Optimization control method and control system for flue gas recirculation and boiler coupling system

Abstract

The present invention relates to the technical field of boiler system control, and specifically to an optimization control method and a control system for a flue gas recirculation and boiler coupling system. A data acquisition module is used to collect boiler operation data in real time, a data preprocessing module is used to preprocess the boiler operation data, and a multivariate dynamic model is constructed based on the preprocessed data by a machine learning algorithm. An optimization target setting module is used to set a multi-target optimization function in combination with a linear weighted method. A hybrid optimization algorithm module is used to globally optimize the target function by combining a particle swarm optimization algorithm with a genetic algorithm to determine an optimal flue gas recirculation flow rate and a recirculation ratio. An adaptive control strategy is used to adjust operation parameters in real time, and a controller parameter is dynamically adjusted by a parameter self-learning mechanism to ensure that the optimal control effect is achieved under different load conditions. The present invention effectively improves the operation efficiency of the boiler system, reduces NOx emissions, and enhances the intelligence and adaptive capabilities of the system.

Description

Technical Field

[0001] The present invention relates to the technical field of boiler system control, and in particular to an optimization control method and a control system for a flue gas recirculation and boiler coupling system. Background Art

[0002] The operating efficiency and environmental performance of the boiler system directly affect the energy utilization rate and pollutant emission level. Among them, flue gas recirculation technology is widely used in boiler systems as an effective means of emission reduction. By reintroducing part of the flue gas into the combustion process, flue gas recirculation technology can reduce the combustion temperature and reduce the generation of nitrogen oxides (NOx), thereby reducing pollutant emissions.

[0003] Existing boiler control systems usually adopt fixed control strategies and cannot adapt to the dynamic changes of boiler load in real time, resulting in low combustion efficiency and increased energy consumption. In addition, the optimization algorithms in the existing technology are mostly single optimization targets, which fail to fully consider the comprehensive balance between NOx emissions, combustion efficiency and energy consumption, and are prone to fall into local optimality and cannot achieve global optimal control, making it difficult for existing boiler systems to simultaneously meet the requirements of efficient combustion, low emissions and low energy consumption in actual operation, leaving a large room for optimization. Summary of the invention

[0004] Based on the above objectives, the present invention provides an optimization control method and control system for a flue gas recirculation and boiler coupling system.

[0005] The optimization control method of the flue gas recirculation and boiler coupling system comprises the following steps:

[0006] S1. Data collection: Multiple sensors are installed at key locations of the boiler system to collect real-time boiler operation data, including flue gas composition, flow rate, temperature, and data related to boiler combustion status. The sensors include flue gas analyzers, flow meters, temperature sensors, and oxygen content detectors;

[0007] S2. Data preprocessing: preprocess the collected data, use data cleaning algorithm to remove noise, use interpolation algorithm to fill in the missing parts of the data, and perform data standardization;

[0008] S3. Model building: Based on the pre-processed data, a multivariable dynamic model of the flue gas recirculation and boiler coupling system is constructed using a machine learning algorithm. The multivariable dynamic model includes a boiler combustion efficiency prediction model, a NOx emission prediction model, and a system energy efficiency prediction model;

[0009] S4. Optimization target setting: According to the operation requirements of the boiler system, set multi-objective optimization targets, including minimizing NOx emissions, maximizing combustion efficiency and minimizing energy consumption, and solve the comprehensive optimization objective function through linear weighted solution;

[0010] S5. Optimization algorithm design: A hybrid optimization algorithm based on a combination of particle swarm optimization algorithm and genetic algorithm is used to globally optimize the model and find the optimal flue gas recirculation flow rate and recirculation ratio;

[0011] S6. Real-time control: Based on the optimal control parameters obtained by the optimization algorithm, the operating parameters of the flue gas recirculation system are adjusted in real time through an adaptive control strategy to ensure that the optimal control effect can be achieved under different boiler loads.

