Combustion stability diagnosis method based on multi-source information fusion and related device

Abstract

The invention discloses a combustion stability diagnosis method based on multi-source information fusion and a related device, and relates to the technical field of combustion, and the method comprises the steps: obtaining flame image parameter information and a hearth outlet pressure signal; determining flame image deep feature data according to the flame image parameter information; according to the hearth outlet pressure signal, sensor monitoring information data is determined; fusing the flame image deep feature data with the sensor monitoring information data to obtain fused data; performing clustering analysis on the fused data to obtain two clusters; calculating the information entropy of each cluster, and determining a stable cluster and an unstable cluster according to the size of the information entropy; the mass centers of the stable clusters and the unstable clusters are calculated, and combustion stability evaluation indexes are established; and determining a combustion stability result according to the combustion stability evaluation index. The method can effectively prevent combustion state abnormity, avoid safety accidents, and timely and accurately grasp combustion stability.

Description

Technical Field

[0001] The present application relates to the field of combustion technology, and in particular to a combustion stability diagnosis method and related devices based on multi-source information fusion. Background Art

[0002] Combustion is an important way to utilize fossil fuels. Establishing and maintaining a stable combustion state is crucial to improving combustion efficiency, reducing pollutant emissions, and ensuring the safe and environmentally friendly operation of the entire combustion system. However, due to the influence of various factors such as fuel quality and combustion method, it is difficult to maintain a stable combustion state for a long time.

[0003] In the actual combustion process, such as boiler combustion in thermal power generating units, it is difficult to obtain a clear and complete flame image in the furnace due to the high temperature and high dust environment in the furnace and the limitation of the installation position, which makes the quantitative analysis method of combustion stability based on flame image analysis face challenges in field application. Traditional visualization technology only provides simple monitoring functions, and the interpretation of image data is highly dependent on the subjective experience of the operator and lacks objective judgment criteria. Although combustion diagnostic technologies such as thermocouples and flame fault detectors have been developed based on the light energy and heat energy released by combustion, these technologies have limitations such as large errors, susceptibility to interference from the monitoring environment, single functions, frequent maintenance and short equipment life.

[0004] In order to effectively prevent abnormal combustion and avoid safety accidents, it is extremely important to grasp the combustion stability in a timely and accurate manner. Although flame images contain rich combustion state information, accurate analysis of pixel information faces challenges. The potential inducing factors of combustion instability are numerous and difficult to characterize, which makes accurate diagnosis of combustion state not easy. Summary of the invention

[0005] The purpose of this application is to provide a combustion stability diagnosis method and related devices based on multi-source information fusion, which can effectively prevent abnormal combustion conditions, avoid safety accidents, and grasp combustion stability in a timely and accurate manner.

[0006] To achieve the above objectives, this application provides the following solutions.

[0007] In a first aspect, the present application provides a combustion stability diagnosis method based on multi-source information fusion, and the combustion stability diagnosis method based on multi-source information fusion includes the following steps.

[0008] Obtain flame image parameter information and furnace outlet pressure signal; the flame image parameter information includes: flame black dragon, flame front and flame temperature.

[0009] According to the flame image parameter information, the flame image deep feature data is determined; the flame image deep feature data includes: black dragon length change rate, flame front area change rate, flame temperature gradient and temperature change rate.

[0010] According to the furnace outlet pressure signal, the sensor monitoring information data is determined; the sensor monitoring information data is the furnace negative pressure fluctuation amount.

[0011] The flame image deep feature data is fused with the sensor monitoring information data to obtain fused data.

[0012] Cluster analysis was performed on the fused data to obtain two clusters.

[0013] The information entropy of each cluster is calculated, and stable clusters and unstable clusters are determined according to the size of the information entropy.

[0014] The centroids of the stable cluster and the unstable cluster are calculated, and a combustion stability evaluation index is established.

[0015] According to the combustion stability evaluation index, a combustion stability result is determined.

[0016] Optionally, the determination of the black dragon length change rate specifically includes the following steps.

[0017] The flame black dragon is identified using image processing methods.

[0018] Calculate the maximum length of the flame black dragon in the image and get the length of the black dragon.

[0019] The peak wave number of the black dragon length curve within 1 minute is taken as the black dragon length change rate.

[0020] The black dragon length change rate is normalized to a dimensionless number of 0-100 in the form of a reciprocal.

[0021] Optionally, the determination of the flame front area change rate specifically includes the following steps.

[0022] The flame front is determined using image processing methods.

[0023] Calculate the flame front area based on the pixel points.

[0024] The peak wave number of the flame front area curve within 1 minute is taken as the flame front area change rate.

[0025] The flame front area change rate is normalized to a dimensionless number ranging from 0 to 100 in a reciprocal form.

[0026] Optionally, the determination of the flame temperature gradient specifically includes the following steps.

[0027] The temperature distribution field is calculated using image grayscale.

[0028] The measured temperature is compared with the temperature distribution field to calculate the temperature of each pixel in the image.

[0029] The maximum value of the temperature change of adjacent pixels in the image is taken as the flame temperature gradient.

[0030] The flame temperature gradient is normalized in reciprocal form to a dimensionless number ranging from 0 to 100.

[0031] Optionally, the determination of the temperature change rate specifically includes the following steps.

