Method and system for monitoring coal cinder cooling process

Abstract

The invention provides a coal cinder cooling process monitoring method and system, and belongs to the technical field of industrial automation and intelligent monitoring. According to the system, multiple sensors are arranged in a hearth of a coal cinder cooling section of a power plant boiler, and multi-dimensional data such as the thickness, temperature, air temperature, air pressure, air volume and hearth images of slag are collected. After data is subjected to filtering processing and normalization processing, visual features are extracted by using a convolutional neural network in combination with an attention mechanism, time sequence features are extracted by using a long-short-term memory network in combination with the attention mechanism, and multi-source features are fused to form comprehensive features. And the real-time state of the cooling process is visually presented through a data display module combining augmented reality and virtual reality technologies. And finally, an operation control module based on a self-adaptive control algorithm and a reinforcement learning algorithm monitors and intelligently adjusts the cooling process in real time. Accurate monitoring and intelligent regulation and control of the coal cinder cooling process can be achieved, the cooling efficiency and stability are improved, and the labor cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial automation and intelligent monitoring, and in particular to a monitoring method and system for a coal slag cooling process. Background Art

[0002] In coal-fired power plants, the ash cooling process is an important part of boiler operation. Traditional ash cooling monitoring mainly relies on a single sensor (such as a temperature sensor or a pressure sensor) to obtain limited operating data and is regulated by manual experience. However, the existing technology has the following shortcomings:

[0003] Single monitoring method: Traditional methods can only obtain local data, such as temperature or pressure, and cannot fully reflect the multi-dimensional state of slag cooling in the furnace (such as slag thickness, cooling air parameters, furnace images, etc.), resulting in incomplete monitoring information.

[0004] Lack of real-time and accuracy: The existing monitoring system is unable to process and analyze complex data in real time, and it is difficult to accurately capture the dynamic changes in the cooling process and detect potential problems in a timely manner, which can easily lead to low cooling efficiency or equipment failure.

[0005] Control relies on manual experience: The control of the traditional cooling process mainly relies on manual experience and lacks intelligent automatic control methods. This model is not only inefficient, but also difficult to achieve refined management, and it is easy to cause control errors due to human factors.

[0006] Lack of visual display: The existing technology lacks intuitive visualization means, making it difficult for operators to quickly and comprehensively understand the real-time status of the cooling process, increasing the difficulty of operation and the risk of misjudgment.

[0007] Inability to self-optimize: The existing system is unable to self-learn and optimize based on historical data and real-time monitoring results, making it difficult to adapt to complex working conditions and unable to achieve long-term stable and efficient operation.

[0008] Therefore, a method and system for monitoring the slag cooling process are urgently needed to solve the above problems. Summary of the invention

[0009] The purpose of the present invention is to provide a monitoring method and system for the coal slag cooling process, which can realize accurate monitoring and intelligent regulation of the coal slag cooling process, improve cooling efficiency and stability, and reduce labor costs.

[0010] To achieve the above object, the present invention provides a method for monitoring a slag cooling process, comprising the following steps:

[0011] Step S1, arranging a variety of sensors in the furnace of the slag cooling section of the power plant boiler, including a three-dimensional laser scanner, an infrared thermal imaging array sensor, an integrated wind parameter sensor and a camera, for acquiring data in the furnace;

[0012] Step S2: filtering the collected furnace data and removing abnormal values ​​through a data processing module;

[0013] Step S3: Perform advanced analysis on the collected data in the data processing module using deep learning algorithms and machine learning models, including:

[0014] Normalizing the acquired furnace data;

[0015] The convolutional neural network (CNN) is used in combination with the attention mechanism to analyze the image data in the furnace and extract visual features.

[0016] The long short-term memory network LSTM is used in combination with the attention mechanism to analyze the slag temperature, wind temperature, wind pressure, and wind volume to extract time series features;

[0017] Further integrate visual features, time series features, and initially integrated features to form comprehensive features;

[0018] Step S4, displaying comprehensive features through a data display module combining augmented reality AR and virtual reality VR;

[0019] Step S5: Based on the comprehensive characteristics displayed, the slag cooling process is monitored and adjusted in real time in the operation control module in combination with the adaptive control algorithm and the reinforcement learning algorithm.