[0012] Optionally, the S1 specifically includes:

[0013] S11, sensor placement: multiple sensors are placed at key locations of the boiler system, including the boiler furnace inlet and outlet, the recirculation flue gas introduction point, the boiler combustion chamber, the economizer outlet, the rear of the air preheater and the flue gas duct. The sensors include flue gas analyzers, flow meters, temperature sensors and oxygen content detectors;

[0014] S12, sensor calibration: calibrating the multiple sensors, including calibrating the flue gas analyzer using standard gas, calibrating the flow meter using a known flow standard, and calibrating the temperature sensor using a temperature calibrator to ensure the accuracy of data collection;

[0015] S13, data acquisition frequency setting: set the data acquisition frequency of the sensor according to the boiler operating conditions.

[0016] Optionally, the S2 specifically includes:

[0017] S21, data cleaning: removing noise from the collected data;

[0018] S22, data interpolation: use linear interpolation algorithm to fill in the missing data;

[0019] S23, data standardization: standardize the cleaned and interpolated data.

[0020] Optionally, the S3 specifically includes:

[0021] S31, data set preparation: based on the data preprocessed in step S2, construct a data set containing multiple variables (such as smoke composition, flow rate, temperature and oxygen content), and divide the data set into a training set and a test set;

[0022] S32, multivariable dynamic model construction: Multilayer perceptron neural network is used as the machine learning algorithm to construct a multivariable dynamic model of the flue gas recirculation and boiler coupling system;

[0023] S33, boiler combustion efficiency prediction model construction: Under the framework of multivariable dynamic model, the boiler combustion efficiency prediction model is constructed using support vector regression algorithm;

[0024] S34, NOx emission prediction model construction: Under the framework of multivariate dynamic model, the NOx emission prediction model is constructed using random forest regression algorithm;

[0025] S35, system energy efficiency prediction model construction: Under the multivariable dynamic model framework, a long short-term memory network is used to construct a system energy efficiency prediction model.

[0026] Optionally, the S4 specifically includes:

[0027] S41, multi-objective optimization objective function setting: based on the operation requirements of the boiler system, multi-objective optimization objectives are set, wherein the optimization objectives include minimizing NOx emissions, maximizing boiler combustion efficiency, and minimizing system energy consumption;

[0028] S42, linear weighted combination of objective functions: using a linear weighted method, the optimization objectives are combined to form a comprehensive optimization objective function, which is expressed as:

[0029] ;

[0030] in, To comprehensively optimize the objective function, , and are the weight coefficients of NOx emission, combustion efficiency and energy consumption, and satisfy 1;

[0031] S43, determination of weight coefficients: determining weight coefficients of various objective functions by analyzing boiler system operation data and evaluating the importance of different optimization objectives;

[0032] S44, comprehensive optimization objective function solution: After setting the weight coefficients of each objective function, the comprehensive optimization objective function is solved by linear weighting , to determine the optimal combination of flue gas recirculation parameters.

[0033] Optionally, the S5 specifically includes:

[0034] S51, selection and initialization of optimization algorithm: a hybrid optimization algorithm based on the combination of particle swarm optimization algorithm and genetic algorithm is used for global optimization;

[0035] S52, Particle Swarm Optimization Algorithm: Use the particle swarm optimization algorithm to update the speed and position of the particles to gradually approach the optimal solution;

[0036] S53, Genetic Algorithm Optimization: Improve the diversity of solutions and avoid local optimality through the selection, crossover and mutation operations of genetic algorithms;

[0037] S54, selection and convergence of the optimal solution: evaluate and select the optimal solution through an iterative optimization process until convergence and the optimal flue gas recirculation parameters are found.

[0038] Optionally, the S6 specifically includes:

[0039] S61, initialization and application of control parameters: input the obtained optimal flue gas recirculation parameters as initial control parameters into the boiler control system to start the preliminary operation of the system;

[0040] S62, Adaptive control strategy design: Adopt adaptive control strategy to adjust the operating parameters of flue gas recirculation system in real time through PID (proportional-integral-differential) controller;

[0041] S63, real-time monitoring and parameter adjustment: real-time monitoring of boiler operating status, including flue gas composition, flow, temperature and combustion efficiency, and dynamic adjustment of PID controller parameters based on monitoring data , , , ensuring that the system adapts to different load conditions and external disturbances;

[0042] S64, real-time implementation and feedback of control strategy: based on the adjustment results of the PID controller, the flue gas recirculation flow rate and recirculation ratio are adjusted in real time.