[0032] Using the image temperature calculation method, the maximum value of the temperature change of the same pixel in the image in the analysis sequence is calculated.

[0033] The peak wave number of the maximum value curve of the temperature change of the same pixel point in the image in the analysis sequence within 1 minute is taken as the temperature change rate.

[0034] The temperature change rate is normalized in reciprocal form to a dimensionless number ranging from 0 to 100.

[0035] Optionally, determining sensor monitoring information data according to the furnace outlet pressure signal specifically includes the following steps.

[0036] The maximum value of the difference between the peak and trough of the furnace outlet pressure signal curve within 1 minute is taken as the furnace negative pressure fluctuation.

[0037] The furnace negative pressure fluctuation amount is normalized in a reciprocal form to a dimensionless number of 0-100.

[0038] Optionally, the calculation formula of the combustion stability evaluation index is shown as follows.

[0039] .

[0040] in, It is a combustion stability evaluation index, a dimensionless number ranging from 0 to 100; is the weight coefficient of furnace negative pressure fluctuation; is the weight coefficient of the change rate of the flame's black dragon length; is the weight coefficient of the flame front area change rate; is the weight coefficient of flame temperature gradient; is the weight coefficient of temperature change rate; ; is the fluctuation of the negative pressure in the furnace, normalized to a dimensionless number of 0-100 in the form of an inverse; is the change rate of the flame's black dragon length, normalized to a dimensionless number of 0-100 in the form of an inverse; is the rate of change of the flame front area, normalized to a dimensionless number between 0 and 100 in the form of an inverse; is the flame temperature gradient, the maximum value of the temperature change of adjacent pixels in the image, representing the temperature change in the image plane space, and is normalized to a dimensionless number of 0-100 in the form of an inverse; It is the temperature change rate, the maximum value of the temperature change of the same pixel in the image in the analysis sequence, which represents the actual temperature change of the image and is normalized to a dimensionless number of 0-100 in the form of an inverse.

[0041] Optionally, determining the combustion stability result according to the combustion stability evaluation index specifically includes the following steps.

[0042] when At this time, the low-load boiler is absolutely stable and belongs to the normal combustion state.

[0043] when At this time, the combustion of low-load boilers is relatively stable and there is no risk of flameout.

[0044] when When the combustion stability of a low-load boiler is weak, it should be adjusted quickly to prevent flameout.

[0045] when When the load is low, the combustion of the boiler is extremely unstable and may go out at any time.

[0046] In a second aspect, the present application provides a combustion stability diagnosis device based on multi-source information fusion, and the combustion stability diagnosis device based on multi-source information fusion includes: a flame detector, a furnace pressure sensor, a data processing unit and a result output unit.

[0047] The flame detector comprises: a plurality of temperature measuring thermocouples and a plurality of flame detector cameras; the flame detector is installed in the area corresponding to the pulverized coal burner of the low-load boiler and is used for real-time monitoring of the flame state.

[0048] The furnace pressure sensor is installed at the furnace outlet and is used to measure the furnace outlet pressure signal.

[0049] The data processing unit is connected to the flame detector and the furnace pressure sensor respectively, and is used to determine the combustion stability result based on any of the above-mentioned combustion stability diagnosis methods based on multi-source information fusion.

[0050] The result output unit is connected to the data processing unit and is used to output the combustion stability result.

[0051] Optionally, the flame detector includes: 4 temperature measurement thermocouples and 4 flame detector cameras.

[0052] Among them, the first flame detector camera is installed at the root of the flame; the second flame detector camera and the third flame detector camera are installed at the middle section of the flame; and the fourth flame detector camera is installed at the tip of the flame.

[0053] Four temperature measuring thermocouples are arranged along the furnace direction; the first temperature measuring thermocouple and the second temperature measuring thermocouple are group A; the third temperature measuring thermocouple and the fourth temperature measuring thermocouple are group B; the group A and the group B are diagonally distributed.

[0054] According to the specific embodiments provided in this application, this application discloses the following technical effects.

[0055] The present application provides a combustion stability diagnosis method and related device based on multi-source information fusion, which determines the deep feature data of the flame image according to the acquired flame image parameter information; determines the sensor monitoring information data according to the acquired furnace outlet pressure signal; fuses the deep feature data of the flame image with the sensor monitoring information data to obtain the fused data; performs cluster analysis on the fused data to obtain two clusters; calculates the information entropy of each cluster, and determines the stable cluster and the unstable cluster according to the size of the information entropy; calculates the centroid of the stable cluster and the unstable cluster, and establishes a combustion stability evaluation index; according to the combustion stability evaluation index, the combustion stability result can be determined. The present application can more comprehensively and accurately evaluate the combustion stability by fusing the deep feature data of the flame image and the monitoring information from various sensors. At the same time, the present application can quantitatively characterize the combustion stability, provide an important guiding basis for the adjustment of the combustion state, and is of great significance for studying the combustion mechanism and exploring the laws of dynamic changes in combustion, and helps to improve the overall performance and reliability of the combustion system. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0057] Figure 1 This is an application environment diagram of a combustion stability diagnosis method based on multi-source information fusion in one embodiment of the present application.

