All-digital flame monitoring and feedback system and method
By performing characteristic analysis and identification of flames, the problem of high flame anomaly prediction cost in the prior art is solved, and efficient and accurate flame monitoring and abnormal handling are achieved.
Patent Information
- Application Number
- CN202510475077.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-16
- Publication Date
- 2025-06-17
AI Technical Summary
When the prior art determines whether there is an abnormality in flame combustion through flame monitoring data, a large amount of data is required to train the neural network model, and the model needs to be repeatedly debugged to ensure the accuracy of the model's prediction of flame anomalies, resulting in high prediction costs.
By collecting flame image data, flame temperature field distribution data and flame light intensity data, the flame characteristic analysis is performed, including flame profile recognition, center of mass movement feature calculation and pulsation frequency analysis, to generate flame recognition feature data, and then identify flame abnormality types and perform corresponding processing schemes.
The flame anomaly type prediction step is simplified, which reduces the possibility that the prediction results are biased towards local optimal solutions, improves the accuracy and efficiency of flame monitoring, and reduces the prediction cost.
Smart Images
Figure CN120160164A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of digital flame monitoring, and particularly to an all-digital flame monitoring and feedback system and method. Background Art
[0002] The all-digital flame monitoring technology is an important direction for the development of industrial automation and intelligence. It realizes the real-time monitoring, analysis, and control of the flame state through technologies such as digital sensors, image processing, artificial intelligence (AI), and the Internet of Things (IoT).
[0003] However, when the existing technology determines whether there is an abnormality in flame combustion through flame monitoring data, it usually requires a large amount of data to train a neural network model and needs to repeatedly debug the model to ensure the accuracy of the model's prediction of flame abnormalities, which greatly increases the cost of prediction. Summary of the Invention
[0004] The purpose of the present invention is to provide an all-digital flame monitoring and feedback system and method to solve the above deficiencies in the existing technology.
[0005] To achieve the above purpose, the present invention provides the following technical solutions:
[0006] An all-digital flame monitoring and feedback method, comprising the following steps:
[0007] S1. Collect flame image data, flame temperature field distribution data, and flame light intensity data;
[0008] S2. Perform flame contour recognition and analysis on the flame image data to generate flame feature data;
[0009] S3. Calculate and process the variance of the flame centroid movement based on the flame feature data to generate flame centroid movement feature data;
[0010] S4. Calculate and analyze the flame pulsation frequency based on the flame light intensity data to generate flame pulsation frequency data;
[0011] S5. Collect and combine the flame centroid movement feature data, flame pulsation frequency data, and flame temperature field distribution data to generate flame recognition feature data;
[0012] S6. Analyze and process the flame abnormality type corresponding to the flame recognition feature data based on the flame recognition feature data and the flame abnormality type recognition feature data to generate flame abnormality type analysis data;
[0013] S7. Search for the processing scheme corresponding to the flame abnormality type based on the flame abnormality type analysis data and the flame abnormality type processing scheme data to generate flame abnormality processing scheme return data;
[0014] S8. Collect and combine the flame recognition feature data, the flame anomaly type analysis data, and the flame anomaly handling solution return data to construct flame monitoring management data, and perform flame monitoring feedback operations based on the flame monitoring management data.
[0015] Further, the S1 includes the following steps:
[0016] S11. Collect flame image data A through a CCD visible light camera, where the CCD visible light camera can be deployed at the combustion chamber observation port to capture changes in flame color and shape;
[0017] S12. Collect flame temperature field distribution data B through an infrared thermal imager. Specifically, the 3 - 5μm band can be used for collection;
[0018] S13. Collect flame light intensity data C through a photoelectric sensor. The photoelectric sensor, such as a photodiode or a photomultiplier tube, can obtain a flame pulsation signal by monitoring changes in the flame light intensity.
