Adaptive intelligent firing power generation DCS regulation control method

Abstract

The present application relates to the technical field of thermal power generation, and particularly relates to a self-adaptive intelligent thermal power generation DCS adjustment control method, comprising the following steps: S1, real-time monitoring and data analysis: real-time monitoring of key parameters in the power station through sensors; S2, dynamic fuel quality evaluation: implementing dynamic fuel quality evaluation; S3, power grid load monitoring: real-time monitoring of changes in power grid demand; S4, automatic adjustment and environmental control: DCS receives adjustment suggestions and automatically adjusts control strategies; S5, effect evaluation and strategy optimization: collecting and analyzing effect evaluation information after adjusting control strategies; S6, abnormal monitoring and emergency plan: setting an abnormal situation processing mechanism, when monitoring key parameters are abnormal, automatically starting an emergency plan to quickly respond and ensure the stable operation of thermal power generation. The present application can timely respond to any sudden changes and maintain the stability of the power generation process.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of thermal power generation technology, and particularly relates to an adaptive intelligent thermal power generation DCS regulation control method. BACKGROUND

[0002] In the field of thermal power generation, traditional distributed control systems (DCS) mainly rely on preset control strategies to manage the power generation process. Although these systems can basically meet the daily operation needs, they often appear to be not flexible enough to adapt to the complex changing operating environment in the face of fuel quality fluctuations, changes in grid load, and increasing environmental protection requirements. For example, the uncertainty of fuel quality may lead to a decrease in combustion efficiency and an increase in environmental emissions, while rapid changes in grid load require the power generation system to quickly adjust its output to ensure the stability of the grid. In addition, increasingly stringent environmental regulations also require power plants to reduce pollutant emissions and improve energy use efficiency during operation.

[0003] Current thermal power generation DCS control methods often lack efficient real-time data processing capabilities, making it difficult to achieve real-time monitoring and dynamic optimization of key operating parameters. This limits the response speed and adaptability of power plants in the face of changes in fuel quality and fluctuations in grid demand. Therefore, there is an urgent need for an intelligent control method that can analyze and dynamically adjust in real time to improve the operational efficiency of thermal power generation, optimize fuel use efficiency, and reduce environmental impact. SUMMARY

[0004] Based on the above purpose, the present application provides an adaptive intelligent thermal power generation DCS regulation control method.

[0005] The adaptive intelligent thermal power generation DCS regulation control method comprises the following steps:

[0006] S1, real-time monitoring and data analysis: real-time monitoring of key parameters in the power plant through sensors, including fuel flow, steam pressure, temperature, and emission data, and analyzing the key parameters to identify the current operating state of the thermal power generation;

[0007] S2, dynamic fuel quality assessment: based on the output of real-time monitoring and data analysis, dynamic fuel quality assessment is implemented, and according to the real-time analysis of fuel sample data, the operating parameters of the combustion equipment are automatically adjusted to adapt to changes in fuel quality;

[0008] S3, grid load monitoring: integrate grid load monitoring mechanism, real-time monitor the changes of grid demand, through capturing real-time data of grid load, provide load change information, provide data support for DCS control strategy adjustment;

[0009] S4, automatic adjustment and environmental control: based on the results of dynamic fuel quality assessment and grid load monitoring, the DCS receives adjustment suggestions and automatically adjusts the control strategy to optimize fuel use efficiency and reduce environmental emissions;

[0010] S5, effect evaluation and strategy optimization: after the DCS executes the adjusted control strategy, it collects and analyzes the effect evaluation information after the adjustment of the control strategy, and adjusts and optimizes the adjusted control strategy based on the analysis results of the effect evaluation information;

[0011] S6, abnormality monitoring and emergency plan: set up an abnormal situation handling mechanism, when the key parameters are abnormal, automatically start the emergency plan to quickly respond and ensure the stable operation of the thermal power generation.

[0012] Further, the real-time monitoring and data analysis in S1 includes:

[0013] S11, sensor deployment and data collection: deploy various sensors in the power station to monitor key parameters in real time, including flow sensors, pressure sensors, temperature sensors, and emission monitoring sensors. The flow sensor is used to monitor fuel flow, the pressure sensor is used to monitor steam pressure, the temperature sensor is used to record temperature, and the emission monitoring sensor is used to monitor environmental emission data;

[0014] S12, data analysis: use linear regression model to analyze the key parameters collected from various sensors in real time to identify data patterns and trends;

[0015] S13, running state diagnosis: based on the results of data analysis, identify the current running state of the thermal power station, including running efficiency, operation anomaly and preventive maintenance needs.