[0058] Figure 2 A flowchart of a combustion stability diagnosis method based on multi-source information fusion is provided in one embodiment of the present application.

[0059] Figure 3 A schematic diagram of the structure of a combustion stability diagnosis device based on multi-source information fusion provided in one embodiment of the present application.

[0060] The reference numerals are as follows.

[0061] 1- low load boiler pulverized coal burner; 2- first temperature measuring thermocouple; 3- second temperature measuring thermocouple; 4- first flame detector camera; 5- second flame detector camera; 6- third flame detector camera; 7- fourth flame detector camera; 8- third temperature measuring thermocouple; 9- fourth temperature measuring thermocouple; 10- furnace pressure sensor; 11- data processing unit; 12- result output unit. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0063] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0064] This application is dedicated to developing a combustion stability diagnosis solution that integrates key technologies such as image acquisition, image feature extraction, and image feature analysis. This solution aims to solve the problems of poor interpretability of deep features of flame images and difficulty in quantifying combustion stability, and provide an important theoretical basis and technical foundation for optimizing the operation of combustion systems. For example, a combustion stability diagnosis method and related device based on multi-source information fusion, which can more comprehensively and accurately evaluate combustion stability by fusing flame images and monitoring information from various sensors (such as temperature sensors, pressure sensors, etc.). The technologies involved in this application include using digital image processing technology to analyze flame images, extract features such as black dragon length, flame front area, flame temperature gradient, etc.; fusing and analyzing multi-source information through relevant algorithms; and establishing a quantitative evaluation model for combustion stability with the help of neural networks, machine learning, etc.

[0065] The combustion stability diagnosis method based on multi-source information fusion provided in the embodiment of the present application can be applied to Figure 1In the application environment shown, the terminal 102 communicates with the server 104 through a network. The data storage system can store data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the acquired flame image parameter information and the furnace outlet pressure signal to the server 104, wherein the flame image parameter information includes: flame black dragon, flame front and flame temperature; after the server 104 receives the flame image parameter information and the furnace outlet pressure signal, the server 104 determines the flame image deep feature data according to the flame image parameter information and the furnace outlet pressure signal; the flame image deep feature data includes: black dragon length change rate, flame front area change rate, flame temperature gradient and temperature change rate; according to the furnace outlet pressure signal, the sensor monitoring information data is determined; the sensor monitoring information data is the furnace negative pressure fluctuation amount; the flame image deep feature data is fused with the sensor monitoring information data to obtain fused data; cluster analysis is performed on the fused data to obtain two clusters; the information entropy of each cluster is calculated, and a stable cluster and an unstable cluster are determined according to the size of the information entropy; the centroid of the stable cluster and the unstable cluster is calculated to establish a combustion stability evaluation index; according to the combustion stability evaluation index, the combustion stability result is determined. The server 104 may feed back the obtained combustion stability result to the terminal 102. In addition, in some embodiments, the combustion stability diagnosis method based on multi-source information fusion may also be implemented by the server 104 or the terminal 102 alone, such as the terminal 102 may directly perform combustion stability diagnosis on the acquired flame image parameter information and furnace outlet pressure signal, or the server 104 may obtain the flame image parameter information and furnace outlet pressure signal from the data storage system, and perform combustion stability diagnosis on the flame image parameter information and furnace outlet pressure signal.

[0066] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, and tablet computers. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or may be a cloud server.

[0067] In an exemplary embodiment, Figure 2 As shown, a combustion stability diagnosis method based on multi-source information fusion is provided. The method is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used for explanation, and the steps include the following steps S1 to S8.

[0068] S1: Obtain flame image parameter information and furnace outlet pressure signal; the flame image parameter information includes: flame black dragon, flame front and flame temperature.

[0069] S2: Determine the flame image deep feature data according to the flame image parameter information; the flame image deep feature data includes: black dragon length change rate, flame front area change rate, flame temperature gradient and temperature change rate.

[0070] S3: Determine sensor monitoring information data according to the furnace outlet pressure signal; the sensor monitoring information data is the furnace negative pressure fluctuation amount.

[0071] S4: Fusing the flame image deep feature data with the sensor monitoring information data to obtain fused data.

[0072] S5: Perform cluster analysis on the fused data to obtain two clusters.

[0073] S6: Calculate the information entropy of each cluster, and determine the stable cluster and the unstable cluster according to the size of the information entropy.

[0074] S7: Calculate the centroids of the stable cluster and the unstable cluster, and establish a combustion stability evaluation index.

[0075] S8: Determine the combustion stability result according to the combustion stability evaluation index.

[0076] By implementing the above steps S1 to S8, the present application can more comprehensively and accurately evaluate the combustion stability by fusing the deep feature data of the flame image and the monitoring information from various sensors. At the same time, the present application can quantitatively characterize the combustion stability and provide an important guiding basis for the adjustment of the combustion state. It is of great significance for studying the combustion mechanism and exploring the laws of dynamic changes in combustion, and helps to improve the overall performance and reliability of the combustion system.

[0077] It should be noted that the combustion stability diagnosis method based on multi-source information fusion can be applied to complex operating conditions including low-load operation conditions of pulverized coal boilers in power plants.

[0078] In an exemplary embodiment, the determination of the black dragon length change rate specifically includes the following steps.