[0020] Preferably, in step S1, the data inside the furnace include: slag thickness obtained by a three-dimensional laser scanner; slag temperature obtained by an infrared thermal imaging array sensor; wind temperature, wind pressure and air volume obtained by an integrated wind parameter sensor; and image data inside the furnace obtained by a camera.

[0021] Preferably, in step S2, the calculation formula for filtering is as follows:

[0022]

[0023] Among them, y z represents the filtered data, N represents the size of the sliding window, x k represents the sensor data at time k, and z represents the index variable;

[0024] The calculation formula for outlier removal is as follows:

[0025] if|x k -μ|>3σ, remove x k ;

[0026] Where μ represents the mean of the sensor data and σ represents the standard deviation of the sensor data.

[0027] Preferably, in step S3, the acquired furnace data is normalized, including:

[0028]

[0029] Among them, y (s) represents the normalized features of the sth sensor data, x (s) represents the raw data of the sth sensor data, μ (s) , σ (s) represent the mean and standard deviation of the sth sensor data respectively.

[0030] Preferably, in step S3, a convolutional neural network (CNN) is used in combination with an attention mechanism to analyze the image data in the furnace to extract visual features, including:

[0031] F vis =CNN att (y (1) );

[0032] Among them, F vis Representing visual features, CNN att represents the CNN model with the attention mechanism introduced, y (1) Represents the normalized image data.

[0033] Preferably, in step S3, the slag temperature, wind temperature, wind pressure, and wind volume are analyzed by using a long short-term memory network LSTM in combination with an attention mechanism to extract time series features, including:

[0034] F temp =LSTM att (y (2) ,y (3) ,y (4) ,y (5) ,);

[0035] Among them, F temp Represents time series features, LSTM att represents the LSTM model with the attention mechanism introduced, y (2) ,y (3) ,y (4) ,y (5) Represents the normalized slag temperature, wind temperature, wind pressure and air volume.

[0036] Preferably, in step S3, the visual features, the time series features, and the initially fused features are further fused to form comprehensive features, including:

[0037]

[0038] Among them, ω vis ,ω temp ,ω pre Respectively represent the weights of visual features, time series features, and initial fusion features, F pre It represents the initial fusion feature, and its calculation formula is as follows:

[0039] F pre =concat(y (1) ,y (2) ,y (3) ,y (4) ,y (5) ).

[0040] Preferably, in step S4, the comprehensive features are displayed by a data display module combining augmented reality AR and virtual reality VR, including:

[0041]

[0042] Among them, color(F fuse ) represents the color mapping function, F low Indicates the lower limit of the healthy state of the cooling process, F high Indicates the upper threshold of the health status of the cooling process.

[0043] Preferably, in step S5, the slag cooling process is monitored and adjusted in real time in the operation control module in combination with the adaptive control algorithm and the reinforcement learning algorithm according to the comprehensive characteristics of the display, including:

[0044] Use adaptive control algorithms to make initial adjustments to the cooling process:

[0045]

[0046] Where u(t) represents the control output at time t, K p , K i , K d Respectively represent proportional, integral, and differential gains, F target represents the expected comprehensive eigenvalue, It reflects the changing trend of error over time;

[0047] Introducing reinforcement learning algorithm to further optimize the control strategy:

[0048]

[0049] Among them, θ k+1 represents the updated policy parameters, π θrepresents the current strategy, κ represents a complete trajectory, T is the time step, γ represents the discount factor, r t Represents the reward function, the reward function r t The definition is as follows:

[0050] r t = -α(F fuse,t -F target ) 2 +βlogπ θ (a t |s t );

[0051] Among them, α and β represent the weight parameters in the reward function, F fuse,t represents the comprehensive eigenvalue at time t, π θ (a t |s t ) means in state s t A t probability.

[0052] Preferably, the slag cooling process is intelligently monitored and controlled using historical data and real-time monitoring results, including:

[0053]

[0054] Among them, ω i (t+1) represents the model weight at time t+1, ω i (t) represents the model weight at time t, represents the learning rate, Represents the weight ω i The partial derivative of Loss new (F fuse (t),F true (t)) represents the value of the new loss function at time t. The new loss function combines the current loss and the historical average loss. The calculation formula is as follows:

[0055]

[0056] Among them, F true represents the true feature value, λ represents the weight parameter, which is used to balance the impact of the current loss and the historical average loss, T' represents the time window length, j represents the index variable used in the summation process, which is used to traverse the time points in the time window, and F fuse (t) represents the features obtained by fusion of multi-source data at time t, F true (t) represents the true eigenvalue at time t, F fuse (j) represents the features obtained by fusion of multi-source data at time j, F true(j) represents the true eigenvalue at time j, Loss(F fuse (j),F true (j)) represents the loss function value calculated at time j.