[0016] Further, the dynamic fuel quality assessment in S2 includes:

[0017] S21, data preprocessing: obtain key parameters of real-time fuel samples from sensors and preprocess key parameters, including data cleaning and denoising;

[0018] S22, feature extraction: feature extraction is performed on the preprocessed key parameters to convert the original key data into feature vectors for evaluation;

[0019] S23, model construction and training: based on the feature vectors, a dynamic fuel quality assessment model is constructed, and known fuel sample data is used as a training set to train the dynamic fuel quality assessment model;

[0020] S24, real-time evaluation: after the dynamic fuel quality evaluation model is trained, input the real-time collected fuel sample data into the dynamic fuel quality evaluation model for real-time evaluation, to predict the quality characteristics and change trend of the fuel;

[0021] S25, parameter adjustment: according to the results of real-time evaluation, automatically adjust the operating parameters of the combustion equipment, including air-fuel ratio and injection speed.

[0022] Further, the dynamic fuel quality evaluation model adopts a Gaussian process regression model (GPR), which includes:

[0023] Define the kernel function: combine the radial basis function (RBF) and Matérn function to adapt to different data smoothness and irregularity, and the calculation formula is:

[0024]

[0025] Where x and x' are input feature vectors, θ1 and θ3 are amplitude parameters of the kernel function, and θ2 and θ4 are length parameters;

[0026] Train and optimize the model: use the maximum likelihood estimation method to optimize the parameters of the kernel function, and the calculation formula is:

[0027]

[0028] Where, K is the kernel matrix, is the noise variance, I is the unit matrix, y is the response variable vector, and X is the input variable matrix;

[0029] Realize dynamic update: introduce online learning mechanism, use new data points (new fuel sample data) to update Gaussian process regression model, when new observation (x new ,y new ) comes, the calculation formula is:

[0030]

[0031]

[0032] The updated Gaussian process regression model uses K new and y new for prediction;

[0033] Where K new is the updated covariance matrix, y new is the updated response variable vector, and k(x,x new ) is the covariance vector between new data points and existing data points;

[0034] Prediction and error estimation: use the optimized Gaussian process regression model to make predictions for new input points x * , use the Gaussian process regression model to make predictions, and calculate the standard deviation of the prediction, the formula is:

[0035]

[0036]

[0037] Where μ(x * ) is the predicted mean, σ 2 (x * ) is the predicted variance.

[0038] Further, the grid load monitoring in S3 includes:

[0039] S31, real-time load data acquisition: real-time monitoring of the load parameters of the power grid, including total load, load fluctuation, peak demand;

[0040] S32, introduction of grid load monitoring mechanism and data analysis: integrate the grid load monitoring mechanism into the DCS, and perform real-time processing and analysis on the collected load parameters to identify load change trends and patterns;

[0041] S33, load change information and data support: based on the results of the introduction of the grid load monitoring mechanism and data analysis, generate a grid load change information report to show the current state and future trends of the grid load, and provide the load change information report directly to the DCS as data support for adjusting power output and optimizing control strategies.

[0042] Further, the real-time load data acquisition in S31 includes:

[0043] Load sensor real-time monitoring of total load: use load sensors to monitor the total load of the power grid in real time, load sensors are installed on transmission lines or power generation equipment to measure current or power, thereby obtaining the total load of the power grid;

[0044] Current sensor monitoring of load fluctuation: use current sensors to monitor the fluctuation of the power grid load in real time, and sense the dynamic changes of the load by monitoring the changes of the current;

[0045] Power sensor detection of peak demand: use power sensors to detect the maximum power demand of the power grid in a predetermined period of time, power sensors are usually installed on generators, substations or transmission lines to measure the power output of the power grid.

[0046] Further, the introduction of the grid load monitoring mechanism and data analysis in S32 includes:

[0047] S321, collecting load parameters and real-time processing: the DCS collects load parameters of the power grid in real time, and performs pre-real-time processing on the collected load parameters, including data cleaning, denoising and correction;

[0048] S322, load data analysis: using a convolutional neural network (CNN) model to analyze the load parameters processed in real time;

[0049] S323, identifying load change trend and mode: based on the results of load data analysis, the DCS identifies the change trend and mode of the power grid load, including discovering the periodic fluctuation of the load, the trend of the load change, and the sudden increase or decrease of the load.

[0050] Further, the convolutional neural network (CNN) model comprises:

[0051] Data arrangement: data arrangement is performed on the load parameters monitored in real time, including normalization and denoising;

[0052] Stacking of convolutional layers and pooling layers: multiple convolutional layers and pooling layers are stacked to extract hierarchical features of load data, and the calculation formula is:

[0053] Convolution operation:

[0054] Pooling operation:

[0055] Wherein, is the feature value at position (i,j) in the output feature map of the lth convolutional layer, F is the size of the convolution kernel, is the parameter of the lth convolution kernel, m and n represent the rows and columns of the convolution kernel respectively, b (l) is the bias term of the lth layer, is the feature value at position (i,j) in the output feature map of the lth pooling layer, s is the size of the pooling kernel;

[0056] Multi-scale convolution: multi-scale convolution operation is introduced to consider the load change mode under different time scales, and the calculation formula is:

[0057]

[0058] Wherein, K is the number of scales of multi-scale convolution, F k is the size of the convolution kernel of the kth scale, is the parameter of the lth multi-scale convolution kernel, m and n represent the rows and columns of the convolution kernel respectively, b (l) is the bias term of the lth layer;

[0059] Learning time sequence features: combined with attention mechanism, the learning of time sequence features is introduced to capture the time dependence of load data, and the calculation formula is:

[0060] α t = softmax(W α ·h t +b α );

[0061]

[0062] Where h t represents the hidden state of each time step, T represents the length of the time sequence, α t is the attention weight of time step t, W α is the attention weight parameter, and b α is the attention weight bias.