[0079] A1: By analyzing each frame of the flame video, the flame black dragon is determined using image processing methods.

[0080] A2: Calculate the maximum length of the flame black dragon in the image and get the length of the black dragon.

[0081] A3: The peak wave number of the black dragon length curve within 1 minute is taken as the black dragon length change rate.

[0082] A4: The black dragon length change rate is normalized to a dimensionless number of 0-100 in the form of an inverse.

[0083] "Flame black dragon" refers to the unburned area formed by the mixed airflow of primary air and coal powder entering the furnace. Because the ignition temperature of the coal powder has not yet reached the ignition point, the mixed airflow diffuses and moves around along the incident direction. From the collected flame image, this area looks like a tongue-shaped black dragon, so this unburned area is also called the black dragon area.

[0084] In step A1, the specific steps of determining the flame black dragon by using the image processing method are as follows.

[0085] A11. Image capture: Use a high-speed camera or other suitable imaging equipment to capture flame images during the combustion process from a suitable angle and position, ensuring that the images clearly reflect the shape and characteristics of the flame and have sufficient resolution and frame rate for subsequent analysis.

[0086] A12. Elimination of interference area: Due to camera installation, there may be low-brightness obstructions at the four corners of the flame combustion image area. By eliminating the pixels with low gray values ​​in the image, this interference is eliminated, thereby determining the effective area for calculating the length of the primary air and coal powder mixed airflow entering the furnace to the pre-combustion stage.

[0087] A13. Black dragon characteristics analysis and edge detection area determination: According to the incident conditions of coal powder in the flame image, the area that has the main influence on the length of the black dragon along the incident angle of the coal powder is determined as the edge detection area. For example, when the primary air and coal powder mixed airflow are sprayed in the horizontal direction, the middle area plays a major role in the length of the black dragon. In the determined edge detection area, n straight lines for detecting the length of the black dragon are determined along the incident direction of the mixed airflow (generally, i straight lines are selected along the upper and lower directions of the coal powder injection, for a total of 2i+1 straight lines), which are used for the subsequent calculation of the edge points of the black dragon.

[0088] A14. Perform edge detection to identify feature areas, thereby determining the edge and direction of the black dragon area extraction. Apply a differential algorithm to search for the pixel point with the maximum grayscale gradient change in the detection area, that is, the black dragon edge point, and determine the length of the black dragon on the flame image through the edge point coordinates.

[0089] A15. Identify the black dragon edge points on all selected straight lines, define the distance from the flame image boundary to the black dragon edge point of each straight line as L_i, and then calculate the average of the measured 2i+1 black dragon lengths as the black dragon length L of the flame image.

[0090] In an exemplary embodiment, the determination of the flame front area change rate specifically includes the following steps.

[0091] B1: By analyzing each frame of the flame video, the flame front is determined using image processing methods.

[0092] B2: Calculate the flame front area based on the pixel points.

[0093] B3: The peak wave number of the flame front area curve within 1 minute is taken as the flame front area change rate.

[0094] B4: The flame front area change rate is normalized in the form of a reciprocal to a dimensionless number of 0-100.

[0095] The flame front refers to a regional interface with specific combustion state and heat transfer characteristics formed during the dynamic change of the mixed airflow after the primary air and coal powder flow enters the furnace and interacts with the flame heat flow in the furnace. In this area, the combustion state, temperature, heat transfer mode, etc. change significantly. For example, the transition area from the mixed airflow absorbing heat and not burning to the beginning of combustion, as well as the position of the flame front during the combustion process, are all related to the flame front.

[0096] In step B1, the specific steps of determining the flame front using the image processing method are similar to the steps of identifying the "flame black dragon".

[0097] B11. Use a CCD camera to capture flame images in the boiler furnace. The images taken by the camera are initially analog images and need to be converted into digital images through digital processing technology.

[0098] B12. Image grayscale processing: The collected RGB image consisting of three channels of red (R), green (G), and blue (B) is grayscaled using the formula (Gray=0.299R+0.587G+0.114B) to turn it into a single-channel image. This can reduce the amount of calculation and will not affect the study of the boiler combustion stability status.

[0099] B13. Image filtering: Use median filtering to preprocess the grayscale image to remove the noise in the flame video while retaining the edge information in the image, avoiding the problem of blurred edge areas caused by mean filtering, insufficient noise suppression by Gaussian filtering, and poor text filtering effect.

[0100] B14. Combustion zone state detection and flame front recognition: Analyze the characteristics of flame images. Since the flames produced by coal powder combustion have irregular shapes and are similar in local and overall forms (higher brightness), for example, the process from the unburned zone to the burned zone is similar, but there are differences in the ignition position and contour length. This characteristic conforms to the fractal characteristics, so the fractal theory is used to segment the flame image. Select a window of a certain size (such as 8×8) (the window moving step is a specific value), calculate the K value of the image in the window (the K value is related to the rate of change of the grayscale surface area of ​​the image with the scale change, and its size can reflect the roughness of the grayscale surface of the image. The K value is larger at the edges of different areas of the flame image, and smaller in other places) and the mean of the grayscale value. There is a corresponding formula for calculating the mean of the grayscale value, and the characteristic data is obtained by calculation. The K-means algorithm (clustering category selection is 3, corresponding to the three areas in the flame image) is used to perform cluster analysis on the image. The grayscale mean of a specific category in the clustering result is used as the threshold for image segmentation. The image is segmented to obtain the combustion area of ​​the coal powder, thereby determining the approximate area where the flame front is located (i.e., the boundary area between the combustion area and the unburned area), and returning the corresponding image pixel coordinates.

