Intelligent acquisition monitoring system and method for preventing boiler coking

Abstract

The invention provides an intelligent collecting and monitoring system and method for preventing boiler coking, and relates to the field of intelligent control. The method comprises the following steps: analyzing a monitoring video of boiler combustion, oxygen content values in a boiler at a plurality of preset time points and temperature values of a boiler hearth opening to obtain a relationship among boiler coking, the temperature values and the oxygen content values, and further judging whether the opening degree of an air door in the boiler needs to be adjusted; the oxygen content in the boiler is controlled by controlling the opening degree of the air door, so that the temperature of a hearth opening of the boiler is slightly lower than the melting point of coal ash, and the coking degree of the boiler is reduced.

Description

Technical Field

[0001] The present application relates to the field of intelligent control, and more specifically, to an intelligent data collection and monitoring system and method for preventing boiler coking. Background Art

[0002] In recent years, China has made major breakthroughs in high-voltage and ultra-high-voltage technologies, which has not only significantly enhanced the international competitiveness of domestic power companies, but also promoted technological progress in the power supporting field. However, while power companies are developing rapidly, they are also facing a series of challenges and problems.

[0003] At present, China's main way of producing electricity is thermal power generation, and as the core equipment of thermal power generation, the operating efficiency and effect of power plant boilers have an important impact on electricity production. Large boilers mainly rely on coal combustion to obtain energy, but a large amount of coal ash and oxides will be produced during the combustion process. The deposition and coking of these coal ash inside the boiler has become a problem that needs to be solved urgently. Coal ash deposition and coking will not only affect the heat transfer efficiency of the boiler and reduce thermal efficiency, but also cause unstable boiler operation and increase the risk of failure. At the same time, removing these deposits and coking requires a lot of manpower and material resources, which increases the operating costs of the enterprise.

[0004] Therefore, an intelligent data collection and monitoring solution is needed to prevent boiler coking. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an intelligent collection and monitoring system and method for preventing boiler coking, which analyzes the monitoring video of boiler combustion and the oxygen content values ​​in the boiler and the temperature values ​​of the boiler furnace mouth at multiple predetermined time points to obtain the relationship between boiler coking and the temperature value and oxygen content value, and then judges whether it is necessary to adjust the opening degree of the air door in the boiler, and controls the oxygen content in the boiler by controlling the opening degree of the air door, thereby ensuring that the temperature of the boiler furnace mouth is slightly lower than the melting point of coal ash, thereby reducing the coking degree of the boiler.

[0006] According to one aspect of the present application, there is provided an intelligent collection and monitoring system for preventing boiler coking, which comprises:

[0007] A boiler combustion data acquisition module is used to obtain the temperature value of the boiler furnace port and the oxygen content value in the boiler at multiple predetermined time points within a predetermined time period, as well as a monitoring video of the boiler combustion within the predetermined time period;

[0008] A boiler combustion data encoding module, used for performing feature encoding on the temperature values ​​of the boiler furnace port and the oxygen content value in the boiler at the plurality of predetermined time points and the monitoring video of the boiler combustion within the predetermined time period to obtain a control feature matrix;

[0009] The control result generation module is used to determine whether it is necessary to adjust the opening degree of the air door in the boiler based on the control characteristic matrix.

[0010] In the above-mentioned intelligent data collection and monitoring system for preventing boiler coking, the boiler combustion data encoding module includes: a video data encoding unit, which is used to extract the coking process features of the monitoring video of the boiler combustion within the predetermined time period to obtain a boiler coking tracking feature matrix; a numerical data encoding unit, which is used to extract the coking factor features of the temperature values ​​of the boiler furnace mouth and the oxygen content values ​​in the boiler at the plurality of predetermined time points to obtain a coking influence factor feature matrix; a feature association unit, which is used to multiply the coking influence factor feature matrix and the boiler coking tracking feature matrix to obtain the control feature matrix.

[0011] In the above-mentioned intelligent collection and monitoring system for preventing boiler coking, the video data encoding unit includes: a key frame extraction subunit, which is used to sample the monitoring video of the boiler combustion within the predetermined time period to obtain a plurality of boiler combustion monitoring key frames; a boiler coking process feature generation subunit, which is used to input the plurality of boiler combustion monitoring key frames into a combustion process encoder based on a temporal attention mechanism to obtain a boiler coking tracking feature map; a feature map dimensionality reduction subunit, which is used to pool the boiler coking tracking feature map along the channel dimension to obtain the boiler coking tracking feature matrix.

[0012] In the above-mentioned intelligent collection and monitoring system for preventing boiler coking, the key frame extraction subunit is used to: sample the monitoring video of boiler combustion within the predetermined time period at a predetermined sampling frequency to obtain the multiple boiler combustion monitoring key frames.

[0013] In the above-mentioned intelligent data collection and monitoring system for preventing boiler coking, the numerical data encoding unit includes: a numerical data structuring subunit, which is used to arrange the temperature values ​​of the boiler furnace mouth and the oxygen content values ​​in the boiler at the multiple predetermined time points into a temperature time series input vector and an oxygen content time series input vector according to the time dimension respectively; a numerical data association subunit, which is used to associate the temperature time series input vector with the oxygen content time series input vector to obtain a temperature-oxygen content association matrix; a coking influence factor feature extraction subunit, which is used to pass the temperature-oxygen content association matrix through a coking influence factor feature extractor with a block structure to obtain the coking influence factor feature matrix.