[0063] Fully connected layer and output layer: the output of the convolutional layer and the time sequence feature extraction part is connected to the fully connected layer, and the final prediction or classification is carried out through the output layer, and the calculation formula is:

[0064] y (l) = σ(w (l) ·x (l-1) +b (l) );

[0065] Where y (l) is the output of the l-th fully connected layer, W (l) is the weight parameter of the l-th layer, x (l-1) is the output of the l-1 layer, and b (l) is the bias term of the l-th layer.

[0066] Further, the automatic adjustment and environmental control in S4 include:

[0067] S41, receiving results: DCS receives result data from dynamic fuel quality evaluation and power grid load monitoring;

[0068] S42, analysis and adjustment suggestions: based on the received result data, DCS analyzes adjustment suggestions to determine control strategies, including adjusting power generation, fuel supply rate and combustion parameters;

[0069] S43, automatic adjustment of control strategy: DCS automatically adjusts the control strategy according to the result of analyzing the adjustment suggestion to optimize fuel use efficiency and reduce environmental emissions, including adjusting fuel mixing ratio, changing combustion temperature or adjusting operation parameters of emission treatment equipment;

[0070] S44, real-time monitoring and feedback: DCS continuously monitors the operation of the adjusted thermal power generation and provides effect evaluation information based on real-time monitoring of key parameters.

[0071] Further, the effect evaluation and strategy optimization in S5 include:

[0072] S51, evaluation information analysis: the collected effect evaluation information is evaluated using mean absolute error (MAE) to measure the deviation between the predicted value and the actual value;

[0073] S52, strategy optimization: based on the results of evaluation information analysis, the control strategy is adjusted and optimized;

[0074] S53, real-time optimization: the optimized control strategy is applied to thermal power generation, and key parameters are continuously collected for monitoring and feedback.

[0075] Advantages of the present application:

[0076] The present application, through real-time monitoring and data analysis, dynamic fuel quality evaluation and power grid load monitoring, can effectively respond to changes in fuel quality and fluctuations in power grid demand. This integrated monitoring and adjustment strategy not only improves the operation efficiency of thermal power generation, but also enhances the adaptability of the system to environmental changes. Real-time data analysis ensures the continuity and efficiency of the power generation process. Through dynamic adjustment strategy, the system can respond to any sudden changes in time and maintain the stability of the power generation process.

[0077] The present application, by precisely adjusting the fuel mixing ratio and combustion temperature, and optimizing the operating parameters of the emission treatment equipment, not only reduces the operating cost, but also significantly reduces the harmful emissions, making a positive contribution to environmental protection and resource conservation, and ensuring the sustainability of power generation activities.

[0078] The present application, by real-time monitoring of key parameters and timely activation of emergency plans, the system can quickly and effectively respond to various abnormal situations, thereby ensuring the stable operation of the power generation system. In addition, the implementation of effect evaluation and strategy optimization stage further improves the safety and efficiency of the system operation. Through continuous learning and optimization, the entire regulation and control method is more accurate and efficient. BRIEF DESCRIPTION OF DRAWINGS

[0079] In order to more clearly illustrate the technical solutions in the present application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only a part of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0080] Fig. 1A control method flowchart for an embodiment of the present application;

[0081] Fig. 2 A dynamic fuel quality assessment schematic for an embodiment of the present application. DETAILED DESCRIPTION

[0082] To make the objectives, technical solutions, and advantages of the present application clearer, further detailed descriptions are provided below in combination with specific embodiments.

[0083] It should be noted that, unless otherwise defined, technical terms or scientific terms used in the present application should be understood as their common meanings to those skilled in the art. The terms "first", "second", and similar terms used in the present application do not represent any order, number, or importance, but are only used to distinguish different components. The terms "include", "contain", and similar terms mean that the components or objects before the terms encompass the components or objects listed after the terms and their equivalents, without excluding other components or objects. The terms "connect" or "connected" and similar terms do not mean physical or mechanical connections, but can include electrical connections, whether direct or indirect. The terms "upper", "lower", "left", "right", and the like only represent relative positional relationships, which can change when the absolute positions of the described objects change.