[0101] In an exemplary embodiment, the determination of the flame temperature gradient specifically includes the following steps.

[0102] C1: By analyzing each frame of the flame video, the temperature distribution field is calculated using the image grayscale.

[0103] Specifically, the flame detection camera is calibrated in a space with known temperature distribution (generally, the temperature field is measured in advance for several fixed load conditions, and then calibrated: a standard radiation source with known temperature, such as a high-precision approximate black body, is used to collect its thermal image at different temperatures using an imaging system to obtain the corresponding data of the image grayscale value and the known temperature, and then the relationship between the two is established through data fitting). The flame image is captured using the calibrated flame detection camera, and the radiation intensity value corresponding to the grayscale value of each pixel in the flame image is calculated based on the grayscale information of the image and the fitted grayscale value and radiation intensity relationship, thereby calculating the temperature distribution.

[0104] C2: Compare and calibrate the measured temperature with the temperature distribution field to calculate the temperature of each pixel in the image.

[0105] C3: The maximum value of the temperature change of adjacent pixels in the image is taken as the flame temperature gradient.

[0106] C4: The flame temperature gradient is normalized in inverse form to a dimensionless number between 0 and 100.

[0107] In an exemplary embodiment, the determination of the temperature change rate specifically includes the following steps.

[0108] D1: By analyzing each frame of the flame video and using the image temperature calculation method, the maximum temperature change of the same pixel in the image in the analysis sequence is calculated.

[0109] Specifically, the temperature field distribution obtained by the grayscale calculation of the above image is used to obtain the temperature field of a group of images, and then the temperature values ​​of the same pixel position at different times are obtained, and then the maximum value of the temperature change of the same pixel point in the image in the analysis sequence is obtained.

[0110] D2: The peak wave number of the maximum value curve of the temperature change of the same pixel point in the image in the analysis sequence within 1 minute is taken as the temperature change rate.

[0111] D3: The temperature change rate is normalized to a dimensionless number of 0-100 in the form of a reciprocal.

[0112] In an exemplary embodiment, step S3 specifically includes the following steps.

[0113] S31: The maximum value of the difference between the peak and the trough of the furnace outlet pressure signal curve within 1 minute is taken as the furnace negative pressure fluctuation amount.

[0114] S32: normalizing the furnace negative pressure fluctuation amount in an inverse form into a dimensionless number of 0-100.

[0115] It should be noted that the furnace negative pressure fluctuation should use the furnace outlet pressure signal, so the furnace pressure sensor should be installed at the furnace outlet. Take the maximum value of the difference between the peak and trough of the pressure curve within 1 minute as the furnace negative pressure fluctuation (p). First, the pressure signal must be accurately obtained through the pressure sensor installed at the furnace outlet. The obtained furnace negative pressure fluctuation (p) needs to be normalized in the inverse form of a dimensionless number of 0-100. This step is to adjust the numerical range of the pressure signal to match other indicators used to calculate combustion stability ( ) parameters (such as the rate of change of the flame's black dragon length, the rate of change of the flame front area, etc.) of the same magnitude for weighted calculation. Pressure sensors need to be regularly maintained and calibrated to ensure the accuracy of their measurements. Maintenance includes cleaning the sensor surface, checking the connection lines, etc. Calibration should be carried out in accordance with relevant standards and specifications, and the sensor should be calibrated using a standard pressure source to ensure the reliability of the measurement data.

[0116] In an exemplary embodiment, the calculation formula of the combustion stability evaluation index is shown as follows.

[0117] .

[0118] in, It is a combustion stability evaluation index, a dimensionless number ranging from 0 to 100; is the weight coefficient of furnace negative pressure fluctuation; is the weight coefficient of the change rate of the flame's black dragon length; is the weight coefficient of the flame front area change rate; is the weight coefficient of flame temperature gradient; is the weight coefficient of temperature change rate; ; is the fluctuation of the negative pressure in the furnace, normalized to a dimensionless number of 0-100 in the form of an inverse; is the change rate of the flame's black dragon length, normalized to a dimensionless number of 0-100 in the form of an inverse; is the rate of change of the flame front area, normalized to a dimensionless number between 0 and 100 in the form of an inverse; is the flame temperature gradient, the maximum value of the temperature change of adjacent pixels in the image, representing the temperature change in the image plane space, and is normalized to a dimensionless number of 0-100 in the form of an inverse; is the temperature change rate, the maximum value of the temperature change of the same pixel in the image in the analysis sequence, which represents the actual temperature change of the image and is normalized to a dimensionless number of 0-100 in the form of a reciprocal. The value range of each weight coefficient is shown in Table 1.

[0119] Table 1 Weight coefficient value range

[0120] According to the calculation formula of combustion stability evaluation index, the The corresponding levels are described below.

[0121] Level S1: When At this time, the low-load boiler is absolutely stable and belongs to the normal combustion state.