[0014] In the above-mentioned intelligent collection and monitoring system for preventing boiler coking, the numerical data association subunit is used to calculate the temperature-oxygen content association matrix between the temperature time series input vector and the oxygen content time series input vector using the following association coding formula; wherein the association coding formula is:

[0015]

[0016] Where V a represents the temperature timing input vector, represents the transposed vector of the temperature timing input vector, V b represents the oxygen content time series input vector, M represents the temperature-oxygen content association matrix, Represents vector multiplication.

[0017] In the above-mentioned intelligent data collection and monitoring system for preventing boiler coking, the control result generation module includes: a feature matrix dimensionality reduction unit, which is used to expand the control feature matrix to obtain a control feature vector; a feature optimization unit, which is used to perform local distribution equilibrium adaptive correction on the control feature vector based on subclass label related metrics to obtain an optimized control feature vector; a classification unit, which is used to pass the optimized control feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether it is necessary to adjust the opening degree of the air damper in the boiler.

[0018] In the above-mentioned intelligent collection and monitoring system for preventing boiler coking, the feature optimization unit is used to: determine the weight learning parameter matrix of the classifier; extract each column vector in the weight learning parameter matrix to obtain a set of weight learning parameter column vectors; and calculate the metric correlation prediction factor between each weight learning parameter column vector in the set of weight learning parameter column vectors and the control feature vector using the following formula to obtain a metric correlation prediction factor modulation vector, wherein the formula is:

[0019]

[0020] Among them, V i represents each weight learning parameter column vector in the set of weight learning parameter column vectors, V c represents the control feature vector, It means subtracting by position. is the superposition norm of the two norms of the eigenvectors, ‖·‖ F is the Frobenius norm of the eigenvector, M p represents the weight learning parameter matrix, exp(·) represents the exponential operation with the natural constant e as the base, V p Representing the metric correlation prediction factor modulation vector; taking the metric correlation prediction factor modulation vector as the weight vector, calculating the position point multiplication between the metric correlation prediction factor modulation vector and the control feature vector to obtain the optimized control feature vector.

[0021] According to another aspect of the present application, there is provided an intelligent collection and monitoring method for preventing boiler coking, comprising:

[0022] Obtaining the temperature value of the boiler furnace port and the oxygen content value in the boiler at multiple predetermined time points within a predetermined time period and the monitoring video of the boiler combustion within the predetermined time period;

[0023] Performing feature coding on the temperature values ​​of the boiler furnace port and the oxygen content values ​​in the boiler at the plurality of predetermined time points and the monitoring video of the boiler combustion within the predetermined time period to obtain a control feature matrix;

[0024] Based on the control characteristic matrix, it is determined whether the opening degree of the air damper in the boiler needs to be adjusted.

[0025] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0027] Figure 1 The system block diagram of the intelligent data collection and monitoring system for preventing boiler coking according to an embodiment of the present application.

[0028] Figure 2 The present invention is a block diagram of a boiler combustion data encoding module in an intelligent collection and monitoring system for preventing boiler coking according to an embodiment of the present application.

[0029] Figure 3 The present invention is a block diagram of a video data encoding unit in an intelligent acquisition and monitoring system for preventing boiler coking according to an embodiment of the present application.

[0030] Figure 4 It is a block diagram of a numerical data encoding unit in an intelligent data collection and monitoring system for preventing boiler coking according to an embodiment of the present application.

[0031] Figure 5 It is a block diagram of a control result generation module in an intelligent collection and monitoring system for preventing boiler coking according to an embodiment of the present application.

[0032] Figure 6 The present invention is a flowchart of an intelligent data collection and monitoring method for preventing boiler coking according to an embodiment of the present application. DETAILED DESCRIPTION

[0033] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0034] As mentioned in the above background technology, in recent years, the country's high-voltage and ultra-high-voltage technologies have made significant progress, which has not only enhanced the international status of domestic power companies, but also made remarkable achievements in the field of power supporting. However, we must also clearly see that while power companies are developing rapidly, they are also facing a series of problems. At present, China's main way of producing electricity is thermal power generation, and the efficiency and effect of power plant boilers as the core equipment of thermal power generation have attracted widespread attention. Large boilers mainly rely on coal combustion to obtain energy, but a large amount of coal ash and oxides will be produced after combustion. These coal ashes are deposited and coked inside the boiler, which is an unavoidable problem. If they are not handled in time, they will inevitably have an adverse effect on power production in the long run. Therefore, an intelligent collection and monitoring solution for preventing boiler coking is expected.

[0035] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also demonstrated a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields. The development of deep learning and neural networks has provided new solutions and solutions for intelligent collection and monitoring to prevent boiler coking.