[0084] As shown in Figs. 1-2 the adaptive intelligent thermal power generation DCS regulation control method includes the following steps:

[0085] S1, real-time monitoring and data analysis: real-time monitoring of key parameters in the power station through sensors, including fuel flow, steam pressure, temperature, and emission data, and analyzing the key parameters to identify the current operating state of thermal power generation;

[0086] S2, dynamic fuel quality assessment: based on the output of real-time monitoring and data analysis, dynamic fuel quality assessment is implemented, and based on real-time analysis of fuel sample data, the operating parameters of the combustion equipment are automatically adjusted to adapt to changes in fuel quality;

[0087] S3, power grid load monitoring: integrating power grid load monitoring mechanisms, real-time monitoring of changes in power grid demand, capturing real-time data of power grid load to provide load change information, and providing data support for DCS control strategy adjustment;

[0088] S4, automatic adjustment and environmental control: based on the results of dynamic fuel quality assessment and power grid load monitoring, DCS receives adjustment suggestions and automatically adjusts control strategies to optimize fuel use efficiency and reduce environmental emissions;

[0089] S5, Effect Evaluation and Strategy Optimization: After the DCS executes the adjustment control strategy, the effect evaluation information after the adjustment control strategy is collected and analyzed, and the adjustment control strategy is adjusted and optimized based on the analysis results of the effect evaluation information, to improve the operation efficiency and response accuracy in the future;

[0090] S6, Abnormality Monitoring and Emergency Plan: Set up an abnormal situation handling mechanism, when the key parameters are abnormal, automatically start the emergency plan to quickly respond and ensure the stable operation of thermal power generation;

[0091] Through the above steps, the operation efficiency and environmental adaptability of thermal power generation are significantly improved. Using real-time data analysis and dynamic adjustment strategy, it can effectively respond to changes in fuel quality and fluctuations in power grid demand, ensuring efficient and stable power generation process. In addition, through the implementation of effective abnormality monitoring and emergency plan, this method not only enhances the reliability of the system, but also optimizes the environmental impact, ensuring sustainable environmental protection and resource conservation.

[0092] The real-time monitoring and data analysis in S1 includes:

[0093] S11, Sensor Deployment and Data Collection: Deploy various sensors inside the power station to monitor key parameters in real time. The sensors include flow sensors, pressure sensors, temperature sensors, and emission monitoring sensors. Flow sensors are used to monitor fuel flow, pressure sensors are used to monitor steam pressure, temperature sensors are used to record temperature, and emission monitoring sensors are used to monitor environmental emission data;

[0094] S12, Data Analysis: Use linear regression model to analyze key parameters collected from various sensors in real time to identify data patterns and trends;

[0095] The calculation formula of linear regression model is:

[0096] y = β0 + β1x1 + β2x2 + … + β n x n + ∈;

[0097] Where y is the dependent variable, x1, x2, …, x n are independent variables, β0, β1, …, β n are model parameters, and ∈ is the error term;

[0098] S13, Running State Diagnosis: Based on the results of data analysis, identify the current running state of thermal power station, including running efficiency, operation abnormality and preventive maintenance demand;

[0099] Through the above steps, the running changes of thermal power station can be identified and responded in time, improving the running safety and efficiency of the system.

[0100] The dynamic fuel quality assessment in S2 includes:

[0101] S21, data preprocessing: acquiring real-time key parameters of fuel samples from sensors and preprocessing the key parameters, which includes data cleaning and denoising;

[0102] S22, feature extraction: extracting features from the preprocessed key parameters, converting the original key data into feature vectors for evaluation, such as average temperature, humidity change rate, etc.;

[0103] The calculation of feature extraction is as follows:

[0104] Mean: calculate the mean of key parameter data to reflect the central tendency of data, the calculation formula is:

[0105]

[0106] Where μ is the mean, N is the number of data samples, x i is the value of the i-th data sample;

[0107] Standard deviation: calculate the standard deviation of key parameter data, which represents the dispersion or variation of data, the calculation formula is:

[0108]

[0109] Where σ is the standard deviation, N is the number of data samples, x i is the value of the i-th data sample, μ is the mean of the data;

[0110] Correlation coefficient: calculate the correlation between key parameter data to explore the association between different parameters, the calculation formula is:

[0111]

[0112] Where r is the correlation coefficient, N is the number of data samples, x i and y i are the i-th data samples of two parameters, and are the means of the two parameters, respectively;

[0113] S23, model construction and training: based on the feature vector, construct a dynamic fuel quality assessment model, and use known fuel sample data as a training set to train the dynamic fuel quality assessment model;

[0114] Fuel sample data includes:

[0115] Fuel composition: refers to the content of various chemical components in fuel, such as the content of elements such as carbon, hydrogen, oxygen, sulfur, etc., usually expressed in percentage or mass fraction;

[0116] Volatile content: refers to the content of combustible substances volatilized during heating in fuel, usually expressed in percentage;

[0117] Ash content: refers to the content of non-combustible substances in fuel, usually expressed in percentage;

[0118] Calorific value: refers to the energy released per unit mass or unit volume of fuel, usually expressed in joules / gram (J / g) or kilojoules / cubic meter (kJ / m 3 ) represents;

[0119] Density: refers to the density of fuel, usually expressed in kilograms / cubic meter or grams / milliliter;