[0057] The present invention also provides a monitoring system for the slag cooling process, comprising:

[0058] The sensor module is used to collect various data in the furnace of the slag cooling section of the power plant boiler. The sensor module includes: a three-dimensional laser scanner to obtain the thickness data of the slag; an infrared thermal imaging array sensor to obtain the temperature data of the slag; an integrated wind parameter sensor to obtain the wind temperature, wind pressure and wind volume data; a camera to obtain the image data in the furnace;

[0059] The data processing module is connected to the sensor module and is used to pre-process the collected data in the furnace, including filtering, outlier removal and normalization;

[0060] The data analysis module is connected to the data processing module and is used to perform advanced analysis on the pre-processed data, including: using the convolutional neural network CNN combined with the attention mechanism to analyze the image data and extract visual features; using the long short-term memory network LSTM combined with the attention mechanism to analyze the slag temperature, wind temperature, wind pressure and wind volume data and extract time series features; further integrating the visual features, time series features and the initially integrated features to form comprehensive features;

[0061] A data display module, connected to the data analysis module, is used to display comprehensive features by combining augmented reality AR and virtual reality VR technologies;

[0062] An operation control module, connected to the data analysis module and the data display module, is used to monitor and adjust the slag cooling process in real time according to the comprehensive characteristics of the display in combination with the adaptive control algorithm and the reinforcement learning algorithm;

[0063] The intelligent learning and optimization module is connected with the data analysis module and the operation control module, and is used to perform intelligent learning and optimization of the monitoring system based on historical data and real-time monitoring results.

[0064] Therefore, the present invention adopts the above-mentioned method and system for monitoring the slag cooling process, and the beneficial technical effects are as follows:

[0065] (1) Realize comprehensive monitoring of multi-dimensional data and improve information integrity:

[0066] Compared with traditional monitoring methods that mainly rely on a single sensor to obtain limited data, the present invention arranges multiple sensors including a three-dimensional laser scanner, an infrared thermal imaging array sensor, an integrated wind parameter sensor and a camera in the furnace of the slag cooling section of the power plant boiler, which can simultaneously collect multi-dimensional data such as slag thickness, temperature, wind temperature, wind pressure, air volume and furnace image. This improvement fully reflects the state of the slag cooling process, significantly improves the integrity and accuracy of the monitoring information, and is a major advancement in the existing technology.

[0067] (2) Enhance real-time performance and accuracy, and improve monitoring efficiency:

[0068] Traditional monitoring systems are unable to process and analyze complex data in real time, and it is difficult to accurately capture dynamic changes in the cooling process. However, the present invention achieves real-time advanced analysis of collected data through filtering, outlier removal, normalization, and the application of deep learning algorithms and machine learning models, accurately extracts visual features and time series features, and forms comprehensive features. This improvement significantly enhances the real-time and accuracy of monitoring, improves monitoring efficiency, and effectively avoids problems such as low cooling efficiency or equipment failure caused by untimely or inaccurate monitoring.

[0069] (3) Introduce intelligent control methods to reduce labor costs:

[0070] The control of the traditional cooling process mainly relies on manual experience and lacks intelligent automatic control means. The present invention realizes real-time monitoring and intelligent adjustment of the slag cooling process through the combination of adaptive control algorithm and reinforcement learning algorithm. This improvement not only reduces the dependence on manual experience, but also improves the intelligent level of control, effectively reduces labor costs, and improves the accuracy and efficiency of control.

[0071] (4) Use intuitive visual display to reduce the difficulty of operation:

[0072] Traditional monitoring systems lack intuitive visualization methods, making it difficult for operators to quickly and comprehensively understand the real-time status of the cooling process. However, the present invention combines augmented reality (AR) and virtual reality (VR) technologies to present comprehensive features to operators in an intuitive and easy-to-understand manner. This improvement significantly reduces the difficulty of operation and improves the convenience and accuracy of operation.