[0036] Figure 1 1 is a system block diagram of an intelligent data collection and monitoring system for preventing boiler coking according to an embodiment of the present application. Figure 1 As shown, in the intelligent data collection and monitoring system 100 for preventing boiler coking, it includes: a boiler combustion data collection module 110, which is used to obtain the temperature value of the boiler furnace mouth and the oxygen content value in the boiler at multiple predetermined time points within a predetermined time period, as well as a monitoring video of the boiler combustion within the predetermined time period; a boiler combustion data encoding module 120, which is used to feature encode the temperature value of the boiler furnace mouth and the oxygen content value in the boiler at the multiple predetermined time points, as well as the monitoring video of the boiler combustion within the predetermined time period to obtain a control feature matrix; a control result generation module 130, which is used to determine whether it is necessary to adjust the opening degree of the air damper in the boiler based on the control feature matrix.

[0037] Specifically, in the technical solution of the present application, first, the temperature value of the boiler furnace mouth at multiple predetermined time points within a predetermined time period and the oxygen content value in the boiler and the monitoring video of the boiler combustion within the predetermined time period are obtained. It should not be difficult to understand that the purpose of this step is to collect key data and video information during the operation of the boiler to provide it for subsequent analysis and processing. Specifically, the temperature value and the oxygen content value are important indicators of the boiler operation state, which can reflect the effect of boiler combustion and the oxygen supply during the combustion process. The temperature value is also a key indicator of the risk of coal ash coking. When the temperature inside the boiler exceeds the melting point of the coal ash, the coal ash will melt and adhere to the surface of the pipes and equipment inside the boiler to form coking materials. These coking materials will cause poor heat conduction, obstruction of flue gas flow, and even blockage of the pipeline, affecting the normal operation of the boiler. The oxygen content is related to the melting point of the coal ash, which can help evaluate the tendency of coal ash coking. When the oxygen content in the boiler is insufficient, a large amount of reducing gas will be generated inside the boiler, and the melting point of the coal ash will also decrease, thereby affecting the coking of the boiler. By collecting these data at multiple predetermined time points, the time series information of the boiler operation status can be obtained for subsequent correlation analysis and feature extraction. At the same time, collecting monitoring videos of boiler combustion can provide more intuitive and comprehensive information. The actual situation of boiler combustion can be observed in the video, including the intensity of coal combustion, flame shape, coal ash generation and movement, etc. By sampling and extracting key frames from the video, the key moments of the boiler combustion process can be obtained for subsequent combustion process coding and coking tracking analysis.

[0038] In an embodiment of the present application, an implementation method for obtaining the temperature value of the boiler furnace mouth and the oxygen content value in the boiler at multiple predetermined time points within a predetermined time period and a monitoring video of the boiler combustion within the predetermined time period may be: 1. Acquisition of temperature value and oxygen content value: a. Install a temperature sensor at the boiler furnace mouth to measure the temperature of the furnace mouth; at the same time, install an oxygen sensor at an appropriate position inside the boiler to measure the oxygen content in the boiler; b: Use a data acquisition device to connect the temperature sensor and the oxygen sensor, and set a suitable sampling frequency and sampling interval, such as a frequency of every 1 minute. The data acquisition device may be a special data acquisition device or a data acquisition module that is compatible with the sensor; c: Record and store the collected temperature value and oxygen content value, and real-time recording and storage may be performed using data acquisition software connected to a computer system. 2. Acquisition of boiler combustion monitoring video: a. Install a monitoring camera in the boiler combustion area to monitor the boiler combustion process in real time. It is worth noting that the camera should be installed in a position that can fully observe the boiler combustion area and ensure clear image quality; b. The video acquisition device can be a special video acquisition card or a video acquisition module that matches the camera; c. Store the acquired monitoring video, and use video acquisition software connected to the computer system for real-time storage.

[0039] Figure 2 FIG. 1 is a block diagram of a boiler combustion data encoding module in an intelligent collection and monitoring system for preventing boiler coking according to an embodiment of the present application. Figure 2 As shown, it includes: a video data encoding unit 121, which is used to extract the coking process features of the monitoring video of the boiler combustion within the predetermined time period to obtain a boiler coking tracking feature matrix; a numerical data encoding unit 122, which is used to extract the coking factor features of the temperature values ​​of the boiler furnace mouth and the oxygen content values ​​in the boiler at the predetermined time points to obtain a coking influence factor feature matrix; a feature fusion unit 123, which is used to fuse the coking influence factor feature matrix and the boiler coking tracking feature matrix to obtain the control feature matrix.

[0040] Figure 3 FIG. 1 is a block diagram of a video data encoding unit in an intelligent acquisition and monitoring system for preventing boiler coking according to an embodiment of the present application. Figure 3 As shown, it includes: a key frame extraction subunit 1211, which is used to sample the monitoring video of the boiler combustion within the predetermined time period to obtain a plurality of boiler combustion monitoring key frames; a boiler coking process feature generation subunit 1212, which is used to input the plurality of boiler combustion monitoring key frames into a combustion process encoder based on a temporal attention mechanism to obtain a boiler coking tracking feature map; a feature map dimensionality reduction subunit 1213, which is used to pool the boiler coking tracking feature map along the channel dimension to obtain the boiler coking tracking feature matrix.