[0122] Level S2: When At this time, the combustion of low-load boilers is relatively stable and there is no risk of flameout.

[0123] Level S3: When When the combustion stability of a low-load boiler is weak, it should be adjusted quickly to prevent flameout.

[0124] Level S4: When When the load is low, the combustion of the boiler is extremely unstable and may go out at any time.

[0125] The comprehensive evaluation of low-load boiler combustion stability is shown in Table 2.

[0126] Table 2 Comprehensive evaluation of combustion stability of low-load boilers

[0127] The present application also provides an application scenario, which applies the above-mentioned combustion stability diagnosis method based on multi-source information fusion. Specifically: the combustion stability diagnosis method based on multi-source information fusion provided in this embodiment can be applied in the combustion stability diagnosis scenario. The combustion stability diagnosis scenario includes: data acquisition link, data fusion link, cluster analysis link and combustion stability determination link; first, flame image parameter information and furnace outlet pressure signal are obtained; the flame image parameter information includes: flame black dragon, flame front and flame temperature; according to the flame image parameter information, the flame image deep feature data is determined; the flame image deep feature data includes: black dragon length change rate, flame front area change rate, flame temperature gradient and temperature change rate; according to the furnace outlet pressure signal, the sensor monitoring information data is determined; the sensor monitoring information data is the furnace negative pressure fluctuation amount; secondly, the flame image deep feature data is fused with the sensor monitoring information data to obtain fused data; then, the fused data is clustered and analyzed to obtain two clusters; the information entropy of each cluster is calculated, and the stable cluster and the unstable cluster are determined according to the size of the information entropy; finally, the centroid of the stable cluster and the unstable cluster is calculated to establish a combustion stability evaluation index; according to the combustion stability evaluation index, the combustion stability result can be determined.

[0128] In another exemplary embodiment of the present application, Figure 3 As shown, a combustion stability diagnosis device based on multi-source information fusion is provided, and the combustion stability diagnosis device based on multi-source information fusion includes: a flame detector, a furnace pressure sensor 10, a data processing unit 11 and a result output unit 12.

[0129] The flame detector includes: a plurality of temperature measuring thermocouples and a plurality of flame detector cameras; the flame detector is installed in the area corresponding to the low-load boiler pulverized coal burner 1, and is used to monitor the state of the flame in real time to obtain flame image parameter information; the flame image parameter information includes: flame black dragon, flame front and flame temperature; these parameters are crucial for a comprehensive understanding of the stability and combustion state of the flame, so the installation position of the flame detector camera should be able to accurately obtain image information related to these parameters.

[0130] The furnace pressure sensor 10 is installed at the furnace outlet to measure the furnace outlet pressure signal and provide the key data of furnace negative pressure fluctuation for combustion stability determination.

[0131] The data processing unit 11 is connected to the flame detector and the furnace pressure sensor 10, respectively, and is used to determine the combustion stability result based on any of the above-mentioned combustion stability diagnosis methods based on multi-source information fusion. The combustion stability diagnosis method based on multi-source information fusion comprehensively considers parameters such as furnace negative pressure fluctuation, flame black dragon length change rate, flame front area change rate, flame temperature gradient, temperature change rate, etc., and calculates the combustion stability discrimination index through a specific weight coefficient.

[0132] The result output unit 12 is connected to the data processing unit 11 and is used to output the combustion stability result. That is, the result output unit 12 can output the combustion stability result in real time to intuitively show whether the current combustion state is absolutely stable ( ), relatively stable ( ), poor stability ( ) or extremely unstable ( ).

[0133] The device meets environmental protection standards and is easy to install and operate, which can bring huge economic benefits to the enterprise.

[0134] It should be noted that each coal burner, oil burner, and gas burner should be equipped with a flame detection device. This is to accurately monitor the flame status of each burner. Because the operating conditions of different burners may vary, setting up a separate detection device can obtain flame information more specifically.

[0135] The flame detector camera should be installed in place and debugged normally to ensure that the boiler can correctly detect the flame under full load and all applicable coal conditions; similarly, the thermocouple should be installed in place and debugged normally to ensure that the boiler can correctly detect the temperature under full load and all applicable coal conditions. Figure 3 The installation measurement point locations are given. Specifically, the flame detector includes: 4 temperature measurement thermocouples and 4 flame detector cameras; The first flame detector camera 4 is installed at the root of the flame; the second flame detector camera 5 and the third flame detector camera 6 are installed at the middle section of the flame; the fourth flame detector camera 7 is installed at the tip of the flame; Four temperature measuring thermocouples are arranged along the furnace direction; the first temperature measuring thermocouple 2 and the second temperature measuring thermocouple 3 are group A; the third temperature measuring thermocouple 8 and the fourth temperature measuring thermocouple 9 are group B; the group A and the group B are diagonally distributed.

[0136] In an exemplary embodiment, the workflow of the data processing unit 11 is as follows.