[0043] Optionally, dynamically adjusting the parameters of the PID controller based on the monitoring data in S63 includes:

[0044] S631, parameter self-learning mechanism design: using the real-time collected boiler operation data, combined with the machine learning algorithm, the parameters of the PID controller are self-learned and optimized, and the parameters of the PID controller are dynamically updated through the recursive least squares method or adaptive gradient algorithm. The parameter update formula is expressed as:

[0045] ;

[0046] in, is the PID controller parameter vector, including , , is the learning rate, is the target loss function;

[0047] S632, real-time parameter adjustment and feedback implementation: according to the results of parameter self-learning, the parameter values ​​of the PID controller are adjusted dynamically in real time , .

[0048] Optionally, the target loss function in the parameter self-learning mechanism of S631 includes:

[0049] Construction of loss function: The target loss function consists of the weighted sum of the control error and the regularization term of the parameter, and the expression is:

[0050] ;

[0051] in, Indicates The control error within the control cycle is the difference between the set value and the actual value. is the total number of control cycles, Indicates PID controller parameters, including the proportional coefficient , integral coefficient and the differential coefficient , is the regularization coefficient, which is used to control the size of the parameter to avoid overfitting;

[0052] Loss function optimization: By minimizing the target loss function, the parameters of the PID controller are dynamically adjusted during the parameter self-learning process. :

[0053] Data acquisition module: used to arrange multiple sensors at key locations of the boiler system to collect boiler operation data in real time, including flue gas composition, flow rate, temperature and boiler combustion status data;

[0054] Data preprocessing module: used to preprocess the collected data, including data cleaning, interpolation and standardization;

[0055] Multivariable dynamic modeling module: Based on the preprocessed data, a multivariable dynamic model of the flue gas recirculation and boiler coupling system is constructed using a machine learning algorithm. The model includes a boiler combustion efficiency prediction model, a NOx emission prediction model, and a system energy efficiency prediction model.

[0056] Optimization target setting module: used to set multi-objective optimization targets, including minimizing NOx emissions, maximizing combustion efficiency and minimizing energy consumption, and constructing a comprehensive optimization objective function through a linear weighted method;

[0057] Hybrid optimization algorithm module: It uses a hybrid optimization algorithm based on a combination of particle swarm optimization algorithm and genetic algorithm to perform global optimization on multi-objective optimization targets and determine the optimal flue gas recirculation flow rate and recirculation ratio;

[0058] Adaptive control module: Based on the optimal control parameters obtained by the optimization algorithm, the PID controller is used to adjust the operating parameters of the flue gas recirculation system in real time through an adaptive control strategy to ensure the optimal control effect under different boiler loads. The adaptive control module includes a parameter self-learning unit, which adjusts the parameters of the PID controller in real time through a parameter self-learning mechanism to optimize the performance of the control system.

[0059] Beneficial effects of the present invention:

[0060] The present invention uses an optimization control method combining a multivariable dynamic model and a PID controller to monitor and adjust key operating parameters of the boiler system in real time, such as flue gas composition, flow rate and temperature. Through an adaptive control strategy and parameter self-learning mechanism, the system can automatically adjust under different load conditions to ensure the maximization of boiler combustion efficiency and the minimization of system energy consumption. As a result, the overall operating efficiency of the system is significantly improved, while maintaining a high degree of operating stability and reducing performance fluctuations caused by load fluctuations.

[0061] The present invention achieves a significant reduction in NOx emissions by setting multi-objective optimization goals and using a hybrid optimization algorithm to accurately control NOx emissions. The optimization algorithm comprehensively considers the boiler's combustion efficiency and system energy consumption, controls NOx emissions to the lowest level while ensuring energy utilization, ensures that the boiler system meets strict environmental regulations, and provides technical support for environmental compliance.