[0120] Moisture content: refers to the water content in fuel, usually expressed in percentage or absolute humidity;

[0121] Particle size distribution: refers to the size distribution of fuel particles, usually expressed in percentage of particle diameter or particle size range;

[0122] S24, real-time evaluation: after the dynamic fuel quality evaluation model is trained, real-time fuel sample data is input into the dynamic fuel quality evaluation model for real-time evaluation, to predict the quality characteristics and change trend of the fuel;

[0123] S25, parameter adjustment: according to the results of real-time evaluation, automatically adjust the operating parameters of the combustion equipment, including air-fuel ratio and injection speed, the adjustment strategy is based on the output of the dynamic fuel quality evaluation model, aiming to optimize the combustion process and adapt to the changes of fuel quality;

[0124] Through the above steps, dynamic evaluation of fuel quality can be realized, and operating parameters of the combustion equipment can be automatically adjusted according to the evaluation results, so as to improve the combustion efficiency and stability of the system.

[0125] The dynamic fuel quality evaluation model adopts Gaussian process regression model (GPR), which includes:

[0126] Define kernel function: combine radial basis function (RBF) and Matérn function to adapt to different data smoothness and irregularity, the formula is:

[0127]

[0128] where x and x' are input feature vectors representing the features of two different data points, θ1 and θ3 are amplitude parameters of the kernel function that determine the magnitude of the function value as x and x' vary, and θ2 and θ4 are length parameters that control the sensitivity of the function value to the difference between x and x';

[0129] Training and optimizing the model: The parameters of the kernel function are optimized using the maximum likelihood estimation method, with the formula:

[0130]

[0131] where, K is the kernel matrix, σ2 is the noise variance, I is the identity matrix, y is the response variable vector containing the output values of all observation points, and X is the input variable matrix containing the feature values of all observation points;

[0132] Implementing dynamic updates: To adapt to the dynamics of fuel parameters changing over time, online learning mechanisms are introduced to update the Gaussian process regression model using new data points (new fuel sample data) without retraining from scratch. When new observations (x new ,y new ) arrive, the formula is:

[0133]

[0134]

[0135] The updated Gaussian process regression model uses K new and y new for prediction;

[0136] where K new is the updated covariance matrix including the covariance of new data points, y new is the updated response variable vector including the response values of new data points, and k(x, x new ) is the covariance vector between new data points and existing data points;

[0137] Prediction and error estimation: The optimized Gaussian process regression model is used for prediction and estimation of the uncertainty of the prediction. For a new input point x * , the Gaussian process regression model is used for prediction, and the standard deviation of the prediction is calculated, with the formula:

[0138]

[0139]

[0140] where μ(x *) is the predicted mean, based on the predicted response of the new input point, σ 2 (x * ) is the predicted variance, representing the uncertainty of the prediction.

[0141] Through the above steps, the GPR model not only can handle complex fuel data characteristics, but also can dynamically adapt to new data, providing continuous fuel quality evaluation, which makes the model more flexible and effective in practical application.

[0142] The grid load monitoring in S3 includes:

[0143] S31, real-time load data collection: real-time monitoring of the load parameters of the grid, including total load, load fluctuation, peak demand;

[0144] S32, introduction of grid load monitoring mechanism and data analysis: integrate the grid load monitoring mechanism into the DCS, and process and analyze the collected load parameters in real time to identify load change trends and patterns;

[0145] S33, load change information and data support: based on the results of the introduction of the grid load monitoring mechanism and data analysis, generate a grid load change information report to show the current state and future trends of the grid load, and provide the load change information report directly to the DCS as data support for adjusting power generation output and optimizing control strategies, to ensure that power generation operations respond to real-time demands of the grid and improve the efficiency and stability of power production;

[0146] Through the above steps, the grid load monitoring not only improves the response capability to the changes in grid demand, but also enhances the flexibility and economic benefits of power station operation, providing important support for the stable operation of the entire power system.

[0147] The real-time load data collection in S31 includes:

[0148] Load sensor real-time monitoring of total load: using load sensors to monitor the total load of the grid in real time, load sensors are installed on power transmission lines or power generation equipment to measure current or power, thereby obtaining the total load of the grid;

[0149] Current sensor monitoring load fluctuation: using current sensors to monitor the fluctuation of the grid load in real time, by monitoring the changes in current to sense the dynamic changes of the load, such as sudden load increase or decrease;

[0150] Power sensor detects peak demand: using power sensors to detect the maximum power demand of the grid within a predetermined period of time, power sensors are usually installed on generators, substations or power transmission lines to measure the power output of the grid.