[0073] (5) The system has the ability to self-optimize and improve operational stability and efficiency:

[0074] Traditional monitoring systems cannot perform self-learning and optimization based on historical data and real-time monitoring results. However, the present invention enables the system to perform self-learning and optimization based on historical data and real-time monitoring results through the setting of intelligent learning and optimization modules, and continuously adapt to changes in working conditions. This improvement significantly improves the operating stability and efficiency of the system, and achieves long-term stable and efficient operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 It is a flow chart of a method for monitoring a slag cooling process of the present invention;

[0076] Figure 2 The present invention is a schematic structural diagram of a monitoring system for a coal slag cooling process. DETAILED DESCRIPTION

[0077] The technical solution of the present invention is further described below through the accompanying drawings and embodiments.

[0078] Unless otherwise defined, technical or scientific terms used in the present invention shall have the common meanings understood by one having ordinary skills in the field to which the present invention belongs.

[0079] Embodiment 1

[0080] like Figure 1 As shown, the present invention provides a flow chart of a method for monitoring a slag cooling process, comprising the following steps:

[0081] Step S1, arranging a variety of sensors in the furnace of the slag cooling section of the power plant boiler, including a three-dimensional laser scanner, an infrared thermal imaging array sensor, an integrated wind parameter sensor and a camera, for acquiring data in the furnace.

[0082] The installation positions of each sensor are as follows:

[0083] 3D laser scanner: Installed on the top of the furnace, used to measure the thickness of the slag in real time. The scanning frequency is set to once per minute, covering the entire bottom of the furnace, and can accurately obtain the thickness distribution data of the slag.

[0084] Infrared thermal imaging array sensor: installed on the side of the furnace, used to monitor the temperature distribution on the slag surface in real time. The sensor samples 10 times per second and can capture the temperature changes during the slag cooling process.

[0085] Integrated wind parameter sensor: installed at the entrance of the cooling air duct, used to measure the temperature, pressure and flow of the cooling air. The sampling frequency of the sensor is 5 times per second, which can reflect the parameter changes of the cooling air in real time.

[0086] Camera: Installed on the side of the furnace, used to obtain image data inside the furnace. The camera resolution is set to 1920×1080 pixels, and the sampling frequency is 1 frame per second, which is used to assist in monitoring the overall status of the furnace.

[0087] Through the collaborative work of these sensors, multi-dimensional data of the slag cooling process can be comprehensively collected, including slag thickness, temperature, cooling air parameters, and furnace images.

[0088] Step S2: The collected raw data usually contains noise and outliers, and needs to be preprocessed to improve the data quality. In this embodiment, the collected furnace data is filtered and outliers are removed through the data processing module.

[0089] The calculation formula for filtering is as follows:

[0090]

[0091] Among them, y z represents the filtered data, N represents the size of the sliding window, x k represents the sensor data at time k, and z represents the index variable;

[0092] The calculation formula for outlier removal is as follows:

[0093] if|x k -μ|>3σ, remove x k ;

[0094] Where μ represents the mean of the sensor data and σ represents the standard deviation of the sensor data.

[0095] Step S3: Use deep learning algorithms and machine learning models to perform advanced analysis on the collected data in the data processing module.

[0096] Normalize the acquired furnace data to extract valuable feature information, and normalize all sensor data to the [0,1] interval to eliminate the impact of different dimensions and ranges, including:

[0097]

[0098] Among them, y (s) represents the normalized features of the sth sensor data, x (s) represents the raw data of the sth sensor data, μ (s) , σ (s) represent the mean and standard deviation of the sth sensor data respectively.

[0099] The convolutional neural network (CNN) is used in combination with the attention mechanism to analyze the image data in the furnace and extract visual features, including:

[0100] F vis =CNN att (y (1) );

[0101] Among them, F vis Representing visual features, CNN att represents the CNN model with the attention mechanism introduced, y (1) Represents the normalized image data.

[0102] The long short-term memory network LSTM is combined with the attention mechanism to analyze the slag temperature, wind temperature, wind pressure, and wind volume, and extract time series features, including:

[0103] F temp =LSTM att (y (2) ,y (3) ,y (4) ,y (5) ,);

[0104] Among them, F temp Represents time series features, LSTM att represents the LSTM model with the attention mechanism introduced, y (2) ,y (3) ,y (4) ,y (5) Represents the normalized slag temperature, wind temperature, wind pressure and air volume.

