On-line detection method and equipment for combustion temperature field in furnace fused with self-learning

Through self-learning model and TCNN technology, the problem that thermal radiation imaging technology cannot be measured online during the cleaning of the image probe in the boiler furnace is solved, and the continuous and effective detection and adaptability of the temperature field distribution in the furnace is achieved, ensuring the stability of the boiler operation Sex and safety.

CN118982641APending Publication Date: 2024-11-19HUAZHONG UNIV OF SCI & TECH +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411013177.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-26
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

The existing thermal radiation imaging technology cannot perform online measurements during the cleaning of the image probe in the boiler furnace, resulting in large errors in the temperature field distribution measurement in the furnace or the measurement results cannot be obtained.

Method used

The self-learning model is adopted to calculate the temperature field distribution in the furnace by inputting the unit operation parameters. The process of establishing the self-learning model includes training of the characteristic working conditions of the unit operation parameters and temperature field data, and constructing it using the transposed convolutional neural network (TCNN) model. The mathematical relationship between the unit operating parameters and the temperature field distribution in the furnace ensures that online detection can still be carried out when the image probe cannot work normally, and the model is updated and optimized after the probe is cleaned to improve generalization capabilities.

Benefits of technology

It realizes continuous and effective online detection of the combustion temperature field distribution in the furnace under harsh environments, improves the reliability and adaptability of the detection, avoids measurement interruptions during probe cleaning, and enhances the ability to adapt to various combustion conditions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118982641A_ABST
    Figure CN118982641A_ABST
Patent Text Reader

Abstract

The invention belongs to the related technical field of combustion detection, and discloses an in-furnace combustion temperature field online detection method and equipment fused with self-learning, and the method comprises the following steps: when an image probe cannot work normally, stopping calculating the in-furnace combustion temperature field from a combustion image obtained by the image probe, and calculating the in-furnace combustion temperature field according to the calculated in-furnace combustion temperature field; the unit operation parameters at the current moment are input into the self-learning model, and the self-learning model outputs in-furnace temperature field distribution; the construction of the self-learning model comprises the following steps: carrying out probability density analysis on unit operation parameters in a preset time period in a special database to obtain a characteristic working condition training data set of the unit operation parameters and in-furnace temperature field data, and carrying out training learning on the self-learning model by adopting characteristic working condition data in the characteristic working condition training data set, and the mathematical relationship between the unit operation parameters and the in-furnace temperature field distribution is obtained. According to the invention, the continuous effectiveness and reliability of the thermal radiation imaging technology on the on-line detection of the distribution of the combustion temperature field in the furnace in a severe measurement environment are ensured.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field related to combustion detection, and more specifically, relates to an online detection method and device for combustion temperature field in a furnace integrating self-learning. Background Art

[0002] At present, coal-fired power generation still occupies a dominant position in power supply, but with the gradual maturity of new energy power generation technology and the continuous increase in the installed capacity of new energy, coal-fired power generation units will undertake more deep peak-shaving tasks. When coal-fired power generation units participate in deep peak-shaving, the flame fullness in the boiler furnace decreases, and problems such as unstable combustion and uneven heat load are prone to occur. The combustion in the furnace is a complex three-dimensional physical and chemical process. The temperature field in the furnace can directly reflect the combustion status in the furnace, and the rationality of its distribution will affect the stable operation of the boiler. Therefore, carrying out online detection of the furnace temperature field is of great practical significance to ensure the safety and economy of the boiler during deep peak-shaving.

[0003] Due to the high combustion temperature in the furnace, commonly used contact temperature measuring elements such as thermocouples and thermistors are prone to wear and tear, and are frequently replaced. They are only suitable for short-term measurements and can only perform single-point measurements, making it difficult to obtain the temperature field distribution in the furnace. Optical and acoustic non-contact temperature measurement methods do not have elements in contact with the combustion flame in the furnace, and can obtain signals with high temporal and spatial resolution. They are the main development direction for temperature distribution measurement in special environments such as the flame in the furnace of a coal-fired power generation unit boiler. The literature [Lou Chun. Engineering Combustion Diagnostics [M]. Beijing: China Electric Power Press, 2016] introduces non-contact measurement technologies that have been used for online detection of furnace temperature fields, mainly including acoustic tomography technology, absorption spectrum tomography technology and thermal radiation imaging technology; the literature [Yan Huibo, Tang Guangtong, Li Lujiang et al. New progress in the algorithm for measuring three-dimensional temperature fields in large furnaces by thermal radiation imaging [J]. Clean Coal Technology, 2022, 28(5): 97-108] points out that among the three technologies, thermal radiation imaging technology does not require external sound waves or laser signals, but only needs to obtain the combustion image signal in the furnace, and has the characteristics of compact system and easy implementation; and the image signal has high pixel resolution and fast response, and can provide temperature field distribution measurement results with high temporal and spatial resolution. Compared with the shortcomings of acoustic and absorption spectrum tomography technologies, whose measurement spatial resolution is limited by path and whose temporal resolution is large, thermal radiation imaging technology has greater application potential.