[0151] The introduction of the grid load monitoring mechanism in S32 and data analysis includes:

[0152] S321, collecting load parameters and real-time processing: DCS collects load parameters of the grid in real time, and performs pre-real-time processing on the collected load parameters. Real-time processing includes data cleaning, denoising and correction to ensure the accuracy and reliability of the data;

[0153] S322, load data analysis: using convolutional neural network (CNN) model to analyze the load parameters processed in real time;

[0154] S323, identifying load change trend and mode: based on the results of load data analysis, DCS identifies the change trend and mode of the grid load, including discovering the periodic fluctuations of the load, the trend of load change and the sudden increase or decrease of the load. Identifying these trends and modes helps to develop corresponding adjustment strategies to ensure the stability and efficiency of the grid operation.

[0155] The convolutional neural network (CNN) model includes:

[0156] Data arrangement: data arrangement is performed on the real-time monitored load parameters, including normalization and denoising, to improve the robustness and accuracy of the model;

[0157] Stacking of convolutional layers and pooling layers: multiple convolutional layers and pooling layers are stacked to extract hierarchical features of load data, and the calculation formula is:

[0158] Convolution operation:

[0159] Pooling operation:

[0160] Wherein, is the feature value at position (i,j) in the output feature map of the lth convolutional layer, F is the size of the convolution kernel, is the parameter of the lth convolution kernel, m and n represent the rows and columns of the convolution kernel respectively, b (l) is the bias term of the lth layer, is the feature value at position (i,j) in the output feature map of the lth pooling layer, s is the size of the pooling kernel;

[0161] Multi-scale convolution: multi-scale convolution operation is introduced to consider load change patterns under different time scales to capture more comprehensive feature information, and the calculation formula is:

[0162]

[0163] Wherein, K is the number of scales of multi-scale convolution, F k is the size of the convolution kernel of the kth scale, is the parameter of the multi-scale convolution kernel of the first layer, m and n represent the rows and columns of the convolution kernel respectively, b (l) is the bias term of the first layer;

[0164] Learning time sequence features: combined with the attention mechanism, the learning of time sequence features is introduced to capture the time dependence of load data, and the calculation formula is:

[0165] α t = softmax(W α ·h t +b α );

[0166]

[0167] where h t represents the hidden state of each time step, T represents the length of the time sequence, α t is the attention weight of time step t, W α is the attention weight parameter, and b α is the attention weight bias;

[0168] Fully connected layer and output layer: the output of the convolution layer and the time sequence feature extraction part is connected to the fully connected layer, and the final prediction or classification is performed through the output layer, and the calculation formula is:

[0169] y (l) = σ(w (l) ·x (l-1) +b (l) );

[0170] where y (l) is the output of the l-th fully connected layer, W (l) is the weight parameter of the l-th layer, x (l-1) is the output of the l-1 layer, and b (l) is the bias term of the l-th layer;

[0171] Through the above, not only the spatial features can be captured, but also the time sequence features can be learned using the attention mechanism, which can better adapt to the dynamic changes of real-time load data, and can more accurately predict the trend and pattern of power grid load, thereby providing more effective support for the operation of power stations and ensuring the stability and reliability of the power grid.

[0172] The automatic adjustment and environmental control in S4 includes:

[0173] S41, receiving results: the DCS receives result data from dynamic fuel quality assessment and power grid load monitoring;

[0174] S42, analyze adjustment suggestions: based on the received result data, the DCS analyzes adjustment suggestions to determine control strategies, including adjusting power generation, fuel supply rate, and combustion parameters;

[0175] S43, automatically adjust control strategies: the DCS automatically adjusts control strategies based on the results of analyzing adjustment suggestions to optimize fuel use efficiency and reduce environmental emissions, including adjusting fuel mixing ratios, changing combustion temperatures, or adjusting operating parameters of emission treatment equipment;

[0176] S44, real-time monitoring and feedback: the DCS continuously monitors the operation of the adjusted thermal power generation and provides effect evaluation information based on real-time monitoring of key parameters;

[0177] Through the above steps, automatic adjustment and environmental control can quickly respond and adjust the control strategy of the power generation system according to the real-time changes of dynamic fuel quality and power grid load, so as to optimize fuel use efficiency and reduce environmental emissions.

[0178] Effect evaluation and strategy optimization in S5 includes:

[0179] S51, evaluation information analysis: the collected effect evaluation information is evaluated using mean absolute error (MAE) to measure the deviation between predicted values and actual values;

[0180] The formula for calculating the mean absolute error (MAE) is:

[0181]

[0182] Where y i is the actual value, is the predicted value, and n is the sample size;

[0183] S52, strategy optimization: based on the results of evaluation information analysis, adjust and optimize the control strategy;

[0184] S53, real-time optimization: the optimized control strategy is applied to thermal power generation to ensure that the system operates in the best state and continues to collect key parameters for monitoring and feedback;

[0185] Through the above steps, the effect evaluation and strategy optimization phase helps to improve the accuracy and response speed of adjusting the control strategy by continuously collecting and analyzing real-time data to adapt to changes in fuel quality and power grid load and other factors, thereby optimizing the operating efficiency and stability of the thermal power generation system.