[0105] The visual features, time series features, and initially fused features are further integrated to form comprehensive features. By assigning weights to different features and comprehensively considering image information and time series information, a comprehensive feature that fully reflects the cooling process state is finally obtained. The calculation formula is:

[0106]

[0107] Among them, ω vis ,ω temp ,ω pre Respectively represent the weights of visual features, time series features, and initial fusion features, F pre It represents the initial fusion feature, and its calculation formula is as follows:

[0108] F pre =concat(y (1) ,y (2) ,y (3) ,y (4) ,y (5) ).

[0109] Step S4: display comprehensive features through a data display module that combines augmented reality AR and virtual reality VR.

[0110] Data mapping: Map the comprehensive feature value to the color space and set the color threshold according to the health status of the cooling process. For example, when the comprehensive feature value is in the normal range, it is displayed in green; when it is close to the warning value, it is displayed in yellow; when it exceeds the warning value, it is displayed in red, as follows:

[0111]

[0112] Among them, color(F fuse ) represents the color mapping function, F low Indicates the lower limit of the healthy state of the cooling process, F high Indicates the upper threshold of the health status of the cooling process.

[0113] AR and VR display: VR devices (such as helmets) provide operators with an immersive view of the furnace interior, while AR technology is used to annotate key data and alarm information in real scenes. Operators can interact with the virtual interface through gestures or voice commands to view the cooling status of slag at different locations in real time.

[0114] Through the intuitive visualization of step S4, operators can more effectively understand and control the slag cooling process, thereby improving cooling efficiency and stability. At the same time, this visualization also provides necessary information support for the intelligent control in step S5, enabling the system to more accurately monitor and adjust in real time, and ultimately achieve intelligent management of the slag cooling process.

[0115] Step S5: Based on the comprehensive characteristics displayed, the slag cooling process is monitored and adjusted in real time in the operation control module in combination with the adaptive control algorithm and the reinforcement learning algorithm, including:

[0116] The proportional-integral-differential (PID) controller is used to make preliminary adjustments to the cooling process. The comprehensive characteristic value is used as the feedback signal, and the cooling air parameters are dynamically adjusted according to the set expected value and error change trend. The control formula is:

[0117]

[0118] Where u(t) represents the control output at time t, K p , K i , K d Respectively represent proportional, integral, and differential gains, F target represents the expected comprehensive eigenvalue, It reflects the changing trend of error over time;

[0119] Introducing reinforcement learning algorithm to further optimize the control strategy:

[0120]

[0121] Among them, θ k+1 represents the updated policy parameters, π θ represents the current strategy, κ represents a complete trajectory, T is the time step, γ represents the discount factor, r t represents the reward function.

[0122] By defining a reward function, corresponding rewards or penalties are given according to the changes in the comprehensive eigenvalues, and the control strategy is adjusted dynamically. The reward function is defined as follows:

[0123] r t = -α(F fuse,t -F target ) 2 +βlogπ θ (a t |s t );

[0124] Among them, α and β represent the weight parameters in the reward function, F fuse,t represents the comprehensive eigenvalue at time t, π θ (a t |s t ) means in state s t A t probability.

[0125] Utilize historical data and real-time monitoring results to intelligently monitor and control the slag cooling process, including:

[0126]

[0127] Among them, ω i (t+1) represents the model weight at time t+1, ω i (t) represents the model weight at time t, represents the learning rate, Represents the weight ω i The partial derivative of Loss new (F fuse (t),F true (t)) represents the value of the new loss function at time t. The new loss function combines the current loss and the historical average loss. The calculation formula is as follows:

[0128]

[0129] Among them, F truerepresents the true feature value, λ represents the weight parameter, which is used to balance the impact of the current loss and the historical average loss, T' represents the time window length, j represents the index variable used in the summation process, which is used to traverse the time points in the time window, and F fuse (t) represents the features obtained by fusion of multi-source data at time t, F true (t) represents the true eigenvalue at time t, F fuse (j) represents the features obtained by fusion of multi-source data at time j, F true (j) represents the true eigenvalue at time j, Loss(F fuse (j),F true (j)) represents the loss function value calculated at time j.

[0130] Through the above method, the model can continuously adjust the weights according to real-time monitoring data and historical data to minimize the loss function, thereby improving the accuracy and efficiency of monitoring and control. This method is particularly suitable for scenarios such as coal slag cooling processes that require real-time response and precise control. It can effectively improve cooling efficiency and stability, reduce labor costs, and reduce the risk of equipment failure.