[0062] The present invention introduces a parameter self-learning mechanism and a real-time feedback adjustment strategy, which enables the system to automatically optimize control parameters based on real-time collected data. This intelligent feature enables the system to autonomously adapt to changes in the external environment and different boiler operating conditions, reducing the need for manual intervention and improving the system's autonomous operation capability and robustness. In addition, through the application of machine learning algorithms, the system can continuously learn and improve its control strategy, further improving control accuracy and overall system performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0064] Figure 1 A schematic diagram of a method flow of an embodiment of the present invention;

[0065] Figure 2 Schematic diagram of a system flow of an embodiment of the present invention. DETAILED DESCRIPTION

[0066] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0067] It should be noted that the references to "one embodiment", "embodiment", "exemplary embodiments", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0068] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0069] like Figure 1 As shown, the optimization control method of the flue gas recirculation and boiler coupling system includes the following steps:

[0070] S1. Data collection: Multiple sensors are installed at key locations of the boiler system to collect real-time boiler operation data, including flue gas composition, flow rate, temperature, and data related to boiler combustion status. The sensors include flue gas analyzers, flow meters, temperature sensors, and oxygen content detectors;

[0071] S2. Data preprocessing: preprocess the collected data, use data cleaning algorithm to remove noise, use interpolation algorithm to fill in the missing parts of the data, and perform data standardization;

[0072] S3. Model building: Based on the pre-processed data, a multivariable dynamic model of the flue gas recirculation and boiler coupling system is constructed using a machine learning algorithm. The multivariable dynamic model includes a boiler combustion efficiency prediction model, a NOx emission prediction model, and a system energy efficiency prediction model;

[0073] S4. Optimization target setting: According to the operation requirements of the boiler system, set multi-objective optimization targets, including minimizing NOx emissions, maximizing combustion efficiency and minimizing energy consumption, and solve the comprehensive optimization objective function through linear weighted solution;

[0074] S5. Optimization algorithm design: A hybrid optimization algorithm based on a combination of particle swarm optimization algorithm and genetic algorithm is used to globally optimize the model and find the optimal flue gas recirculation flow rate and recirculation ratio;

[0075] S6. Real-time control: Based on the optimal control parameters obtained by the optimization algorithm, the operating parameters of the flue gas recirculation system are adjusted in real time through an adaptive control strategy to ensure that the optimal control effect can be achieved under different boiler loads.

[0076] S1 specifically includes:

[0077] S11, sensor layout: multiple sensors are arranged at key locations of the boiler system, including the boiler furnace inlet and outlet, the recirculation flue gas introduction point, the boiler combustion chamber, the economizer outlet, the rear of the air preheater and the flue gas duct. The sensors include flue gas analyzers, flow meters, temperature sensors and oxygen content detectors. The specific installation is as follows:

[0078] Boiler furnace inlet and outlet: Install temperature sensors and flue gas analyzers to monitor the composition and temperature of incoming and outgoing flue gas in real time to evaluate combustion conditions and emission characteristics;

[0079] Recirculation flue gas introduction point: install flow meters and temperature sensors to monitor the flow and temperature of the recirculation flue gas to ensure the optimization of the recirculation ratio;

[0080] In the boiler combustion chamber: install an oxygen content detector to monitor the oxygen concentration in the combustion chamber in real time to ensure the completeness of combustion;

[0081] At the economizer outlet: install a temperature sensor to monitor the flue gas temperature after preheating to evaluate the heat exchange efficiency of the boiler;

[0082] After the air preheater: install temperature sensors and flue gas analyzers to monitor the flue gas temperature and composition after the air preheater and optimize the performance of the preheater;

[0083] Smoke exhaust duct: Install flue gas analyzers and flow meters to monitor the final flue gas composition and flow rate to ensure emission compliance;

[0084] S12, sensor calibration: calibrate multiple sensors, including calibrating the flue gas analyzer with standard gas, calibrating the flow meter with known flow standards, and calibrating the temperature sensor with a temperature calibrator to ensure the accuracy of data collection;

[0085] S13, data collection frequency setting: set the data collection frequency of the sensor according to the boiler operating conditions. The default frequency is once per second, and the collection frequency can be adjusted dynamically according to the system;

[0086] By rationally arranging sensors at key locations in the boiler system, calibrating the sensors and setting the data acquisition frequency, the accuracy and reliability of system data acquisition are ensured, laying the foundation for subsequent data processing and optimized control.