[0186] Those skilled in the art should understand that the above discussion of any embodiment is merely exemplary in nature and is not intended to imply that the present application is limited to these examples; any of the above embodiments or technical features among different embodiments can be combined, and steps can be implemented in any order, under the idea of the present application, and there are many other changes of different aspects of the present application as described above, which are not provided in details for the sake of brevity.

[0187] The present application is intended to cover all such alternatives, modifications, and variations as fall within the broad scope of the claims. Accordingly, any and all such alternatives, modifications and variations should be included within the scope of the present application.

Claims

1. A method for adaptive intelligent power plant DCS regulation control, characterized in that, The method comprises the following steps: S1, real-time monitoring and data analysis: real-time monitoring of key parameters in the power plant through sensors, including fuel flow, steam pressure, temperature and emission data, and analyzing the key parameters to identify the current operating state of the thermal power plant; S2, dynamic fuel quality assessment: based on the output of real-time monitoring and data analysis, dynamic fuel quality assessment is implemented, and the operating parameters of the combustion equipment are automatically adjusted according to the real-time analysis of the fuel sample data to adapt to the changes in fuel quality; S3, grid load monitoring: integrate grid load monitoring mechanism into DCS, real-time monitor the changes of grid demand, collect load parameters of the grid, real-time process the collected load parameters, use convolutional neural network model to analyze the real-time processed load parameters, identify the trend and pattern of load changes, provide data support for DCS control strategy adjustment; the convolutional neural network model includes introducing multi-scale convolution operation, considering load change pattern under different time scales, represented as: wherein K is the number of scales of the multi-scale convolution, F k is the kernel size of the kth scale, is the parameter of the multi-scale convolution kernel of the lth layer, m and n represent the rows and columns of the convolution kernel, respectively, b (l) is the bias term of the lth layer; Learning time series features: combined with attention mechanism, learning time series features is introduced to capture the time dependence of load data, the calculation formula is: alpha t = softmax(W α · h t + b α ); where h t denotes the hidden state at each time step, T denotes the length of the time series, a is the attention weight at time step t, W α is the attention weight parameter, b α is the attention weight bias; Full connection layer and output layer: connect the output of convolution layer and time series feature extraction part to full connection layer, and make final prediction or classification through output layer, the calculation formula is: y (l) = σ(W (l) · x (l-1) + b (l) ); where y (l) is the output of the full connection layer of the lth layer, W (l) is the weight parameter of the lth layer, x (l-1) is the output of the l-1th layer, b (l) is the bias term of the lth layer; S4, automatic adjustment and environmental control: based on the results of dynamic fuel quality assessment and grid load monitoring, DCS receives adjustment suggestions and automatically adjusts control strategy to optimize fuel use efficiency and reduce environmental emissions; S5, effect evaluation and strategy optimization: after DCS executes the adjusted control strategy, collect and analyze the effect evaluation information after the adjustment of the control strategy, and adjust and optimize the adjustment control strategy based on the analysis results of the effect evaluation information; S6, abnormality monitoring and emergency plan: set up abnormality handling mechanism, when abnormality of key parameters is monitored, automatically start emergency plan to quickly respond and ensure stable operation of thermal power plant.

2. The adaptive intelligent thermal power plant DCS governing control method of claim 1, wherein, The real-time monitoring and data analysis in S1 comprises: S11, sensor deployment and data collection: deploy multiple sensors in the power plant to monitor key parameters in real time, including flow sensors, pressure sensors, temperature sensors and emission monitoring sensors, flow sensors for monitoring fuel flow, pressure sensors for monitoring steam pressure, temperature sensors for recording temperature, and emission monitoring sensors for monitoring environmental emission data; S12, data analysis: use linear regression model to analyze key parameters collected from each sensor in real time to identify data patterns and trends; S13, operating state diagnosis: based on the results of data analysis, identify the current operating state of the thermal power plant, including operating efficiency, operation abnormality and preventive maintenance requirements.

3. The adaptive intelligent thermal power plant DCS governing control method of claim 1, wherein, The dynamic fuel quality assessment in S2 comprises: S21, data preprocessing: obtain key parameters of real-time fuel samples from sensors and preprocess the key parameters, including data cleaning and denoising; S22, feature extraction: feature extraction is performed on the preprocessed key parameters, and the original key data is converted into a feature vector for evaluation; S23, model construction and training: a dynamic fuel quality evaluation model is constructed based on the feature vector, and known fuel sample data is used as a training set to train the dynamic fuel quality evaluation model; S24, real-time evaluation: after the dynamic fuel quality evaluation model is trained, real-time collected fuel sample data is input into the dynamic fuel quality evaluation model for real-time evaluation to predict the quality characteristics and change trend of the fuel; S25, parameter adjustment: based on the results of real-time evaluation, the operating parameters of the combustion equipment are automatically adjusted, including air-fuel ratio and injection speed.