[0019] Further, the S2 includes the following steps:
[0020] S21. Based on the Canny edge detection algorithm, identify and analyze the flame in the flame image data A, identify the flame height, flame area, flame centroid position, and flame swing amplitude, and generate flame feature data D=(d1,..., d p ,..., d υ ), p = 1, 2, 3,..., υ, d p represents the flame feature data of the p-th flame, and υ represents the maximum number of flames. The flame feature data includes flame height feature data, flame area feature data, flame centroid position feature data, and flame swing amplitude feature data; among them, the flame height feature data, flame area feature data, flame centroid position feature data, and flame swing amplitude feature data are the feature data obtained by normalizing the corresponding data of the flame height, flame area, flame centroid position, and flame swing amplitude respectively.
[0021] Further, the S3 includes the following steps:
[0022] S31. Based on the flame feature data D, calculate and process the degree of discreteness of the centroid position of each flame over time in consecutive image frames, and generate a set of flame centroid movement feature data E=(e1,..., e p ,..., e υ ), e pIt represents the characteristic data of the centroid movement of the p-th flame. The characteristic data of the centroid movement of the flame consists of the variance of the centroid movement of the flame. Therefore, the calculation formula for the characteristic data of the centroid movement of each flame is as follows:
[0023] Calculate the variance of the centroid of the flame in the x direction:
[0024]
[0025] Calculate the variance of the centroid of the flame in the y direction:
[0026]
[0027] Calculate the total variance of the centroid of the flame:
[0028] Var total = Var(X) + Var(Y)
[0029] Where, N represents the number of sampling frames (i.e., the number of flame images collected), i represents the i-th frame of the flame image, x represents the x-axis coordinate of the centroid of the flame, y represents the y-axis coordinate of the centroid of the flame, represents the mean value of the x-axis coordinates, that is represents the mean value of the Y-axis coordinates, that is e p It includes the total variance of the centroid of the flame corresponding to the flame images from frame 1 to frame N of the p-th flame.
[0030] Furthermore, the step S4 includes the following steps:
[0031] S41. Preprocess the flame light intensity data C to generate preprocessed flame light intensity data. The preprocessing can include low-pass filtering: used to filter out high-frequency noise (such as electromagnetic interference); band-pass filtering: used to retain the target frequency band (for example, 10Hz - 500Hz); wavelet denoising: used for non-stationary signals to separate noise and effective components; trend term elimination: used to remove the slow drift in the signal (such as the baseline shift caused by the change of combustion load) through moving average or polynomial fitting; signal segmentation: used to segment long-time signals (for example, each segment is 10 seconds) to ensure that the analysis focuses on the stable combustion stage;
[0032] S42. Perform frequency-domain analysis on the preprocessed data of the flame light intensity through Fourier transform, short-time Fourier transform, and wavelet transform to generate frequency-domain analysis data of the flame light intensity. Among them, Fourier transform is used to convert the time-domain signal into the frequency domain, output the power spectral density diagram, and identify the main frequency peak, that is, the flame pulsation frequency. Short-time Fourier transform is used to analyze the change of frequency over time (suitable for transient combustion processes). Wavelet transform is used for multi-scale decomposition of signals and is suitable for capturing mutations or local features. It can also be through: Zoom FFT to improve the resolution for a specific frequency band and accurately capture adjacent frequencies; Welch method to perform segmented averaging to reduce spectral noise.
[0033] S43. Based on the frequency-domain analysis data of the flame light intensity, extract the frequency component with the largest amplitude as the main frequency, and extract the frequencies that are integer multiples or fractional multiples of the main frequency to generate the flame pulsation frequency data F.
[0034] Further, the S5 includes the following steps:
[0035] S51. Collect and combine the flame centroid movement feature data set E, the flame pulsation frequency data F, and the flame temperature field distribution data B to generate the flame recognition feature data G = (E, F, B).