[0087] S2 specifically includes:

[0088] S21, data cleaning: remove noise from the collected data using a data cleaning algorithm, expressed as:

[0089] ;

[0090] in, is the original data, is the noise signal, is the noise attenuation coefficient;

[0091] S22, data interpolation: linear interpolation algorithm is used to fill the missing data. The interpolation formula is expressed as:

[0092] ;

[0093] in, and are adjacent valid data points, is the interpolated data;

[0094] S23, data standardization: standardize the cleaned and interpolated data. The standardization formula is as follows:

[0095] ;

[0096] in, is the mean value of the data, is the standard deviation of the data;

[0097] Through data cleaning, interpolation and standardization, we can effectively remove noise, fill in data gaps and standardize data, providing high-quality data input for subsequent model building.

[0098] S3 specifically includes:

[0099] S31, data set preparation: based on the data preprocessed in step S2, a data set containing multiple variables (such as flue gas composition, flow rate, temperature and oxygen content) is constructed, and the data set is used to capture the dynamic relationship between different variables in the boiler system, and the data set is divided into a training set and a test set;

[0100] S32, multivariable dynamic model construction: Multilayer perceptron neural network is used as the machine learning algorithm to build a multivariable dynamic model of the flue gas recirculation and boiler coupling system. The multivariable dynamic model predicts the boiler's combustion efficiency, NOx emissions and system energy efficiency by inputting historical data. The specific steps are as follows:

[0101] (1) Input layer definition: The input layer receives the preprocessed multivariate data set ,Each input feature (such as smoke composition, flow, temperature, etc.) corresponds to a neuron, and the input data includes the current time step Data and previous The data of time steps is used to capture the dynamic characteristics of the system. The input vector form is expressed as:

[0102] ,in is the number of input features;

[0103] (2) Hidden layer design: A multilayer perceptron contains multiple hidden layers, each layer consists of a number of neurons, and the neurons in each hidden layer are connected through a weight matrix And activation function (such as ReLU or tanh) to process the input data, the output of the hidden layer is:

[0104] ;

[0105] in, It is The output of the layer, is the activation function, and are the weight matrix and bias vector respectively;

[0106] (3) Output layer definition: The output layer converts the final output of the hidden layer into the multivariate prediction result of the system. , the number of neurons in the output layer is consistent with the number of variables predicted by the model (such as boiler combustion efficiency, NOx emissions and system energy efficiency), and the output results are expressed as:

[0107] ;

[0108] in, and is the weight matrix and bias vector of the output layer;

[0109] (4) Model training: Use the back propagation algorithm and Adam optimizer to train the multivariate dynamic model based on the training set data. The goal is to minimize the loss function between the predicted value and the actual value. The loss function is defined as: ;

[0110] in, is the actual value, is the predicted value, is the sample size, are model parameters;

[0111] S33, boiler combustion efficiency prediction model construction: Under the framework of multivariable dynamic model, support vector regression algorithm is used to construct boiler combustion efficiency prediction model. The specific steps are as follows:

[0112] (1) Input variable selection: Select input variables that are closely related to boiler combustion efficiency, such as fuel type, oxygen content, flue gas temperature, etc. The input vector form is expressed as: ,in, is the number of features selected;

[0113] (2) Model construction, define the objective function of support vector regression, expressed as:

[0114] ;

[0115] in, is the weight vector, is the bias term, is the penalty coefficient, and is the slack variable;

[0116] Introducing kernel function To handle nonlinear relationships, the prediction output is expressed as:

[0117] ;

[0118] in, is the Lagrange multiplier, is the support vector;

[0119] (3) Model training and validation: Use the training set data to train the boiler combustion efficiency prediction model, determine the model parameters by minimizing the objective function, use cross-validation to evaluate the model performance, and adjust the penalty coefficient through grid search. and kernel function parameters;

[0120] S34, NOx emission prediction model construction: Under the multivariate dynamic model framework, the random forest regression algorithm is used to construct the NOx emission prediction model. The specific steps are as follows:

[0121] (1) Input variable selection: Select input variables that are closely related to NOx emissions, such as fuel flow, oxygen content, furnace temperature, and flue gas flow. The input vector form is expressed as: ,in, is the number of features selected;

[0122] (2) Model construction: The random forest consists of multiple decision trees. Each tree is trained using a subset randomly selected from the training set data. Each decision tree is constructed by recursively partitioning the data in the feature space and is used to predict NOx emissions.