[0041] The boiler combustion process is a continuous dynamic process, and the monitoring video records various changes in the combustion process. However, in order to effectively analyze and process video data, it is usually not necessary to process the entire video, but to select key frames at key moments for sampling. Through sampling, the video data can be compressed into a key frame sequence to reduce the amount of data stored and processed. A key frame is a frame with important information or significant changes in the video sequence, which can represent the key state of the entire combustion process. That is, the monitoring video of the boiler combustion within the predetermined time period is sampled to obtain a plurality of boiler combustion monitoring key frames.

[0042] In an embodiment of the present application, a possible implementation method for sampling the monitoring video of the boiler combustion within the predetermined time period to obtain multiple boiler combustion monitoring key frames may be: sampling the monitoring video of the boiler combustion within the predetermined time period at a predetermined sampling frequency to obtain the multiple boiler combustion monitoring key frames. It is worth mentioning that the predetermined sampling frequency here is not always fixed, but depends on the specific usage scenario. The frequency of sampling the monitoring video needs to be the same as the difference between two adjacent time points of the temperature value and the oxygen content value at multiple predetermined time points.

[0043] In order to extract and encode the key information of the combustion process from the boiler combustion monitoring key frame sequence, in the technical solution of the present application, the multiple boiler combustion monitoring key frames are input into the combustion process encoder based on the time attention mechanism to obtain the boiler coking tracking feature map. In the technical solution of the present application, the essence of the combustion process encoder based on the time attention mechanism is the convolutional neural network model using the time attention mechanism. The key information of the boiler combustion process is usually concentrated in the key moments of the combustion process, which are very important for tracking and monitoring the coking problem. As mentioned above, the key frame sequence is a representative and important frame selected from the entire video sequence, which can capture the key states and changes in the combustion process. Ordinary technicians in this field should know that the time attention mechanism is a machine learning method for automatically paying attention to and capturing important time steps in time series data. Multiple boiler combustion monitoring key frames are input into the combustion process encoder based on the time attention mechanism, and a series of attention-weighted key frame features can be obtained. These key frame features can be combined into a boiler coking tracking feature map, which can be used to represent the coking situation in the combustion process. The boiler coking tracking feature map can provide information about the time evolution of coking in the boiler combustion process, helping to monitor the development trend and degree of the boiler coking problem.

[0044] In an embodiment of the present application, an implementation method of inputting the multiple boiler combustion monitoring key frames into a combustion process encoder based on a time attention mechanism to obtain a boiler coking tracking feature map may be: extracting adjacent first boiler combustion monitoring key frames and second boiler combustion monitoring key frames from the multiple boiler combustion monitoring key frames; passing the first boiler combustion monitoring key frame and the second boiler combustion monitoring key frame through the first convolution layer and the second convolution layer of the combustion process encoder respectively to obtain a first boiler combustion convolution feature map and a second boiler combustion convolution feature map; calculating the position point multiplication between the first boiler combustion convolution feature map and the second boiler combustion convolution feature map to obtain a boiler combustion time attention map; inputting the boiler combustion time attention map into a Softmax activation function to obtain a boiler combustion time attention feature map; passing the second boiler combustion monitoring key frame through the third convolution layer of the combustion process encoder to obtain a third boiler combustion convolution feature map; calculating the position point multiplication between the third boiler combustion convolution feature map and the boiler combustion time attention feature map to obtain a boiler combustion time attention feature map corresponding to the second boiler combustion monitoring key frame.

[0045] In order to reduce the dimension of the boiler coking tracking feature map, extract a more compact and representative feature representation, and facilitate the subsequent analysis and processing of the coking problem, it is necessary to perform a dimensionality reduction operation on the boiler coking tracking feature map. That is, the boiler coking tracking feature map is pooled along the channel dimension to obtain a boiler coking tracking feature matrix. It should be well understood that the boiler coking tracking feature map is usually a three-dimensional tensor with two spatial dimensions (such as height and width) and a channel dimension (representing different feature channels). By pooling the channel dimension of the feature map, the features of each channel are statistically summarized to obtain a lower dimensional feature representation. In an embodiment of the present application, pooling can select maximum pooling or average pooling. Maximum pooling selects the maximum value in each channel as a representative feature, while average pooling calculates the average value of the features in each channel. Pooling operations can reduce the dimension of the feature map, remove redundant information, and extract more important and representative features.

[0046] Figure 4 FIG. 1 is a block diagram of a numerical data encoding unit in an intelligent data collection and monitoring system for preventing boiler coking according to an embodiment of the present application. Figure 4 As shown, it includes: a numerical data structuring subunit 1221, which is used to arrange the temperature values ​​of the boiler furnace mouth and the oxygen content values ​​in the boiler at the multiple predetermined time points into a temperature time series input vector and an oxygen content time series input vector according to the time dimension; a numerical data association subunit 1222, which is used to associate the temperature time series input vector with the oxygen content time series input vector to obtain a temperature-oxygen content association matrix; a coking influence factor feature extraction subunit 1223, which is used to pass the temperature-oxygen content association matrix through a coking influence factor feature extractor with a block structure to obtain the coking influence factor feature matrix.