[0004] In practical applications, thermal radiation imaging technology requires opening observation holes in the boiler furnace and placing image probes at these observation holes. Although instrument air has been arranged at these measuring points to cool and purge the image probes, when the coal quality changes or the combustion is incomplete, the observation holes are prone to coking and other phenomena, blocking the normal transmission of the thermal radiation beam and causing the image signal to be distorted by dust and dirt. In severe cases, more severe dust and dirt problems such as dust accumulation and slagging will appear on the image probe, resulting in large errors in the online measurement of the temperature field distribution in the furnace, or even no measurement results can be obtained at all. The observation hole or image probe needs to be cleaned before the entire temperature measurement system can be put into normal use again. It can be seen that the existing thermal radiation imaging technology cannot carry out online measurement of the temperature field distribution in the furnace during the cleaning of the probe. Summary of the invention

[0005] In response to the above defects or improvement needs of the prior art, the present invention provides a method and device for online detection of the combustion temperature field in a furnace integrating self-learning, which aims to solve the problem that thermal radiation imaging technology cannot carry out online measurement of the temperature field distribution in the furnace during the cleaning of the probe.

[0006] To achieve the above object, according to one aspect of the present invention, a method for online detection of combustion temperature field in a furnace integrating self-learning is provided, and the detection method comprises the following steps:

[0007] When the image probe fails to work properly, the calculation of the combustion temperature field in the furnace from the combustion image obtained by the graphic probe is stopped, and the unit operation parameters at the current moment are input into the self-learning model, and the self-learning model outputs the distribution of the temperature field in the furnace;

[0008] Among them, the construction of the self-learning model includes the following steps: performing probability density analysis on the unit operating parameters within a predetermined time period in a dedicated database to obtain a characteristic operating condition training data set of unit operating parameters and furnace temperature field data, and using the characteristic operating condition data in the characteristic operating condition training data set to train the self-learning model to obtain the mathematical relationship between the unit operating parameters and the furnace temperature field distribution; wherein the dedicated database stores grouped image dust index, furnace temperature field data and unit operating parameters.

[0009] Furthermore, the detection method also includes the following steps: after the image probe is cleaned and the combustion image begins to be acquired normally, the dedicated database and the characteristic operating condition training data set are updated, and the characteristic operating condition data are repeatedly trained and learned to optimize the self-learning model.

[0010] Furthermore, according to the thermal radiation imaging model, the combustion temperature field in the furnace is calculated using the combustion image obtained from the image probe. At the same time, the combustion image is partitioned and recognized, and the image dust index is given by statistical averaging. The image dust index obtained online is compared with the set image dust threshold. The comparison result is used to determine whether the combustion image obtained by the image probe is normal, and then a dedicated database is constructed.

[0011] Furthermore, the image dirtiness index k obtained online is j The image gray threshold λ is set k Make a comparison to determine whether the combustion image obtained by the image probe is normal; if k j Less than λ k , the image probe works normally, the acquired combustion image is normal, and the calculated combustion temperature field in the furnace is valid; if k j Greater than λ k , then there is a dust stain problem on the image probe.

[0012] Furthermore, the unit operating parameters include load, total coal volume, feed water volume, main steam flow, total primary air volume, total secondary air volume, and secondary air adjustment damper opening; the dedicated database includes three main data tables: image ash pollution index table, furnace temperature field data table, and unit key operating parameter table; each table contains a unique identifier and a timestamp for data association and query.