[0131] Embodiment 2

[0132] like Figure 2 As shown, the present invention also provides a structural schematic diagram of a monitoring system for a slag cooling process, including:

[0133] The sensor module is used to collect various data in the furnace of the slag cooling section of the power plant boiler. The sensor module includes: a three-dimensional laser scanner to obtain the thickness data of the slag; an infrared thermal imaging array sensor to obtain the temperature data of the slag; an integrated wind parameter sensor to obtain the wind temperature, wind pressure and wind volume data; a camera to obtain the image data in the furnace;

[0134] The data processing module is connected to the sensor module and is used to pre-process the collected data in the furnace, including filtering, outlier removal and normalization;

[0135] The data analysis module is connected to the data processing module and is used to perform advanced analysis on the pre-processed data, including: using the convolutional neural network CNN combined with the attention mechanism to analyze the image data and extract visual features; using the long short-term memory network LSTM combined with the attention mechanism to analyze the slag temperature, wind temperature, wind pressure and wind volume data and extract time series features; further integrating the visual features, time series features and the initially integrated features to form comprehensive features;

[0136] A data display module, connected to the data analysis module, is used to display comprehensive features by combining augmented reality AR and virtual reality VR technologies;

[0137] An operation control module, connected to the data analysis module and the data display module, is used to monitor and adjust the slag cooling process in real time according to the comprehensive characteristics of the display in combination with the adaptive control algorithm and the reinforcement learning algorithm;

[0138] The intelligent learning and optimization module is connected with the data analysis module and the operation control module, and is used to perform intelligent learning and optimization of the monitoring system based on historical data and real-time monitoring results.

[0139] It is worth noting that the contents not elaborated in detail in the present invention are all prior art and are well known to those skilled in the art.

[0140] Therefore, the present invention adopts the above-mentioned coal slag cooling process monitoring method and system, which can realize accurate monitoring and intelligent regulation of the coal slag cooling process, improve cooling efficiency and stability, and reduce labor costs.

[0141] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for monitoring a slag cooling process, characterized in that: The following steps are involved: Step S1, arranging a variety of sensors in the furnace of the slag cooling section of the power plant boiler, including a three-dimensional laser scanner, an infrared thermal imaging array sensor, an integrated wind parameter sensor and a camera, for acquiring data in the furnace; Step S2: filtering the collected furnace data and removing abnormal values ​​through a data processing module; Step S3: Perform advanced analysis on the collected data in the data processing module using deep learning algorithms and machine learning models, including: Normalizing the acquired furnace data; The convolutional neural network (CNN) is used in combination with the attention mechanism to analyze the image data in the furnace and extract visual features. The long short-term memory network LSTM is used in combination with the attention mechanism to analyze the slag temperature, wind temperature, wind pressure, and wind volume to extract time series features; Further integrate visual features, time series features, and initially integrated features to form comprehensive features; Step S4, displaying comprehensive features through a data display module combining augmented reality AR and virtual reality VR; Step S5: Based on the comprehensive characteristics displayed, the slag cooling process is monitored and adjusted in real time in the operation control module in combination with the adaptive control algorithm and the reinforcement learning algorithm.

2. The method for monitoring the slag cooling process according to claim 1, characterized in that: In step S1, the data inside the furnace include: slag thickness obtained by a three-dimensional laser scanner; slag temperature obtained by an infrared thermal imaging array sensor; wind temperature, wind pressure and air volume obtained by an integrated wind parameter sensor; and image data inside the furnace obtained by a camera.

3. The method for monitoring the slag cooling process according to claim 1, characterized in that: In step S2, the calculation formula for filtering is as follows: Among them, y z represents the filtered data, N represents the size of the sliding window, x k represents the sensor data at time k, and z represents the index variable; The calculation formula for outlier removal is as follows: if|x k -μ|>3σ, remove x k ; Where μ represents the mean of the sensor data and σ represents the standard deviation of the sensor data.

4. The method for monitoring the slag cooling process according to claim 1, characterized in that: In step S3, the acquired furnace data is normalized, including: Among them, y (s) represents the normalized features of the sth sensor data, x (s) represents the raw data of the sth sensor data, μ (s) , σ (s) represent the mean and standard deviation of the sth sensor data respectively.

5. The method for monitoring the slag cooling process according to claim 4, characterized in that: In step S3, the image data in the furnace is analyzed using a convolutional neural network (CNN) combined with an attention mechanism to extract visual features, including: F vis =CNN att (y (1) ); Among them, F vis Representing visual features, CNN att represents the CNN model with the attention mechanism introduced, y (1) Represents the normalized image data.