[0036] Further, the S6 includes the following steps:
[0037] S61. Collect the historical flame recognition feature data corresponding to each flame abnormal type to generate the flame abnormal type recognition feature data H = (h1,..., h q ,..., h τ ), q = 1, 2, 3,..., τ, h q is the q-th type of flame abnormal type recognition feature data, and τ is the maximum number of categories of flame abnormal types;
[0038] S62. In the search space of the flame abnormal type recognition feature data H, search for the flame abnormal type recognition feature data h q that matches the flame recognition feature data G, including the following steps:
[0039] S621. Initialize the algorithm parameters;
[0040] S622. Randomly generate N search particles in the H search space;
[0041] S623. Calculate the fitness value of each particle, that is, calculate the matching degree between the h q corresponding to the position of each particle in the H search space and the G. The fitness value is inversely proportional to the matching degree, that is, the higher the matching degree, the lower the fitness value, and the better the fitness;
[0042] S624. Update the position of each search particle using the position corresponding to the historical optimal fitness of the search particle to obtain the optimal position of the particle, and update the global optimal position using the optimal position of the particle with the best fitness among all search particles;
[0043] S625. Determine whether the algorithm has reached the set maximum number of iterations T. If not, return to S623. If so, output the flame anomaly type recognition feature data h corresponding to the global optimal position q to generate the flame anomaly type analysis data H fenxi .
[0044] Furthermore, the S7 includes the following steps:
[0045] S71. Collect the processing scheme data corresponding to the flame anomaly type to generate the flame anomaly type processing scheme data set J = (j1,..., j q ,..., j τ ), where j q represents the flame anomaly type processing scheme data corresponding to the q-th type of flame anomaly type;
[0046] S72. Search the flame anomaly type processing scheme data set J and the flame anomaly type analysis data H fenxi using binary search to generate the flame anomaly processing scheme return data J fanhui .
[0047] Furthermore, the S8 includes the following steps:
[0048] S81. Collect and combine the flame recognition feature data G, the flame anomaly type analysis data H fenxi and the flame anomaly processing scheme return data J fanhui to construct the flame monitoring management data K = (G, H fenxi , J fanhui ), and execute the corresponding flame anomaly processing scheme return data J fenxi corresponding to the flame recognition feature data G and the flame anomaly type analysis data H fanhui operation according to the flame monitoring management data K.
[0049] A fully digital flame monitoring and feedback system includes a CCD visible light camera, an infrared thermal imager, a photoelectric sensor, a data input module, a communication module, a storage, and a processor;
[0050] The CCD visible light camera is used to collect flame image data;
[0051] The infrared thermal imager is used to collect flame temperature field distribution data;
[0052] The photoelectric sensor is used to collect flame light intensity data;
[0053] The data input module is used to collect historical flame recognition feature data corresponding to each flame abnormal type and processing scheme data corresponding to the flame abnormal type;
[0054] The storage is used to store computer programs and collected flame image data, flame temperature field distribution data, flame light intensity data, historical flame recognition feature data corresponding to each flame abnormal type, and processing scheme data corresponding to the flame abnormal type;
[0055] The communication module is used to feedback the flame monitoring management data to the staff;
[0056] The processor is used to execute the computer program and execute a full-digital flame monitoring and feedback method.
[0057] 1. Compared with the prior art, a full-digital flame monitoring and feedback system and method provided by the present invention analyzes the flame height, flame area, flame centroid position, flame swing amplitude, flame centroid movement characteristics, and flame pulsation frequency by collecting flame image data, flame temperature field distribution data, and flame light intensity data, and then identifies the flame abnormal type based on the characteristics of the flame temperature field distribution data, flame centroid movement characteristics, and flame pulsation frequency, so as to timely issue a reminder according to the flame abnormal type and implement the corresponding processing scheme to ensure the safety of flame combustion.
[0058] 2. Compared with the prior art, a full-digital flame monitoring and feedback system and method provided by the present invention uses the flame temperature field distribution data, flame centroid movement characteristics, and flame pulsation frequency as search particles, and obtains the abnormal feature type matched by the flame based on the search for the global optimal position of the search particles, simplifies the abnormal type prediction steps, and reduces the possibility that the prediction result biases towards the local optimal solution. BRIEF DESCRIPTION OF THE DRAWINGS
[0059] 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 for use 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.
[0060] Figure 1 It is a method step diagram provided by an embodiment of the present invention;
[0061] Figure 2 It is a system structure block diagram provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0062] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.
[0063] In the following, example embodiments will be described more fully with reference to the accompanying drawings. However, the example embodiments may be embodied in different forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
[0064] In the case of no conflict, the embodiments of the present disclosure and the features in the embodiments may be combined with each other.