[0013] Furthermore, the grayscale index k of each measuring point image is calculated by the following formula: j ,j=1,2,…l:

[0014]

[0015] Where N t ′ N is the number of pixels in each area of ​​the burning image whose brightness value is lower than the grayscale threshold. t The total number of pixels in each area.

[0016] Furthermore, a probability density curve is drawn with the load data to determine the load range, and characteristic operating conditions such as the minimum operating load, the maximum operating load, and the normal operating load are screened out; then, a probability density analysis is performed on the different total coal amounts under each load to screen out the characteristic operating conditions of different total coal amounts; then, the different water feed amounts under each total coal amount are screened and classified, and so on, and finally the characteristic operating conditions of the key operating parameters of the unit are determined as input data; the calculated furnace temperature field distribution data corresponding to each set of characteristic operating conditions is found as output data, and the characteristic operating condition training data set is constructed based on this.

[0017] Furthermore, the mathematical relationship between the unit operating parameters and the temperature field distribution in the furnace is:

[0018] T i =f(P1,P2,…,P x ) (2)

[0019] Where, T i is the temperature field data in the furnace; P1, P2, ..., P x It is the unit operating parameters.

[0020] The present invention also provides an online detection system for the combustion temperature field in a furnace that integrates self-learning. The system includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it executes the online detection method for the combustion temperature field in a furnace that integrates self-learning as described above.

[0021] The present invention also provides a computer-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the above-mentioned method for online detection of combustion temperature field in a furnace integrating self-learning.

[0022] In general, compared with the prior art, the above technical solutions conceived by the present invention mainly have the following advantages:

[0023] Beneficial effects:

[0024] 1. By establishing a self-learning model to self-learn the detection results of the combustion temperature field distribution in the furnace, the mathematical relationship between the unit operating parameters and the temperature field distribution in the furnace is obtained. When the image probe of the thermal radiation imaging temperature measurement technology cannot work normally, the established self-learning model is called, and the unit operating parameters are input to give the temperature field distribution in the furnace online, ensuring the continuous effectiveness and reliability of the thermal radiation imaging technology in the online detection of the combustion temperature field distribution in the furnace under harsh measurement environments.

[0025] 2. The detection method also includes the following steps: after the image probe is cleaned and the combustion image is acquired normally, the dedicated database and the characteristic operating condition training data set are updated, and the characteristic operating condition data are repeatedly trained and learned to optimize the self-learning model, thereby improving the generalization ability of the self-learning model, so that the mathematical relationship between the unit operating parameters and the temperature field distribution in the furnace can adapt to various combustion conditions.

[0026] 3. TCNN can effectively utilize low-dimensional input data, extract its low-dimensional features to map high-dimensional features, and then predict high-dimensional output data. Moreover, the method only needs to compile relevant software programs to realize the functions, without adding hardware equipment, and is easy to implement. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a flow chart of a method for online detection of combustion temperature field in a furnace integrating self-learning provided by the present invention;

[0028] Figure 2 This is a structural diagram of the TCNN model involved in the present invention. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0030] See also Figure 1 and Figure 2 The present invention provides an online detection method for the combustion temperature field in a furnace integrated with self-learning. The detection method stores the temperature field data detected by the image probe when it is working normally and the corresponding unit operating parameters in a dedicated database, and constructs a self-learning model. The mathematical relationship between the temperature field distribution in the furnace and the unit operating parameters is obtained by preprocessing and training the data of the characteristic working conditions. When the image probe cannot work normally, the established self-learning model is called, and the unit operating parameters are input to give the temperature field distribution in the furnace online. After the image probe is cleaned and starts to work normally, the above process is repeated, and the self-learning model is optimized by updating the data and repeatedly learning, so that the self-learning model can continuously adapt to various combustion working conditions, ensuring the continuous effectiveness and reliability of the online detection of the combustion temperature field distribution in the furnace by the thermal radiation imaging technology under harsh measurement environments.

[0031] The detection method comprises the following steps: when the image probe fails to work properly, the calculation of the combustion temperature field in the furnace is stopped from the combustion image acquired by the graphic probe, and the unit operation parameters at the current moment are input into the self-learning model, and the self-learning model outputs the distribution of the temperature field in the furnace;

[0032] Among them, the construction of the self-learning model includes the following steps: performing probability density analysis on the unit operating parameters within a predetermined time period in a dedicated database to obtain a characteristic operating condition training data set of unit operating parameters and furnace temperature field data, and using the characteristic operating condition data in the characteristic operating condition training data set to train the self-learning model to obtain the mathematical relationship between the unit operating parameters and the furnace temperature field distribution; wherein the dedicated database stores grouped image dust index, furnace temperature field data and unit operating parameters.