[0047] The combustion process of the boiler is a process that changes over time. Temperature and oxygen content, as key parameters, will fluctuate over time. In order to more conveniently capture the dynamic changes of the boiler combustion process, in the technical solution of the present application, the temperature values ​​of the boiler furnace mouth at the multiple predetermined time points and the oxygen content values ​​in the boiler are arranged as temperature time series input vectors and oxygen content time series input vectors according to the time dimension. The temperature time series input vector is a vector composed of temperature values ​​at multiple time points, wherein the temperature value at each time point represents the boiler furnace mouth temperature at that time point. Such a time series can reflect the temperature change trend and volatility, helping us understand the thermal state changes of the boiler. Similarly, the oxygen content time series input vector is a vector composed of oxygen content values ​​at multiple time points, wherein the oxygen content value at each time point represents the oxygen supply in the boiler at that time point. Such a time series can reveal the change trend and volatility of oxygen content, helping us understand the oxygen supply status of the boiler. By arranging these parameters as time series input vectors according to the time dimension, the information of the time series can be introduced into the model, so as to better analyze and predict the combustion state and coking risk of the boiler.

[0048] In the boiler combustion process, temperature and oxygen content are two important parameters, which have a direct impact on the melting point and coking risk of coal ash. The increase in temperature can cause the melting point of coal ash to decrease, thereby increasing the risk of coking. At the same time, changes in oxygen content will also affect the melting point of coal ash. Too high or too low oxygen content may lead to the occurrence of coking problems. Therefore, in the technical solution of the present application, the temperature time series input vector and the oxygen content time series input vector are associated to obtain a temperature-oxygen content association matrix. By correlating and analyzing the temperature and oxygen content, a more comprehensive understanding of their interaction and the comprehensive impact on coal ash coking can be obtained.

[0049] In an embodiment of the present application, a possible implementation method of associating the temperature time series input vector and the oxygen content time series input vector to obtain a temperature-oxygen content association matrix may be: calculating the temperature-oxygen content association matrix between the temperature time series input vector and the oxygen content time series input vector using the following association coding formula; wherein the association coding formula is:

[0050]

[0051] Where V a represents the temperature timing input vector, represents the transposed vector of the temperature timing input vector, V b represents the oxygen content time series input vector, M represents the temperature-oxygen content association matrix, Represents vector multiplication.

[0052] During the boiler combustion process, changes in temperature and oxygen content have a direct impact on the coking risk. However, the direct analysis of the temperature-oxygen content association matrix may be limited by the large matrix dimension, and there may be complex nonlinear relationships between the elements in the matrix. In order to extract the relevant features of the coking risk from the association matrix, the temperature-oxygen content association matrix is ​​passed through a coking influence factor feature extractor with a block structure to obtain a coking influence factor feature matrix. In the technical solution of the present application, the essence of the coking influence factor feature extractor with a block structure is a convolutional neural network model with a block structure. The convolutional neural network model with a block structure can effectively capture the local relationships and patterns in the matrix. It identifies and extracts features in the matrix through convolution operations and pooling operations. In this case, the size and stride of the convolution kernel can be designed according to the characteristics of the association matrix to capture the local relationship and change pattern between temperature and oxygen content. Through the coking influence factor feature extractor, features with coking risk correlation can be extracted from the temperature-oxygen content association matrix.

[0053] In an embodiment of the present application, a possible implementation method of passing the temperature-oxygen content correlation matrix through a coking influence factor feature extractor with a block structure to obtain a coking influence factor feature matrix can be: 1. Constructing a coking influence factor feature extractor: defining a convolutional neural network model with a block structure, including a convolution layer and a pooling layer; designing the size and stride of the convolution kernel to adapt to the characteristics of the correlation matrix; determining the number and size of the convolution layer and the pooling layer to adapt to task requirements and computing resources; adding appropriate nonlinear activation functions and regularization methods, such as ReLU activation function and Dropout regularization. 2. Feature extraction: taking the temperature-oxygen content correlation matrix as input, forward propagating through the coking influence factor feature extractor, in the convolution layer, using convolution operations to capture local features and patterns in the correlation matrix; in the pooling layer, using pooling operations to reduce feature dimensions and retain key information. 3. Generation of the coking influence factor feature matrix: The convolved and pooled feature maps are connected into a feature matrix; fully connected layers and activation functions can be further applied to extract more advanced features; finally, the coking influence factor feature matrix is ​​output.

[0054] The coking influencing factor feature matrix is ​​the coking risk related features extracted from the temperature-oxygen content association matrix by a coking influencing factor feature extractor with a block structure. These features reflect the mutual influence between temperature and oxygen content and the correlation of coking risk. The boiler coking tracking feature matrix is ​​a feature matrix obtained by monitoring and tracking the operating status of the boiler. It reflects the coking situation during the combustion process. In order to comprehensively consider the coking influencing factor characteristics and the boiler status characteristics, in the technical solution of the present application, the coking influencing factor feature matrix and the boiler coking tracking feature matrix are multiplied to obtain the control feature matrix.

[0055] Figure 5 FIG. 1 is a block diagram of a control result generation module in an intelligent collection and monitoring system for preventing boiler coking according to an embodiment of the present application. Figure 5 As shown, the control result generation module 130 includes: a feature matrix dimensionality reduction unit 131, used to expand the control feature matrix to obtain a control feature vector; a feature optimization unit 132, used to perform local distribution equilibrium adaptive correction on the control feature vector based on subclass label related metrics to obtain an optimized control feature vector; a classification unit 133, used to pass the optimized control feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether it is necessary to adjust the opening degree of the air damper in the boiler.