[0065] As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.
[0066] The terms used herein are only for the purpose of describing specific embodiments and are not intended to limit the present disclosure. As used herein, the singular forms "a" and "the" are also intended to include the plural forms unless the context clearly indicates otherwise. It will also be understood that when the terms "comprises" and / or "consists of" are used in this specification, it specifies the presence of the stated features, wholes, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or groups thereof.
[0067] The embodiments described herein may be described with reference to the plan views and / or cross-sectional views by means of the ideal schematic diagrams of the present disclosure. Therefore, the example illustrations may be modified according to the manufacturing technology and / or tolerances. Therefore, the embodiments are not limited to the embodiments shown in the drawings, but include modifications of the configurations formed based on the manufacturing process. Therefore, the regions illustrated in the drawings have schematic attributes, and the shapes of the regions shown in the drawings illustrate the specific shapes of the regions of the elements, but are not intended to be restrictive.
[0068] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and the present disclosure, and will not be interpreted as having an idealized or overly formal meaning unless expressly so defined herein.
[0069] Please refer to Figure 1 , a full digital flame monitoring and feedback method, comprising the following steps:
[0070] S1. Collect flame image data, flame temperature field distribution data, and flame light intensity data;
[0071] S11. Collect the flame image data A through a CCD visible light camera, where the CCD visible light camera can be deployed at the observation port of the combustion chamber to capture the changes in flame color and shape;
[0072] S12. Collect the flame temperature field distribution data B through an infrared thermal imager. Specifically, the 3-5μm band can be used for collection;
[0073] S13. Collect the flame light intensity data C through a photoelectric sensor. The photoelectric sensor such as a photodiode, a photomultiplier tube, etc. can obtain the flame pulsation signal by monitoring the change in flame light intensity.
[0074] S2. Perform flame contour recognition and analysis on the flame image data to generate flame feature data;
[0075] S21. Based on the Canny edge detection algorithm, identify and analyze the flame in the flame image data A, identify the flame height, flame area, flame centroid position, and flame swing amplitude, and generate the flame feature data D=(d1,...,d p ,...,d υ ), p = 1, 2, 3,..., υ, d p represents the flame feature data of the p-th flame, and υ represents the maximum number of flames. Among them, the flame feature data includes flame height feature data, flame area feature data, flame centroid position feature data, and flame swing amplitude feature data; the flame height feature data, flame area feature data, flame centroid position feature data, and flame swing amplitude feature data are the feature data obtained by normalizing the corresponding data of the flame height, flame area, flame centroid position, and flame swing amplitude respectively.
[0076] S3. Calculate and process the variance of the flame centroid movement based on the flame feature data to generate the flame centroid movement feature data;
[0077] S31. Based on the flame feature data D, calculate and process the degree of dispersion of the centroid position of each flame over time in consecutive image frames to generate the flame centroid movement feature data set E=(e1,...,e p ,...,e υ ), e p represents the flame centroid movement feature data of the p-th flame. The flame centroid movement feature data is composed of the variance of the flame centroid movement. Therefore, the calculation formula for the flame centroid movement feature data of each flame is as follows:
[0078] Calculate the variance of the flame centroid in the x direction:
[0079]
[0080] Calculate the variance of the flame centroid in the y direction:
[0081]
[0082] Calculate the total variance of the flame centroid:
[0083] Var total = Var(X) + Var(Y)
[0084] Where N represents the number of sampled frames (i.e., the number of flame images collected), i represents the i-th frame of the flame image, x represents the x-axis coordinate of the flame centroid, and y represents the y-axis coordinate of the flame centroid. represents the mean value of the x-axis coordinates, that is represents the mean value of the Y-axis coordinates, that is e p includes the total variance of the flame centroid corresponding to the flame images from frame 1 to frame N of the p-th flame.