[0033] In one embodiment, the detection method also includes the following steps: after the image probe is cleaned and the combustion image is acquired normally, the dedicated database and the characteristic operating condition training data set are updated, and the characteristic operating condition data are repeatedly trained and learned to optimize the self-learning model, thereby improving the generalization ability of the self-learning model so that the mathematical relationship between the unit operating parameters and the temperature field distribution in the furnace can adapt to various combustion conditions.

[0034] The construction of the special database: according to the thermal radiation imaging model, the combustion temperature field in the furnace is calculated using the combustion image obtained from the image probe; at the same time, the combustion image is partitioned and identified and the image dust index is given by statistical averaging, and the image dust index obtained online is compared with the set image dust threshold, and the combustion image obtained by the image probe is judged according to the comparison result to determine whether it is normal, and then a special database is constructed, and the image dust index, furnace temperature field data and unit operation parameters are stored in the special database. Specifically, it includes the following sub-steps:

[0035] S11: Set l measuring points at different heights of the boiler, obtain the combustion flame image through the image probe installed at the measuring point, and use the thermal radiation imaging method to obtain the three-dimensional temperature field data T of the m grid units in the furnace. i ,i=1,2,…m.

[0036] S12: Divide the combustion image obtained by the image probe into n areas, and the exposure time is t e , perform statistical processing on the brightness value R of each pixel in the area under this exposure time, if the brightness value of the pixel is lower than the grayscale threshold λ R , then the image signal at that location is judged to be gray and distorted. The gray and distorted index k of each measuring point image is calculated by the following formula: j ,j=1,2,…l:

[0037]

[0038] Where N t ′ N is the number of pixels in each area of ​​the burning image whose brightness value is lower than the grayscale threshold. t The total number of pixels in each area.

[0039] The image dirtiness index k obtained online j The image gray threshold λ is set k Make a comparison to determine whether the combustion image obtained by the image probe is normal. j Less than λ k , the image probe works normally, the acquired combustion image is normal, and the calculated combustion temperature field in the furnace is effective and reliable; if k j Greater than λ k, then there will be a serious dust pollution problem on the image probe, which will cause a large error in the online measurement of the temperature field distribution in the furnace or the inability to obtain the measurement results.

[0040] S13: Design and establish a dedicated database to record the image dust index k when the image probe is working normally j , furnace temperature field data T i And the key operating parameters of the unit P1, P2, ..., P x (load, total coal volume, feed water volume, main steam flow, total primary air volume, total secondary air volume, secondary air damper opening) are stored in a dedicated database. The dedicated database includes three main data tables: image ash pollution index table, furnace temperature field data table and unit key operating parameter table; each table contains a unique identifier and a timestamp for data association and query.

[0041] Construction of self-learning model: When the image probe is working normally, the probability density analysis of the unit operating parameters within a certain time range in the dedicated database is performed to construct a characteristic operating condition training data set including the unit operating parameters and the furnace temperature field data. The characteristic operating condition data is preprocessed and the self-learning model is trained to obtain the mathematical relationship between the unit operating parameters and the furnace temperature field distribution. The specific steps include the following:

[0042] S21: When the image probe is working normally, a probability density analysis is performed on the unit operating parameters within a certain time range t in the dedicated database to construct a furnace temperature field data T i 、Key operating parameters of the unit P1, P2, ..., P x Characteristic operating condition training data set. Draw a probability density curve with load data, determine the load range, and screen out characteristic operating conditions such as the minimum operating load, maximum operating load, and normal operating load; then perform probability density analysis on different total coal amounts under each load, and screen out characteristic operating conditions with different total coal amounts; then screen and classify different water feed amounts under each total coal amount, and so on, and finally determine the characteristic operating conditions of the key operating parameters of the unit as input data; find out the calculated furnace temperature field distribution data corresponding to each set of characteristic operating conditions as output data, and use this to construct the characteristic operating condition training data set.