[0123] (3) Decision tree construction process: random selection of the features indivual( , and select the optimal split point among these features, divide the data into two parts based on the split point, and recursively repeat this process until the stopping condition (such as the maximum depth of the tree or the number of samples in the leaf node) is met. The output of the random forest is the average of the outputs of each decision tree, expressed as:

[0124] ;

[0125] in, is the number of decision trees, It is The output of the tree;

[0126] Model training and validation: Build each decision tree using the training set data, randomly generate each tree in the forest during model training, use cross-validation to evaluate model performance, and optimize the model by adjusting the number and depth of decision trees;

[0127] S35, system energy efficiency prediction model construction: Under the multivariable dynamic model framework, a long short-term memory network is used to construct a system energy efficiency prediction model. The specific steps are as follows:

[0128] (1) Input data preparation: Select input variables that are closely related to system energy efficiency, such as total flue gas volume, combustion efficiency, and tin furnace outlet temperature, and organize them into time series data. The input series form is expressed as:

[0129] ,in, is the number of input variables;

[0130] (2) Model construction: The LSTM network consists of an input layer, a LSTM hidden layer, and an output layer. The LSTM network unit controls the flow of information through three gates (input gate, forget gate, and output gate) to solve the problem of long-term dependence.

[0131] Input gate: controls the input of new information;

[0132] Forget gate: determines how much past information to forget;

[0133] Output gate: determines how much information to pass to the next time step;

[0134] The hidden layer output of the long short-term memory network is expressed as:

[0135] ;

[0136] in, is the output of the hidden layer, is the output of the output gate, is the cell state, is the element-wise product;

[0137] (3) Model training and verification: The system energy efficiency prediction model is trained using the training set data, and the parameters are adjusted using the back propagation through time (BPTT) algorithm. After the model training is completed, the model performance is verified using the test set data, and the training parameters are adjusted based on the verification results.

[0138] By adopting machine learning algorithms such as multilayer perceptron neural network, support vector regression, random forest regression and long short-term memory network, a multivariable dynamic model of the system and its included prediction sub-model were constructed, which achieved accurate prediction of the key performance indicators of the boiler and provided strong data support and theoretical basis for system optimization control.

[0139] S4 specifically includes:

[0140] S41, multi-objective optimization objective function setting: based on the operation requirements of the boiler system, multi-objective optimization objectives are set, and the optimization objectives include minimizing NOx emissions, maximizing boiler combustion efficiency and minimizing system energy consumption, among which;

[0141] Minimize NOx emissions ( :Meet environmental protection standards by reducing NOx emissions from boilers;

[0142] Maximize boiler combustion efficiency ( ): Improve the thermal efficiency of the boiler by optimizing the combustion process;

[0143] Minimize system energy consumption ( ): Improve energy utilization by reducing the overall energy consumption of the system;

[0144] S42, linear weighted combination of objective functions: using a linear weighted method, the optimization objectives are combined to form a comprehensive optimization objective function, which is expressed as:

[0145] ;

[0146] in, To comprehensively optimize the objective function, , and are the weight coefficients of NOx emission, combustion efficiency and energy consumption, and satisfy 1;

[0147] S43, determination of weight coefficients: determining weight coefficients of various objective functions by analyzing boiler system operation data and evaluating the importance of different optimization objectives;

[0148] The value can be set according to the requirements of environmental regulations on NOx emissions;

[0149] The value of can be adjusted according to the priority of combustion efficiency to ensure efficient operation of the boiler;

[0150] The value of can be set according to the needs of energy saving and consumption reduction, especially in scenarios with high energy costs;