[0085] S4. Calculate and analyze the flame pulsation frequency based on the flame light intensity data to generate flame pulsation frequency data;
[0086] S41. Preprocess the flame light intensity data C to generate preprocessed flame light intensity data. The preprocessing can include low-pass filtering: used to filter out high-frequency noise (such as electromagnetic interference); band-pass filtering: used to retain the target frequency band (for example, 10 Hz - 500 Hz); wavelet denoising: used for non-stationary signals to separate noise and effective components; trend term elimination: used to remove the slow drift in the signal (such as the baseline shift caused by the change in combustion load) through moving average or polynomial fitting; signal segmentation: used to segment long-time signals (for example, each segment is 10 seconds) to ensure that the analysis focuses on the stable combustion stage.
[0087] S42. Perform frequency domain analysis on the preprocessed flame light intensity data through Fourier transform, short-time Fourier transform, and wavelet transform to generate frequency domain analysis data of the flame light intensity. Among them, the Fourier transform is used to convert the time domain signal into the frequency domain, output the power spectral density diagram, and identify the main frequency peak, that is, the flame pulsation frequency; the short-time Fourier transform is used to analyze the change of frequency with time (suitable for transient combustion processes); the wavelet transform is used for multi-scale decomposition of signals, which is suitable for capturing mutations or local features. It can also be through: Zoom FFT to improve the resolution for a specific frequency band to accurately capture adjacent frequencies; Welch method for segment averaging to reduce spectral noise.
[0088] S43. Based on the frequency domain analysis data of the flame light intensity, extract the frequency component with the largest amplitude as the main frequency, and extract the frequencies that are integer multiples or fractional multiples of the main frequency to generate the flame pulsation frequency data F.
[0089] S5. Collect and combine the flame centroid movement characteristic data, flame pulsation frequency data, and flame temperature field distribution data to generate flame recognition characteristic data;
[0090] S51. Collect and combine the flame centroid movement characteristic data set E, flame pulsation frequency data F, and flame temperature field distribution data B to generate flame recognition characteristic data G = (E, F, B).
[0091] S6. Based on the flame recognition characteristic data and the flame abnormal type recognition characteristic data, analyze and process the flame abnormal type corresponding to the flame recognition characteristic data to generate flame abnormal type analysis data;
[0092] S61. Collect the historical flame recognition characteristic data corresponding to each flame abnormal type to generate flame abnormal type recognition characteristic data H = (h1,..., h q ,..., h τ ), q = 1, 2, 3,..., τ, h q is the q-th type of flame abnormal type recognition characteristic data, and τ is the maximum number of categories of flame abnormal types;
[0093] S62. In the search space of the flame abnormal type recognition characteristic data H, search for the flame abnormal type recognition characteristic data h q that matches the flame recognition characteristic data G, including the following steps:
[0094] S621. Initialize the algorithm parameters;
[0095] S622. Randomly generate N search particles in the H search space;
[0096] S623. Calculate the fitness value of each particle, that is, calculate the matching degree between h q corresponding to the position of each particle in the H search space and G. The fitness value is inversely proportional to the matching degree, that is, the higher the matching degree, the lower the fitness value, and the better the fitness;
[0097] S624. Use the position corresponding to the historical best fitness of each search particle to update the position of the search particle to obtain the particle optimal position, and use the particle optimal position with the best fitness among all search particles to update the global optimal position;
[0098] S625. Determine whether the algorithm has reached the set maximum number of iterations T. If not, return to S623. If so, output the flame abnormal type recognition characteristic data h q corresponding to the global optimal position to generate flame abnormal type analysis data H fenxi .
[0099] S7. Based on the flame anomaly type analysis data and the flame anomaly type processing solution data, perform a search process for the corresponding processing solution of the flame anomaly type, and generate the flame anomaly processing solution return data;
[0100] S71. Collect the processing solution data corresponding to the flame anomaly type, and generate a set of flame anomaly type processing solution data J = (j1,..., j q ,..., j τ ), where j q represents the flame anomaly type processing solution data corresponding to the qth type of flame anomaly type;
[0101] S72. Based on binary search, search the set of flame anomaly type processing solution data J and the flame anomaly type analysis data H fenxi to generate the flame anomaly processing solution return data J fanhui .