[0043] S22: Establish a transpose convolutional neural network (TCNN) model with unit operating parameters as input and furnace temperature field data as output. Standardize the characteristic operating condition data, input the preprocessed unit operating parameters into the fully connected layer for nonlinear transformation, reconstruct the output vector and pass it through the transpose convolution layer to complete the upsampling operation, and output the three-dimensional temperature field data in the furnace. During the training process, according to the error between the output temperature field data and the target temperature field data, the weights of the fully connected layer and the transpose convolution layer are updated through the back propagation process, and the training and learning process of the TCNN model is completed by repeated iterations to obtain the mathematical relationship between the unit operating parameters and the temperature field distribution in the furnace:

[0044] T i =f(P1,P2,…,P x ) (2)

[0045] The calling of the self-learning model includes the following steps: when the image probe cannot work normally, stop calculating the combustion temperature field in the furnace from the combustion image, call the established self-learning model, input the current unit operating parameters, and output the temperature field distribution in the furnace. At the same time, compare the output furnace temperature field distribution with the furnace temperature field data under similar working conditions in the dedicated database, give the statistical average error between the two, and store the furnace temperature field data with a statistical average error less than the error threshold and the corresponding key unit operating parameters in the dedicated database. Specifically includes the following sub-steps:

[0046] S31: When the image grayscale index k j Greater than the image gray threshold λ k When the image probe cannot work normally, the calculation of the combustion temperature field in the furnace from the combustion image is stopped, and the established TCNN model (self-learning model) is called to input the current key operating parameters of the unit P1, P2, ..., P x , you can output the temperature field distribution T in the furnace i .

[0047] S32: Compare the output furnace temperature field distribution with the furnace temperature field data under similar working conditions in the dedicated database, give the statistical average error σ between the two, and set the statistical average error to be less than the error threshold λ. σ The furnace temperature field data and the corresponding key operating parameters of the unit are stored in a dedicated database. The statistical average error is given by the following formula:

[0048]

[0049] Where, T i is the temperature field data of m grid units in the furnace output by the TCNN model, It is the furnace temperature field data under similar working conditions in the special database.

[0050] The iterative optimization of the self-learning model includes the following steps: after the image probe is cleaned and the combustion image is normally acquired, the above steps are repeated, and the data are repeatedly learned by updating the dedicated database and the characteristic working condition training data set, and the self-learning model is optimized to improve its generalization ability, so that the mathematical relationship between the unit operating parameters and the temperature field distribution in the furnace can continuously adapt to various combustion working conditions. Specifically, it includes the following sub-steps:

[0051] After the image probe is cleaned and the combustion image is acquired normally, the combustion temperature field T in the furnace is recalculated. i and image dirt index k j , judge whether the acquired combustion image is normal, and update the calculation results to the dedicated database and characteristic working condition training data set to ensure the diversity and coverage of the data. By updating the dedicated database and characteristic working condition training data set, repeatedly learning data, optimizing the self-learning model, reducing the statistical average error of the furnace temperature field distribution, and improving its generalization ability, the mathematical relationship between the unit operating parameters and the furnace temperature field distribution can continuously adapt to various combustion conditions.

[0052] The present invention also provides an online detection system for the combustion temperature field in a furnace that integrates self-learning. The system includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, it executes the online detection method for the combustion temperature field in a furnace that integrates self-learning as described above.

[0053] The present invention also provides a computer-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the above-mentioned method for online detection of combustion temperature field in a furnace integrating self-learning.

[0054] The present invention is further described in detail below with reference to specific embodiments.

[0055] Example 1

[0056] (1) Furnace combustion image processing

[0057] This example takes a boiler in Xinjiang as an example. The rated power of the boiler is 350MW, and the combustion mode is four-corner tangential combustion. A total of 12 measuring points are set on the front and rear walls of the boiler at heights of 27m, 34m, and 41m, with 4 measuring points at each height. The image probe is used to obtain the combustion flame image, and the three-dimensional combustion temperature field data T of 10×10×10 grid units in the furnace are obtained by thermal radiation imaging method. iThe pixel size of the burning flame image is 706 (width) × 576 (height). The burning flame image is divided into five regions along the width direction, with pixel ranges of 1-160, 161-300, 301-420, 421-550, and 551-706 respectively; the exposure time t e Set to 1 / 400s, gray threshold λ for 5 regions R They are 50, 70, 90, 110, and 130 respectively, and the image gray threshold λ R is 0.4, and the image dust index k of the 12 measuring points is calculated by formula (1): j If the value is less than 0.4, the image probe is considered to be working properly. If the value is greater than 0.4, the image probe is considered to have a dust pollution problem. The calculated image dust pollution index, furnace temperature field data, and unit operation parameters are stored in three tables of a dedicated database. Each data item in the table has an auto-incrementing ID as an identifier, and the timestamp records the time when the data is calculated.