[0151] S44, comprehensive optimization objective function solution: After setting the weight coefficients of each objective function, the comprehensive optimization objective function is solved by linear weighting To determine the optimal combination of flue gas recirculation parameters (including flow rate and ratio), the steps are as follows:

[0152] (1) Initialization: setting the initial weight coefficients and parameter values;

[0153] (2) Iterative solution: Use gradient descent or other optimization algorithms to solve the objective function Iterate the solution until a solution is found that Achieving minimum or maximum parameter combinations;

[0154] (3) Optimal solution selection: Determine the optimal solution. The minimized parameter combination is the optimal operating condition of the system;

[0155] By setting multi-objective optimization goals and adopting a linear weighted method, different optimization objective functions are integrated, and finally by solving the comprehensive objective function, the optimal operating parameter combination of the boiler system under the conditions of meeting the NOx emission, combustion efficiency and energy consumption goals is obtained, which provides a basis for the overall optimization control of the system.

[0156] S5 specifically includes:

[0157] S51, selection and initialization of optimization algorithm: a hybrid optimization algorithm based on a combination of particle swarm optimization algorithm and genetic algorithm is used for global optimization, where;

[0158] The particle swarm optimization algorithm is used for global search. The initial particle swarm is randomly generated in the entire search space, and each particle represents a possible combination of flue gas recirculation flow rate and recirculation ratio.

[0159] Genetic algorithms are used for local search, and crossover and mutation operations are used to enhance the diversity of search and avoid falling into local optimality;

[0160] S52, Particle Swarm Optimization Algorithm: Use the particle swarm optimization algorithm to update the speed and position of the particles to gradually approach the optimal solution. The specific steps are as follows:

[0161] Particle velocity and position initialization: The initial velocity and position of each particle are randomly determined in the search space. The position of the particle represents the specific value of the flue gas recirculation flow rate and ratio, and the velocity represents the change trend.

[0162] Fitness calculation: by comprehensively optimizing the objective function , calculate the fitness value of each particle to measure the degree to which it meets the optimization goal;

[0163] Speed ​​and position update: The particle speed and position update formula is used to adjust the position of the particle swarm to move closer to the optimal solution. The update formula is expressed as:

[0164] ;

[0165] ;

[0166] in, is the particle velocity, is the particle position, is the best historical position of the particle, is the global best position, is the inertia weight, and is the learning factor, and is a random number;

[0167] S53, genetic algorithm optimization: Improve the diversity of solutions and avoid local optimality through the selection, crossover and mutation operations of genetic algorithms. The specific steps are as follows:

[0168] Selection operation: According to the fitness value, select the particle with the highest fitness as the parent particle to generate the next generation;

[0169] Crossover operation: Through the crossover operation, some features of the parent particles are combined to generate new child particles. The particles after the crossover show new possible solutions in the feature space.

[0170] Mutation operation: Through mutation operation, some characteristics of the offspring particles are randomly changed to increase the diversity of solutions and avoid falling into local optimality;

[0171] S54, selection and convergence of the optimal solution: evaluate and select the optimal solution through an iterative optimization process until convergence and the optimal flue gas recirculation parameters are found. The steps are as follows:

[0172] Fitness evaluation: The fitness of the new particles generated by the genetic algorithm is evaluated and compared with the current global optimal solution. If the fitness of the new particles is better than the current optimal solution, the optimal solution is updated;

[0173] Iteration and convergence judgment: Repeat the alternating iteration of particle swarm optimization and genetic algorithm until the fitness value reaches the preset threshold or the optimization process converges, and find the optimal combination of flue gas recirculation flow and recirculation ratio;

[0174] Output optimal parameters: After the iteration is completed, the output makes the comprehensive optimization objective function Achieve the optimal combination of flue gas recirculation parameters;

[0175] By combining the hybrid optimization strategy of particle swarm optimization algorithm and genetic algorithm and making full use of the advantages of global search and local search, the global optimization of the flue gas recirculation flow and ratio in the boiler system is achieved, thereby ensuring that the system can achieve the optimal control effect under different loads.

