , there is a risk of crack leakage in the water wall tubes, superheater tubes, reheater tubes and economizer tubes in the boiler, a second-level leakage alarm is generated, and a maintenance plan for the risk of crack leakage is executed.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and the computer program can be executed by a processor to implement a heat-absorbing surface tube screen leakage alarm online monitoring system according to any one of claims 1-9 above.

Images