[0058] (2) Modeling of self-learning model of temperature field in furnace

[0059] A probability density analysis was performed on the unit operating parameters in the dedicated database for the past three months, and a probability density curve of the load was drawn, with the load as the horizontal axis and the frequency as the vertical axis. Three load characteristic operating conditions of 310-330MW, 270-290MW, and 230-250MW were screened out; then a probability density analysis was performed on the total coal quantity under each load, and two total coal quantity characteristic operating conditions of 140-165t / h and 165-190t / h were screened out; and so on, 800-900t / h, 900-1000t / h, 1000- The three characteristic operating conditions of water supply volume of 1100t / h, the three characteristic operating conditions of total primary air volume of 380-400t / h, 400-440t / h, and 440-460t / h, the three characteristic operating conditions of total secondary air volume of 470-610t / h, 610-750t / h, and 750-890t / h are selected. The calculated furnace temperature field data corresponding to each set of characteristic operating conditions are found out, and a characteristic operating condition data set including 162 sets of unit operating parameters and furnace temperature field data is constructed. 80% of the data set is taken as the training set, and the remaining 20% ​​of the data set is taken as the test set.

[0060] A transposed convolutional neural network model was established to standardize the characteristic operating condition data. The preprocessed unit operating parameters were input into the fully connected layer for nonlinear transformation. The Adam algorithm was used as the optimization method, the Relu function was used as the activation function, and the mean square error (MSE) was used as the loss function. The vector output by the fully connected layer was reconstructed into a 5×5 matrix, and then the three-dimensional temperature field data in the furnace was output after the upsampling operation of the transposed convolutional layer. In order to prevent the model from overfitting, a Dropout layer was added to the model to improve the generalization ability of the model. 1500 training iterations were performed to optimize the model parameters and reduce the training loss function to 0.10023. The mathematical relationship between the key operating parameters of the unit and the temperature field distribution in the furnace was obtained. The test set was used to test the prediction ability of the TCNN model. The three load characteristic conditions were predicted and compared with the corresponding target values. After calculation, the average relative errors of the prediction results under the load characteristic conditions of 310-330MW, 270-290MW, and 230-250MW were 2.386%, 2.713%, and 2.748%, respectively.

[0061] (3) Calling the self-learning model of the temperature field in the furnace

[0062] When the image dirt index k j When it is greater than 0.4, that is, the image probe cannot work normally, stop calculating the furnace combustion temperature field from the combustion image, call the trained TCNN model, input the current load, total coal volume, water supply, main steam flow, total primary air volume, total secondary air volume, and secondary air damper opening of the unit, and then output the furnace temperature field distribution. Compare the output furnace temperature field distribution with the furnace temperature field data under similar working conditions in the dedicated database, and calculate the statistical average error σ and error threshold λ through formula (2) σ It is set to 0.05, and the furnace temperature field data with a statistical average error less than 0.05 and the corresponding unit operating parameters are stored in a dedicated database.

[0063] (4) Iterative optimization of the self-learning model of the temperature field in the furnace

[0064] After the image probe is cleaned and the combustion image is acquired, the image dust index k of the 12 measuring points is calculated by formula (1): j , and the image gray threshold λ kFor comparison, the image dust index, furnace temperature field data and unit operating parameters when the value is less than 0.4 are updated to the dedicated database, and the characteristic working condition training data set is updated at the same time to train the TCNN model. Through continuous updating and learning of data, the self-learning model is optimized, the statistical average error of the furnace temperature field distribution is reduced, and its generalization ability is improved, so that the mathematical relationship between the key operating parameters of the unit and the furnace temperature field distribution can better adapt to various combustion conditions.