4. The adaptive intelligent thermal power plant DCS governing control method of claim 3, wherein, The dynamic fuel quality evaluation model adopts a Gaussian process regression model, which includes: Defining a kernel function: combining a radial basis function and a Matérn function to adapt to different data smoothness and irregularities, the calculation formula is: Where x and x' are input feature vectors, θ1 and θ3 are amplitude parameters of the kernel function, and θ2 and θ4 are length parameters; Training and optimizing the model: the parameters of the kernel function are optimized using the maximum likelihood estimation method, and the calculation formula is: wherein K is a kernel matrix, is the noise variance, I is the identity matrix, y is the response variable vector, and X is the input variable matrix; Implement dynamic update: Introduce online learning mechanism, use new data points to update Gaussian process regression model, when new observation (x new ,y new ) comes, the calculation formula is: The updated Gaussian process regression model uses K new and y new to make predictions; where K new is the updated covariance matrix, y new is the updated response variable vector, k(x, x new ) is the covariance vector between the new data point and the existing data points; Prediction and error estimation: Prediction is performed using the optimized Gaussian process regression model for a new input point x * Prediction is performed using the Gaussian process regression model and the standard deviation of the prediction is calculated as: where μ(x * ) is the predicted mean and σ 2 (x * ) is the predicted variance.

5. The adaptive intelligent thermal power plant DCS governing control method of claim 1, wherein, The power grid load monitoring in S3 includes: S31, real-time load data acquisition: real-time monitoring of the load parameters of the power grid, including total load, load fluctuation, peak demand; S32, load change information and data support: based on the introduction of the power grid load monitoring mechanism and the results of data analysis, a power grid load change information report is generated to show the current state and future trend of the power grid load, and the load change information report is directly provided to the DCS as data support for adjusting power generation output and optimizing control strategies.

6. The adaptive intelligent thermal power plant DCS governing control method of claim 5, wherein, The real-time load data acquisition in S31 includes: Real-time monitoring of total load by load sensor: using a load sensor to monitor the total load of the power grid in real time, the load sensor is installed on the power transmission line or power generation equipment to measure current or power, thereby obtaining the total load condition of the power grid; Current sensor monitors load fluctuation: using a current sensor to monitor the fluctuation of the power grid load in real time, and sensing the dynamic change of the load by monitoring the change of the current; Power sensor detects peak demand: using a power sensor to detect the maximum power demand of the power grid in a predetermined time period, the power sensor is usually installed on the generator, substation or power transmission line to measure the power output condition of the power grid.

7. The adaptive intelligent thermal power plant DCS governing control method of claim 6, wherein, The introduction of the power grid load monitoring mechanism and data analysis in S32 includes: S321, acquisition of load parameters and real-time processing: the DCS acquires the load parameters of the power grid in real time and performs pre-real-time processing on the acquired load parameters, including data cleaning, denoising and correction; S322, identifying load change trend and pattern: based on the results of load data analysis, the DCS identifies the change trend and pattern of the power grid load, including discovering the periodic fluctuation of the load, the trend of load change and sudden increase or decrease of the load.

8. The adaptive intelligent thermal power plant DCS governing control method of claim 7, wherein, The convolutional neural network model includes: Data arrangement: data arrangement is performed on the real-time monitored load parameters, including normalization and denoising; Stacking of convolutional layers and pooling layers: Stack multiple convolutional layers and pooling layers to extract hierarchical features of the load data, with the formula: Convolution operation: Pooling operations: wherein, is the feature value at position (i, j) in the output feature map of the lth convolutional layer, F is the size of the convolutional kernel, is the parameter of the lth convolutional kernel, m and n represent the row and column of the convolutional kernel, respectively, b (l) is the bias term of the lth layer, is the feature value at position (i, j) in the output feature map of the lth pooling layer, s is the size of the pooling kernel.

9. The adaptive intelligent thermal power plant DCS governing control method of claim 1, wherein, The automatic adjustment and environmental control in S4 includes: S41, receiving results: DCS receives result data from dynamic fuel quality assessment and grid load monitoring; S42, analyzing adjustment suggestions: based on the received result data, DCS analyzes adjustment suggestions to determine control strategies, including adjusting power generation, fuel supply rate, and combustion parameters; S43, automatic adjustment of control strategies: DCS automatically adjusts control strategies based on the results of analyzing adjustment suggestions to optimize fuel use efficiency and reduce environmental emissions, including adjusting fuel mixing ratio, changing combustion temperature, or adjusting operating parameters of emission treatment equipment; S44, real-time monitoring and feedback: DCS continuously monitors the operation of the adjusted thermal power generation and provides effect evaluation information based on real-time monitoring of key parameters.

10. The adaptive intelligent thermal power plant DCS governing control method of claim 9, wherein, The effect evaluation and strategy optimization in S5 includes: S51, evaluation information analysis: use mean absolute error (MAE) to evaluate the collected effect evaluation information to measure the deviation between predicted values and actual values; S52, strategy optimization: adjust and optimize the control strategy based on the results of evaluation information analysis; S53, real-time optimization: the optimized control strategy is applied to thermal power generation, and key parameters continue to be collected for monitoring and feedback.

Images