[0102] S8. Collect and combine the flame recognition feature data, the flame anomaly type analysis data, and the flame anomaly processing solution return data to construct the flame monitoring management data, and execute the flame monitoring feedback operation based on the flame monitoring management data;
[0103] S81. Collect and combine the flame recognition feature data G, the flame anomaly type analysis data H fenxi and the flame anomaly processing solution return data J fanhui to construct the flame monitoring management data K = (G, H fenxi , J fanhui ), and execute the corresponding flame anomaly processing solution return data J fenxi for the flame corresponding to the flame recognition feature data G and the flame anomaly type analysis data H fanhui operation.
[0104] Please refer to Figure 2 , the present invention also provides a full digital flame monitoring and feedback system, including a CCD visible light camera, an infrared thermal imager, a photoelectric sensor, a data input module, a communication module, a storage, and a processor;
[0105] The CCD visible light camera is used to collect the flame image data;
[0106] The infrared thermal imager is used to collect the flame temperature field distribution data;
[0107] The photoelectric sensor is used to collect the flame light intensity data;
[0108] The data input module is used to collect the historical flame recognition feature data corresponding to each flame anomaly type and the processing solution data corresponding to the flame anomaly type;
[0109] The memory is used to store computer programs and the collected flame image data, flame temperature field distribution data, flame light intensity data, historical flame recognition feature data corresponding to each flame anomaly type, and processing scheme data corresponding to the flame anomaly type;
[0110] The communication module is used to feedback the flame monitoring and management data to the staff;
[0111] The processor is used to execute the computer program and implement a full-digital flame monitoring and feedback method provided by the present invention.
[0112] Only some exemplary embodiments of the present invention are described by way of illustration above. Undoubtedly, for those of ordinary skill in the art, without departing from the spirit and scope of the present invention, the described embodiments can be modified in various different ways. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the protection scope of the claims of the present invention.
Claims
1. A fully digital flame monitoring and feedback method, characterized in that: The following steps are involved: S1, collecting flame image data, flame temperature field distribution data, and flame light intensity data; S2, performing flame contour recognition analysis on the flame image data to generate flame feature data; S3, calculating and processing the flame centroid movement variance based on the flame characteristic data to generate flame centroid movement characteristic data; S4. Calculating and analyzing the flame pulsation frequency based on the flame light intensity data to generate flame pulsation frequency data; S5, collecting and combining the flame mass center movement characteristic data, the flame pulsation frequency data and the flame temperature field distribution data to generate flame identification characteristic data; S6. Based on the flame identification feature data and the flame abnormality type identification feature data, analyzing and processing the flame abnormality type corresponding to the flame identification feature data to generate flame abnormality type analysis data; S7, based on the flame abnormality type analysis data and the flame abnormality type processing solution data, searching for the processing solution corresponding to the flame abnormality type, and generating flame abnormality processing solution return data; S8. Collect and combine the flame identification feature data, flame anomaly type analysis data and flame anomaly processing solution return data to construct flame monitoring management data, and perform flame monitoring feedback operations based on the flame monitoring management data.
2. A fully digital flame monitoring and feedback method according to claim 1, characterized in that: The S1 comprises the following steps: S11, collecting flame image data A through a CCD visible light camera; S12, collecting flame temperature field distribution data B through an infrared thermal imager; S13. Collect flame light intensity data C through a photoelectric sensor.
3. A fully digital flame monitoring and feedback method according to claim 2, characterized in that: The S2 comprises the following steps: S21, based on the Canny edge detection algorithm, the flame in the flame image data A is identified and analyzed to identify the flame height, flame area, flame centroid position, and flame swing amplitude, and generate flame feature data D=(d1, ..., d p , …, d υ ), p=1, 2, 3,…, υ, d p represents the flame characteristic data of the pth flame, and υ represents the maximum number of flames.
4. A fully digital flame monitoring and feedback method according to claim 3, characterized in that: The S3 comprises the following steps: S31, based on the flame characteristic data D, the discrete degree of the centroid position of each flame in the continuous image frames over time is calculated and processed to generate a flame centroid movement characteristic data set E = (e1, ..., e p ,…,e υ ), e p Represents the flame centroid movement characteristic data of the pth flame.