[0137] First, flame image parameter information and furnace outlet pressure signal are obtained; the flame image parameter information includes: flame black dragon, flame front and flame temperature; according to the flame image parameter information, the flame image deep feature data is determined; the flame image deep feature data includes: black dragon length change rate, flame front area change rate, flame temperature gradient and temperature change rate; according to the furnace outlet pressure signal, the sensor monitoring information data is determined; the sensor monitoring information data is the furnace negative pressure fluctuation amount, and then, the flame image deep feature data is fused with the sensor monitoring information data through the concatenate function of numpy. The K-Means algorithm is used to perform cluster analysis on the fused data and divide the data into two clusters. The information entropy of each cluster is calculated, and the stable cluster and the unstable cluster are defined according to the size of the information entropy. By calculating the centroids of the stable cluster and the unstable cluster, the combustion stability evaluation index is established (here the Euclidean distance between the two centroids is simply used as an example). Finally, the combustion stability is determined according to the combustion stability evaluation index. In practical applications, it is necessary to select appropriate feature extraction methods, clustering algorithms, information entropy calculation methods, and methods for constructing combustion stability evaluation indicators based on specific image data features. At the same time, it is also necessary to preprocess and analyze the data more carefully to improve the accuracy and reliability of the model. In practical applications, it is necessary to use real flame image deep features and sensor monitoring information data, and make corresponding adjustments and optimizations based on the characteristics of the data. Moreover, the implementation of complex feature extraction networks such as anti-noise autoencoders requires the use of deep learning frameworks such as TensorFlow or PyTorch. In addition, the part of the correlation coefficient correction delay time is not specifically reflected in the code, which requires calculating the correlation coefficient between the actual sensor data and flame image features, and correcting the delay time based on the correlation coefficient. The specific implementation method will depend on the characteristics and processing methods of the data. The complete code implementation of the combustion stability diagnosis method based on multi-source information fusion is a relatively complex task, which comprehensively considers multiple aspects such as data processing, feature extraction, model training and evaluation, and needs to be customized according to the specific application scenario and data. Development and optimization. The simplified case code is given as follows.

[0138] import numpy as np import pandas as pd from sklearn.cluster import KMeans from sklearn.metrics import entropy # Assume we have deep feature data of flame images (example data) flame_image_features = np.random.rand(100, 5) # 100 samples, each with 5 features # Sensor monitoring information data (sample data) sensor_monitoring_info = np.random.rand(100, 4) # 100 samples, each with 4 sensor monitoring information # Step 1: Data preprocessing # Here you can perform more detailed data preprocessing according to the actual situation, such as normalization, standardization, etc. # Step 2: Feature Fusion image_sensor_features = np.concatenate((flame_image_features, sensor_monitoring_info), axis=1) # Step 3: Cluster analysis kmeans = KMeans(n_clusters=2) # Divided into 2 clusters kmeans.fit(image_sensor_features) cluster_labels = kmeans.labels_ # Step 4: Calculate the information entropy of each cluster cluster_0_features = image_sensor_features[cluster_labels == 0] cluster_1_features = image_sensor_features[cluster_labels == 1] entropy_0 = entropy(np.bincount(cluster_0_features[:, 0])) # Calculate information entropy using the first column of features as an example entropy_1 = entropy(np.bincount(cluster_1_features[:, 0])) # Calculate information entropy using the first column of features as an example # Step 5: Define stable and unstable clusters stable_cluster = 0 if entropy_0 <entropy_1 else 1 unstable_cluster = 1 - stable_cluster # Step 6: Calculate the centroids of stable and unstable clusters stable_cluster_center = np.mean(image_sensor_features[cluster_labels== stable_cluster], axis=0) unstable_cluster_center = np.mean(image_sensor_features[cluster_labels == unstable_cluster], axis=0) # Step 7: Establish combustion stability evaluation index (example) # Here you can build a combustion stability evaluation index based on the center of mass or other related features stability_indicator = np.linalg.norm(stable_cluster_center -unstable_cluster_center) # Step 8: Combustion stability analysis based on evaluation indicators if stability_indicator>some_threshold: # Set the threshold according to the specific situation print("Combustion is unstable") else: print("Combustion is stable").

[0139] In summary, the present application can more comprehensively and accurately evaluate combustion stability by fusing flame images and monitoring information from various sensors (such as temperature sensors, pressure sensors, etc.).

[0140] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0141] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A combustion stability diagnosis method based on multi-source information fusion, characterized in that: The combustion stability diagnosis method based on multi-source information fusion includes: Obtaining flame image parameter information and furnace outlet pressure signal; the flame image parameter information includes: flame black dragon, flame front and flame temperature; Determine the flame image deep feature data according to the flame image parameter information; the flame image deep feature data includes: the black dragon length change rate, the flame front area change rate, the flame temperature gradient and the temperature change rate; Determine sensor monitoring information data according to the furnace outlet pressure signal; the sensor monitoring information data is the furnace negative pressure fluctuation amount; Fusing the flame image deep feature data with the sensor monitoring information data to obtain fused data; Performing cluster analysis on the fused data, obtaining two clusters; Calculate the information entropy of each cluster, and determine the stable cluster and unstable cluster according to the size of the information entropy; Calculating the centroids of the stable cluster and the unstable cluster, and establishing a combustion stability evaluation index; According to the combustion stability evaluation index, a combustion stability result is determined.