[0056] In order to improve computing efficiency and save computing resources, the control feature matrix needs to be expanded in the technical solution of the present application to obtain the control feature vector. It should be understood that the operation of the two-dimensional feature matrix is ​​more complicated than the one-dimensional vector operation, so the required computing time will be more. By expanding the feature matrix, the dimensionality of the data is reduced, and during the expansion process, the key feature information contained in the feature matrix is ​​not lost, and will not affect the classification result.

[0057] In particular, in the technical solution of the present application, it is considered that the temperature value and the internal oxygen content value at the furnace mouth of the boiler may have inconsistent data quality due to factors such as sensor accuracy, installation location, and environmental interference. The boiler combustion process is affected by many factors, such as fuel type, quality, air intake, etc. Changes in these factors may cause inconsistent local distribution of coking tracking feature maps under different conditions. The combustion process encoder based on the time attention mechanism may be more sensitive to certain features in the video key frames, while ignoring other equally important information, resulting in differences in the local distribution of feature maps. The coking influencing factor feature extractor with a block structure may not be able to fully capture all subtle features related to coking, especially when coking develops slowly or is not significant. When multiplying the coking influencing factor feature matrix with the boiler coking tracking feature matrix to obtain the control feature matrix, the weights and importance of different features may not be properly processed, which may lead to inconsistent local distribution of the control feature vector. The control feature vector has local distribution differences, which leads to the problem that when it is classified by the classifier, there will be a class harmony offset relative to the predetermined class label of the classifier, affecting the classification iteration effect. Therefore, in the technical solution of the present application, the control feature vector is subjected to local distribution balancing adaptive correction based on subclass label related metrics.

[0058] Specifically, in an embodiment of the present application, the control feature vector is subjected to a local distribution equalization adaptive correction based on a subclass label related metric to obtain an optimized control feature vector, including: determining a weight learning parameter matrix of the classifier; extracting each column vector in the weight learning parameter matrix to obtain a set of weight learning parameter column vectors; and using the following formula to respectively calculate a metric correlation prediction factor between each weight learning parameter column vector in the set of weight learning parameter column vectors and the control feature vector to obtain a metric correlation prediction factor modulation vector, wherein the formula is:

[0059]

[0060] Among them, V i represents each weight learning parameter column vector in the set of weight learning parameter column vectors, V c represents the control feature vector, It means subtracting by position. is the superposition norm of the two norms of the eigenvectors, ‖·‖ F is the Frobenius norm of the eigenvector, M p represents the weight learning parameter matrix, exp(·) represents the exponential operation with the natural constant e as the base, V pRepresenting the metric correlation prediction factor modulation vector; taking the metric correlation prediction factor modulation vector as the weight vector, calculating the position point multiplication between the metric correlation prediction factor modulation vector and the control feature vector to obtain the optimized control feature vector.

[0061] In the technical solution of the present application, it is considered that there are local distribution differences in the control feature vector, which leads to the problem that when it is classified by the classifier, there will be a class harmony offset relative to the predetermined class label of the classifier, affecting the classification iteration effect. Based on this, in the technical solution of the present application, the control feature vector is subjected to local distribution balancing adaptive correction based on the subclass label related metric, which uses the superposition paradigm of the two norms between each column vector in the weight learning parameter matrix of the classifier and the control feature vector as the subclass label related metric integral factor and the matrix norm of the weight learning parameter of the classifier as the differential factor, and then uses the association paradigm between the subclass label related metric integral factor and the differential factor to perform bidirectional linear manifold modulation on the control feature vector, thereby improving the consistency of the local feature distribution of the control feature vector by adaptively stretching and compressing the control feature vector, thereby reducing the harmony offset of the control feature vector relative to the predetermined class label of the classifier, thereby improving the accuracy of the classification result.

[0062] Finally, the optimized control feature vector is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether it is necessary to adjust the opening degree of the damper in the boiler. It is not difficult to understand that the opening degree of the damper in the boiler is an important parameter for adjusting the oxygen supply during the combustion process of the boiler. By adjusting the opening degree of the damper, the oxygen content in the boiler combustion process can be controlled, thereby affecting the coking risk. When the coking risk in the boiler combustion process is high, the oxygen supply can be increased by increasing the opening degree of the damper to reduce the coking risk; when the coking risk is low, the oxygen supply can be reduced by reducing the opening degree of the damper to improve the combustion efficiency. The classifier is a machine learning model that can learn and predict the association between the characteristic value and the coking risk during the boiler combustion process. Based on the output results of the classifier, the operator can adjust the opening degree of the damper in the boiler in time to achieve optimal control of the boiler combustion process and reduce the coking risk. Such an intelligent collection and monitoring method can improve the operating efficiency and reliability of the boiler, reduce the occurrence of coking problems, and thus improve the power production efficiency and the life of the equipment.