5. A fully digital flame monitoring and feedback method according to claim 4, characterized in that: The S4 comprises the following steps: S41, preprocessing the flame light intensity data C to generate flame light intensity preprocessing data; S42, performing frequency domain analysis processing on the flame light intensity preprocessing data by Fourier transform, short-time Fourier transform, and wavelet transform to generate flame light intensity frequency domain analysis data; S43. Based on the flame light intensity frequency domain analysis data, extract the frequency component with the largest amplitude as the main frequency, and extract the frequencies of integer multiples or fractional multiples of the main frequency to generate flame pulsation frequency data F.
6. A fully digital flame monitoring and feedback method according to claim 5, characterized in that: The S5 comprises the following steps: S51. Collect and combine the flame mass center movement characteristic data set E, flame pulsation frequency data F and flame temperature field distribution data B to generate flame identification characteristic data G=(E, F, B).
7. A fully digital flame monitoring and feedback method according to claim 6, characterized in that: The S6 comprises the following steps: S61, collecting historical flame identification feature data corresponding to each flame abnormality type, generating flame abnormality type identification feature data H = (h1, ..., h q ,…,h τ ), q=1, 2, 3,..., τ, h q is the qth type of flame anomaly identification feature data, τ is the maximum number of flame anomaly types; S62: Searching for flame abnormality type identification feature data h matching the flame identification feature data G in the flame abnormality type identification feature data H search space. q , including the following steps: S621, initializing algorithm parameters; S622, randomly generating N search particles in the H search space; S623, calculate the fitness value of each particle, that is, calculate the h corresponding to the position of each particle in the H search space q The matching degree with the G, the fitness value is inversely proportional to the matching degree, that is, the higher the matching degree, the lower the fitness value, and the better the fitness; S624, using the position corresponding to the historical optimal fitness of each search particle, updating the position of the search particle to obtain the optimal position of the particle, and using the optimal position of the particle with the best fitness among all search particles to update the global optimal position; S625: Determine whether the algorithm has reached the set maximum number of iterations T. If not, return to S623. If yes, output the flame abnormality type identification feature data h corresponding to the global optimal position. q , generate flame anomaly type analysis data H fenxi .
8. A fully digital flame monitoring and feedback method according to claim 7, characterized in that: The S7 comprises the following steps: S71, collect the processing solution data corresponding to the flame abnormality type, and generate the flame abnormality type processing solution data set J=(j1, ..., j q , …, j τ ), j q Indicates flame abnormality type processing solution data corresponding to the qth flame abnormality type; S72, based on binary search, the flame abnormality type processing solution data set J and the flame abnormality type analysis data set H fenxi Search and generate flame anomaly handling solution to return data J fanhui .
9. A fully digital flame monitoring and feedback method according to claim 8, characterized in that: The S8 comprises the following steps: S81, analyzing the flame recognition feature data G and the flame abnormality type data H fenxi and flame exception handling scheme return data J fanhui Collect and combine to construct flame monitoring management data K = (G, H fenxi , J fanhui ), and according to the flame monitoring management data K, the flame identification feature data G and the flame abnormality type analysis data H fenxi The corresponding flame executes the corresponding flame exception handling solution and returns data J fanhui Operation.
10. A fully digital flame monitoring and feedback system, used to execute a fully digital flame monitoring and feedback method according to any one of claims 1 to 9, characterized in that: Including CCD visible light camera, infrared thermal imager, photoelectric sensor, data input module, communication module, storage, processor; The CCD visible light camera is used to collect flame image data; The infrared thermal imager is used to collect flame temperature field distribution data; The photoelectric sensor is used to collect flame light intensity data; The data input module is used to collect historical flame identification feature data corresponding to each flame abnormality type and processing solution data corresponding to the flame abnormality type; The storage device is used to store computer programs and collected flame image data, flame temperature field distribution data, flame light intensity data, historical flame identification feature data corresponding to each flame abnormality type, and processing solution data corresponding to the flame abnormality type; The communication module is used to feed back the flame monitoring management data to the staff; The processor is used to execute the computer program to execute the fully digital flame monitoring and feedback method as described in any one of claims 1-9.