2. The combustion stability diagnosis method based on multi-source information fusion according to claim 1 is characterized in that: The determination of the black dragon length change rate specifically includes: The flame black dragon is identified by using image processing method; Calculate the maximum length of the flame black dragon in the image and get the length of the black dragon; The peak wave number of the black dragon length curve within 1 minute is taken as the black dragon length change rate; The black dragon length change rate is normalized to a dimensionless number of 0-100 in the form of a reciprocal.

3. The combustion stability diagnosis method based on multi-source information fusion according to claim 1 is characterized in that: The determination of the flame front area change rate specifically includes: The flame front is determined by using image processing methods; Calculate the flame front area based on the pixel points; The peak wave number of the flame front area curve within 1 minute is taken as the flame front area change rate; The flame front area change rate is normalized to a dimensionless number ranging from 0 to 100 in a reciprocal form.

4. The combustion stability diagnosis method based on multi-source information fusion according to claim 1 is characterized in that: The determination of the flame temperature gradient specifically includes: Calculate the temperature distribution field using image grayscale; The measured temperature is compared with the temperature distribution field to calculate the temperature of each pixel in the image; The maximum value of the temperature change of adjacent pixels in the image is taken as the flame temperature gradient; The flame temperature gradient is normalized in reciprocal form to a dimensionless number ranging from 0 to 100.

5. The combustion stability diagnosis method based on multi-source information fusion according to claim 1 is characterized in that: The determination of the temperature change rate specifically includes: Using the image temperature calculation method, the maximum value of the temperature change of the same pixel in the image in the analysis sequence is calculated; The peak wave number of the maximum value curve of the temperature change of the same pixel point in the image in the analysis sequence within 1 minute is taken as the temperature change rate; The temperature change rate is normalized in reciprocal form to a dimensionless number ranging from 0 to 100.

6. The combustion stability diagnosis method based on multi-source information fusion according to claim 1 is characterized in that: According to the furnace outlet pressure signal, the sensor monitoring information data is determined, specifically including: The maximum value of the difference between the peak and trough of the furnace outlet pressure signal curve within 1 minute is taken as the furnace negative pressure fluctuation amount; The furnace negative pressure fluctuation amount is normalized in a reciprocal form to a dimensionless number of 0-100.

7. The combustion stability diagnosis method based on multi-source information fusion according to claim 1 is characterized in that: The calculation formula of the combustion stability evaluation index is: ； in, It is a combustion stability evaluation index, a dimensionless number ranging from 0 to 100; is the weight coefficient of furnace negative pressure fluctuation; is the weight coefficient of the change rate of the flame's black dragon length; is the weight coefficient of the flame front area change rate; is the weight coefficient of flame temperature gradient; is the weight coefficient of temperature change rate; ; is the fluctuation of the negative pressure in the furnace, normalized to a dimensionless number of 0-100 in the form of an inverse; is the change rate of the flame's black dragon length, normalized to a dimensionless number between 0 and 100 in inverse form; is the rate of change of the flame front area, normalized to a dimensionless number between 0 and 100 in the form of an inverse; is the flame temperature gradient, the maximum value of the temperature change of adjacent pixels in the image, representing the temperature change in the image plane space, and is normalized to a dimensionless number of 0-100 in the form of an inverse; It is the temperature change rate, the maximum value of the temperature change of the same pixel in the image in the analysis sequence, which represents the actual temperature change of the image and is normalized to a dimensionless number of 0-100 in the form of an inverse.

8. The combustion stability diagnosis method based on multi-source information fusion according to claim 7 is characterized in that: According to the combustion stability evaluation index, the combustion stability result is determined, specifically including: when When the low-load boiler is absolutely stable, it is in a normal combustion state; when When the load is low, the combustion of the boiler is relatively stable, without the risk of flameout; when When the combustion stability of low-load boilers is weak, they should be adjusted quickly to prevent flameout; when When the load is low, the combustion of the boiler is extremely unstable and may go out at any time.

9. A combustion stability diagnosis device based on multi-source information fusion, characterized in that: The combustion stability diagnosis device based on multi-source information fusion includes: a flame detector, a furnace pressure sensor, a data processing unit and a result output unit; The flame detector includes: a plurality of temperature measuring thermocouples and a plurality of flame detector cameras; the flame detector is installed in the area corresponding to the pulverized coal burner of the low-load boiler and is used to monitor the state of the flame in real time; The furnace pressure sensor is installed at the furnace outlet and is used to measure the furnace outlet pressure signal; The data processing unit is connected to the flame detector and the furnace pressure sensor respectively, and is used to determine the combustion stability result based on the combustion stability diagnosis method based on multi-source information fusion according to any one of claims 1 to 8; The result output unit is connected to the data processing unit and is used to output the combustion stability result.

10. The combustion stability diagnosis device based on multi-source information fusion according to claim 9, characterized in that: The flame detector includes: 4 temperature measuring thermocouples and 4 flame detector cameras; The first flame detector camera is installed at the flame root; the second flame detector camera and the third flame detector camera are installed at the flame middle section; the fourth flame detector camera is installed at the flame tip; Four temperature measuring thermocouples are arranged along the furnace direction; the first temperature measuring thermocouple and the second temperature measuring thermocouple are group A; the third temperature measuring thermocouple and the fourth temperature measuring thermocouple are group B; the group A and the group B are diagonally distributed.

Images