[0063] To summarize, an intelligent data collection and monitoring system 100 for preventing boiler coking based on an embodiment of the present application is explained, which analyzes the monitoring video of boiler combustion and the oxygen content values ​​in the boiler and the temperature values ​​of the boiler furnace mouth at multiple predetermined time points to obtain the relationship between boiler coking and the temperature value and the oxygen content value, and then determines whether it is necessary to adjust the opening degree of the damper in the boiler. By controlling the opening degree of the damper, the oxygen content in the boiler is controlled, thereby ensuring that the temperature of the boiler furnace mouth is slightly lower than the melting point of the coal ash, thereby reducing the coking degree of the boiler.

[0064] As described above, the intelligent data collection and monitoring system 100 for preventing boiler coking according to the embodiment of the present application can be implemented in various terminal devices, such as a server for intelligent data collection and monitoring for preventing boiler coking. In one example, the intelligent data collection and monitoring system 100 for preventing boiler coking according to the embodiment of the present application can be integrated into the terminal device as a software module and / or a hardware module. For example, the intelligent data collection and monitoring system 100 for preventing boiler coking can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the intelligent data collection and monitoring system 100 for preventing boiler coking can also be one of the many hardware modules of the terminal device.

[0065] Alternatively, in another example, the intelligent data collection and monitoring system 100 for preventing boiler coking and the terminal device may also be separate devices, and the intelligent data collection and monitoring system 100 for preventing boiler coking may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0066] Figure 6 FIG. 1 is a flow chart of an intelligent data collection and monitoring method for preventing boiler coking according to an embodiment of the present application. Figure 6 As shown, in the intelligent data collection and monitoring method for preventing boiler coking, the method includes: S110, obtaining the temperature value of the boiler furnace mouth and the oxygen content value in the boiler at multiple predetermined time points within a predetermined time period, as well as the monitoring video of the boiler combustion within the predetermined time period; S120, feature encoding the temperature value of the boiler furnace mouth and the oxygen content value in the boiler at the multiple predetermined time points, as well as the monitoring video of the boiler combustion within the predetermined time period to obtain a control feature matrix; S130, judging whether it is necessary to adjust the opening degree of the air damper in the boiler based on the control feature matrix.

[0067] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned intelligent collection and monitoring method for preventing boiler coking have been referred to above. Figures 1 to 5 The description of the intelligent collection and monitoring system for preventing boiler coking has been introduced in detail, and therefore, its repeated description will be omitted.

[0068] To summarize, an intelligent data collection and monitoring method for preventing boiler coking based on an embodiment of the present application is explained, which analyzes the monitoring video of boiler combustion and the oxygen content values ​​in the boiler and the temperature values ​​of the boiler furnace mouth at multiple predetermined time points to obtain the relationship between boiler coking and the temperature value and the oxygen content value, and then determines whether it is necessary to adjust the opening degree of the damper in the boiler. By controlling the opening degree of the damper, the oxygen content in the boiler is controlled, thereby ensuring that the temperature of the boiler furnace mouth is slightly lower than the melting point of the coal ash, thereby reducing the coking degree of the boiler.

[0069] It should be understood that, for the sake of clarity, embodiments of the present application are described with reference to different functional units. However, it will be apparent that, without departing from the present application, the functionality of each functional unit can be implemented in a single unit, in multiple units or as a part of other functional units. For example, the functionality that is described as being performed by a single unit can be performed by multiple different units. Therefore, reference to a specific functional unit is only considered as a reference to the appropriate unit for providing the described functionality, rather than indicating a strict logical or physical structure or organization. Therefore, the present application can be implemented in a single unit, or can be physically and functionally distributed between different units and circuits.

[0070] Although the present application has been described in conjunction with some embodiments, it is not intended to be limited to the specific form set forth herein. On the contrary, the scope of the present application is limited only by the appended claims. Additionally, although individual features may be included in different claims, these may possibly be advantageously combined, and the inclusion in different claims does not imply that the combination of features is not feasible and / or advantageous. The order of features in the claims does not imply any specific order in which the features must work. In addition, in the claims, the word "comprising" does not exclude other elements, and the term "one" or "an" does not exclude a plurality.

Claims

1. An intelligent collection and monitoring system for preventing boiler coking, characterized in that: include: A boiler combustion data acquisition module is used to obtain the temperature value of the boiler furnace port and the oxygen content value in the boiler at multiple predetermined time points within a predetermined time period, as well as a monitoring video of the boiler combustion within the predetermined time period; A boiler combustion data encoding module, used for performing feature encoding on the temperature values ​​of the boiler furnace port and the oxygen content value in the boiler at the plurality of predetermined time points and the monitoring video of the boiler combustion within the predetermined time period to obtain a control feature matrix; The control result generation module is used to determine whether it is necessary to adjust the opening degree of the air door in the boiler based on the control characteristic matrix.

2. The intelligent data collection and monitoring system for preventing boiler coking according to claim 1 is characterized in that: The boiler combustion data encoding module comprises: A video data encoding unit, used for extracting coking process features from the monitoring video of boiler combustion within the predetermined time period to obtain a boiler coking tracking feature matrix; A numerical data encoding unit, used for extracting coking factor characteristics from the temperature values ​​of the boiler furnace port and the oxygen content values ​​in the boiler at the plurality of predetermined time points to obtain a coking influencing factor characteristic matrix; A feature association unit is used to multiply the coking influencing factor feature matrix and the boiler coking tracking feature matrix to obtain the control feature matrix.