[0065] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for online detection of combustion temperature field in furnace integrating self-learning, It is characterized in that The detection method comprises the following steps: When the image probe fails to work properly, the calculation of the combustion temperature field in the furnace from the combustion image obtained by the graphic probe is stopped, and the unit operation parameters at the current moment are input into the self-learning model, and the self-learning model outputs the distribution of the temperature field in the furnace; Among them, the construction of the self-learning model includes the following steps: performing probability density analysis on the unit operating parameters within a predetermined time period in a dedicated database to obtain a characteristic operating condition training data set of unit operating parameters and furnace temperature field data, and using the characteristic operating condition data in the characteristic operating condition training data set to train the self-learning model to obtain the mathematical relationship between the unit operating parameters and the furnace temperature field distribution; wherein the dedicated database stores grouped image dust index, furnace temperature field data and unit operating parameters.

2. The method for online detection of combustion temperature field in a furnace integrating self-learning as claimed in claim 1, Features: The detection method also includes the following steps: after the image probe is cleaned and the combustion image is normally acquired, the dedicated database and the characteristic operating condition training data set are updated, and the characteristic operating condition data are repeatedly trained and learned to optimize the self-learning model.

3. The method for online detection of combustion temperature field in a furnace integrating self-learning as claimed in claim 1, Features: According to the thermal radiation imaging model, the combustion temperature field in the furnace is calculated using the combustion image obtained from the image probe. At the same time, the combustion image is partitioned and recognized, and the image dirt index is given by statistical averaging. The image dirt index obtained online is compared with the set image dirt threshold. The comparison result is used to determine whether the combustion image obtained by the image probe is normal, and then a dedicated database is constructed.

4. The method for online detection of combustion temperature field in a furnace integrating self-learning as claimed in claim 3, Features: The image dirtiness index k obtained online j The image gray threshold λ is set k Make a comparison to determine whether the combustion image obtained by the image probe is normal; if k j Less than λ k , the image probe works normally, the acquired combustion image is normal, and the calculated combustion temperature field in the furnace is valid; if k j Greater than λ k , then there is a dust stain problem on the image probe.

5. The method for online detection of combustion temperature field in a furnace integrating self-learning as claimed in claim 1, Features: The unit operating parameters include load, total coal volume, feed water volume, main steam flow, total primary air volume, total secondary air volume, and secondary air regulating damper opening; The dedicated database includes three main data tables: image dust index table, furnace temperature field data table and unit key operating parameter table; each table contains a unique identifier and a timestamp to facilitate data association and query.

6. The method for online detection of combustion temperature field in a furnace integrating self-learning as claimed in claim 4, Features: The gray dirt index k of each measuring point image is calculated by the following formula: j ,j=1,2,…l: Where N′ t N is the number of pixels in each area of ​​the burning image whose brightness value is lower than the grayscale threshold. t The total number of pixels in each area.

7. The method for online detection of combustion temperature field in a furnace integrating self-learning as claimed in claim 1, Features: Draw a probability density curve with load data, determine the load range, and screen out characteristic operating conditions such as the minimum operating load, maximum operating load, and normal operating load; then perform probability density analysis on different total coal amounts under each load to screen out characteristic operating conditions with different total coal amounts; then screen and classify different water feed amounts under each total coal amount, and so on, and finally determine the characteristic operating conditions of the key operating parameters of the unit as input data; find out the calculated furnace temperature field distribution data corresponding to each set of characteristic operating conditions as output data, and use this to construct a characteristic operating condition training data set.

8. The method for online detection of combustion temperature field in a furnace by integrating self-learning according to any one of claims 1 to 7, Features: The mathematical relationship between the unit operating parameters and the temperature field distribution in the furnace is: T i =f(P 1 ,P 2 ,…,P x ) (2) Where, T i is the temperature field data in the furnace; P 1 ,P 2 ,…,P x It is the unit operating parameters.

9. An online detection system for combustion temperature field in furnace integrating self-learning, Features: The system includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the on-line detection method for the combustion temperature field in a furnace integrating self-learning as described in any one of claims 1 to 8 is executed.

10. A computer-readable storage medium, Features: The computer-readable storage medium stores machine-executable instructions. When the machine-executable instructions are called and executed by the processor, the machine-executable instructions prompt the processor to implement the online detection method for the combustion temperature field in a furnace integrating self-learning as described in any one of claims 1-8.