6. A method for monitoring the slag cooling process according to claim 5, characterized in that: In step S3, the LSTM network is used in combination with the attention mechanism to analyze the slag temperature, wind temperature, wind pressure, and wind volume to extract time series features, including: F temp =LSTM att (y (2) ,y (3) ,y (4) ,y (5) ,); Among them, F temp Represents time series features, LSTM att represents the LSTM model with the attention mechanism introduced, y (2) ,y (3) ,y (4) ,y (5) Represents the normalized slag temperature, wind temperature, wind pressure and air volume.

7. A method for monitoring the slag cooling process according to claim 6, characterized in that: In step S3, the visual features, time series features, and initially fused features are further fused to form comprehensive features, including: Among them, ω vis ,ω temp ,ω pre Respectively represent the weights of visual features, time series features, and initial fusion features, F pre It represents the initial fusion feature, and its calculation formula is as follows: F pre =concat(and (1) ,and (2) ,and (3) ,and (4) ,and (5) )。 8. The method for monitoring the slag cooling process according to claim 7, characterized in that: In step S4, the comprehensive features are displayed through a data display module combining augmented reality AR and virtual reality VR, including: Among them, color(F fuse ) represents the color mapping function, F low Indicates the lower limit of the healthy state of the cooling process, F high Indicates the upper threshold of the health status of the cooling process.

9. A method for monitoring the slag cooling process according to claim 8, characterized in that: In step S5, based on the comprehensive characteristics displayed, the slag cooling process is monitored and adjusted in real time in the operation control module in combination with the adaptive control algorithm and the reinforcement learning algorithm, including: Use adaptive control algorithms to make initial adjustments to the cooling process: Where u(t) represents the control output at time t, K p , K i , K d Respectively represent proportional, integral, and differential gains, F target represents the expected comprehensive eigenvalue, It reflects the changing trend of error over time; Introducing reinforcement learning algorithm to further optimize the control strategy: Among them, θ k+1 represents the updated policy parameters, π θ represents the current strategy, κ represents a complete trajectory, T is the time step, γ represents the discount factor, r t Represents the reward function, the reward function r t The definition is as follows: r t =-α(F fuse,t -F target ) 2 +blog θ (a t |s t ); Among them, α and β represent the weight parameters in the reward function, F fuse,t represents the comprehensive eigenvalue at time t, π θ (a t |s t ) means in state s t A t probability.

10. A method for monitoring the slag cooling process according to claim 9, characterized in that: Utilize historical data and real-time monitoring results to intelligently monitor and control the slag cooling process, including: Among them, ω i (t+1) represents the model weight at time t+1, ω i (t) represents the model weight at time t, represents the learning rate, Represents the weight ω i The partial derivative of Loss new (F fuse (t),F true (t)) represents the value of the new loss function at time t. The new loss function combines the current loss and the historical average loss. The calculation formula is as follows: Among them, F true represents the true feature value, λ represents the weight parameter, which is used to balance the impact of the current loss and the historical average loss, T' represents the time window length, j represents the index variable used in the summation process, which is used to traverse the time points in the time window, and F fuse (t) represents the features obtained by fusion of multi-source data at time t, F true (t) represents the true eigenvalue at time t, F fuse (j) represents the features obtained by fusion of multi-source data at time j, F true (j) represents the true eigenvalue at time j, Loss(F fuse (j),F true (j)) represents the loss function value calculated at time j.

11. A monitoring system for coal slag cooling process, characterized in that: include: Sensor modules for collecting various data in the furnace of the slag cooling section of a power plant boiler; A data processing module, connected to the sensor module, for preprocessing the collected data in the furnace; A data analysis module, connected to the data processing module, for performing advanced analysis on the pre-processed data; A data display module, connected to the data analysis module, is used to display comprehensive features by combining augmented reality AR and virtual reality VR technologies; An operation control module, connected to the data analysis module and the data display module, is used to monitor and adjust the slag cooling process in real time according to the comprehensive characteristics of the display in combination with the adaptive control algorithm and the reinforcement learning algorithm; The intelligent learning and optimization module is connected with the data analysis module and the operation control module, and is used to perform intelligent learning and optimization of the monitoring system based on historical data and real-time monitoring results.

Images