3. The intelligent data collection and monitoring system for preventing boiler coking according to claim 2 is characterized in that: The video data encoding unit comprises: A key frame extraction subunit, used for sampling the monitoring video of boiler combustion within the predetermined time period to obtain a plurality of boiler combustion monitoring key frames; A boiler coking process feature generation subunit, configured to input the plurality of boiler combustion monitoring key frames into a combustion process encoder based on a temporal attention mechanism to obtain a boiler coking tracking feature map; The feature map dimensionality reduction subunit is used to pool the boiler coking tracking feature map along the channel dimension to obtain the boiler coking tracking feature matrix.

4. The intelligent data collection and monitoring system for preventing boiler coking according to claim 3 is characterized in that: The key frame extraction subunit is used to: sample the monitoring video of boiler combustion within the predetermined time period at a predetermined sampling frequency to obtain the plurality of boiler combustion monitoring key frames.

5. The intelligent data collection and monitoring system for preventing boiler coking according to claim 4 is characterized in that: The numerical data encoding unit comprises: A numerical data structuring subunit, for arranging the temperature values ​​of the boiler furnace port and the oxygen content values ​​in the boiler at the plurality of predetermined time points into a temperature time series input vector and an oxygen content time series input vector according to the time dimension respectively; A numerical data association subunit, used for associating the temperature time series input vector with the oxygen content time series input vector to obtain a temperature-oxygen content association matrix; The coking influencing factor feature extraction subunit is used to obtain the coking influencing factor feature matrix by passing the temperature-oxygen content correlation matrix through a coking influencing factor feature extractor with a block structure.

6. The intelligent data collection and monitoring system for preventing boiler coking according to claim 5 is characterized in that: The numerical data association subunit is used to: calculate the temperature-oxygen content association matrix between the temperature time series input vector and the oxygen content time series input vector using the following association coding formula; Wherein, the association coding formula is: Where V a represents the temperature timing input vector, represents the transposed vector of the temperature timing input vector, V b represents the oxygen content time series input vector, M represents the temperature-oxygen content association matrix, Represents vector multiplication.

7. The intelligent data collection and monitoring system for preventing boiler coking according to claim 6 is characterized in that: The control result generating module comprises: A feature matrix dimension reduction unit, used for expanding the control feature matrix to obtain a control feature vector; A feature optimization unit, used for performing local distribution equilibrium adaptive correction on the control feature vector based on subclass label related metrics to obtain an optimized control feature vector; The classification unit is used to pass the optimized control feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether it is necessary to adjust the opening degree of the air damper in the boiler.

8. The intelligent data collection and monitoring system for preventing boiler coking according to claim 7 is characterized in that: The feature optimization unit is used to: Determining a weight learning parameter matrix of the classifier; Extracting each column vector in the weight learning parameter matrix to obtain a set of weight learning parameter column vectors; The metric correlation prediction factor between each weight learning parameter column vector in the set of the weight learning parameter column vector and the control feature vector is calculated by the following formula to obtain a metric correlation prediction factor modulation vector, wherein the formula is: Among them, V i represents each weight learning parameter column vector in the set of weight learning parameter column vectors, V c represents the control feature vector, It means subtracting by position. is the superposition norm of the two norms of the eigenvectors, ‖·‖ F is the Frobenius norm of the eigenvector, M p represents the weight learning parameter matrix, exp(·) represents the exponential operation with the natural constant e as the base, V p A modulation vector representing a predictor of the metric relevance; The metric correlation prediction factor modulation vector is used as a weight vector, and the position point multiplication between the metric correlation prediction factor modulation vector and the control feature vector is calculated to obtain the optimized control feature vector.

9. An intelligent collection and monitoring method for preventing boiler coking, characterized in that: include: Obtaining the temperature value of the boiler furnace port and the oxygen content value in the boiler at multiple predetermined time points within a predetermined time period and the monitoring video of the boiler combustion within the predetermined time period; Performing feature coding on the temperature values ​​of the boiler furnace port and the oxygen content values ​​in the boiler at the plurality of predetermined time points and the monitoring video of the boiler combustion within the predetermined time period to obtain a control feature matrix; Based on the control characteristic matrix, it is determined whether the opening degree of the air damper in the boiler needs to be adjusted.

10. The intelligent data collection and monitoring method for preventing boiler coking according to claim 9, characterized in that: The temperature values ​​of the boiler furnace port and the oxygen content values ​​in the boiler at the plurality of predetermined time points and the monitoring video of the boiler combustion within the predetermined time period are feature-encoded to obtain a control feature matrix, including: Extracting coking process features from the monitoring video of boiler combustion within the predetermined time period to obtain a boiler coking tracking feature matrix; Extracting coking factor characteristics from the temperature values ​​of the boiler furnace port and the oxygen content values ​​in the boiler at the plurality of predetermined time points to obtain a coking influencing factor characteristic matrix; The coking influencing factor characteristic matrix and the boiler coking tracking characteristic matrix are integrated to obtain the control characteristic matrix.

Images