[0176] S6 specifically includes:

[0177] S61, initialization and application of control parameters: input the obtained optimal flue gas recirculation parameters as initial control parameters into the boiler control system to start the preliminary operation of the system;

[0178] S62, Adaptive control strategy design: Adopt adaptive control strategy, adjust the operating parameters of flue gas recirculation system in real time through PID (proportional-integral-differential) controller to cope with changes in boiler load and external disturbances, and ensure that the system can maintain optimal control effect under different loads;

[0179] Output of PID controller Defined as:

[0180] ;

[0181] in, is the control error between the set value and the actual value, , , are the proportional, integral and differential coefficients respectively;

[0182] S63, real-time monitoring and parameter adjustment: real-time monitoring of boiler operating status, including flue gas composition, flow, temperature and combustion efficiency, and dynamic adjustment of PID controller parameters based on monitoring data , , , ensuring that the system adapts to different load conditions and external disturbances;

[0183] S64, real-time implementation and feedback of control strategy: Based on the adjustment results of the PID controller, the flue gas recirculation flow rate and recirculation ratio are adjusted in real time to ensure that the system operates in the optimal state. The control error is continuously monitored and corrected through the feedback mechanism to ensure the stability and efficiency of the boiler system under all load conditions;

[0184] Through adaptive control strategy and real-time monitoring feedback mechanism, dynamic adjustment and optimal control of the flue gas recirculation system under different boiler load conditions are realized, ensuring efficient and stable operation of the system.

[0185] In S63, the parameters of the PID controller are dynamically adjusted based on the monitoring data, including:

[0186] S631, parameter self-learning mechanism design: using the real-time collected boiler operation data, combined with the machine learning algorithm, the parameters of the PID controller are self-learned and optimized, and the parameters of the PID controller are dynamically updated through the recursive least squares method or adaptive gradient algorithm. The parameter update formula is expressed as:

[0187] ;

[0188] in, is the PID controller parameter vector, including , , is the learning rate, is the target loss function;

[0189] S632, real-time parameter adjustment and feedback implementation: according to the results of parameter self-learning, the parameter values ​​of the PID controller are adjusted dynamically in real time , , ensuring that the controller can adapt to changes in boiler operating conditions and further improve the control accuracy and response speed of the system;

[0190] The controller output will be adjusted in real time based on the feedback mechanism to ensure the best performance of the system under different load conditions;

[0191] By introducing parameter self-learning mechanism and real-time feedback adjustment strategy, the adaptability and accuracy of PID controller in boiler system are improved, thus achieving more optimized control effect.

[0192] The objective loss function in the S631 parameter self-learning mechanism includes:

[0193] Construction of loss function: target loss function It consists of the weighted sum of the control error and the regularization term of the parameter, and the expression is:

[0194] ;

[0195] in, Indicates The control error within the control cycle is the difference between the set value and the actual value. is the total number of control cycles, Indicates PID controller parameters, including the proportional coefficient , integral coefficient and the differential coefficient , is the regularization coefficient, which is used to control the size of the parameter to avoid overfitting;

[0196] Loss function optimization: By minimizing the objective loss function , dynamically adjust the parameters of the PID controller during the parameter self-learning process , ensuring that the control system achieves optimal control accuracy and system stability under different boiler loads.

[0197] like Figure 2 As shown, the optimization control system of the flue gas recirculation and boiler coupling system is used to implement the above-mentioned optimization control method of the flue gas recirculation and boiler coupling system, and includes the following modules:

[0198] Data acquisition module: used to arrange multiple sensors at key locations of the boiler system to collect boiler operation data in real time, including flue gas composition, flow rate, temperature and boiler combustion status data;

[0199] Data preprocessing module: used to preprocess the collected data, including data cleaning, interpolation and standardization;

[0200] Multivariable dynamic modeling module: Based on the preprocessed data, a multivariable dynamic model of the flue gas recirculation and boiler coupling system is constructed using a machine learning algorithm. The model includes a boiler combustion efficiency prediction model, a NOx emission prediction model, and a system energy efficiency prediction model.

[0201] Optimization target setting module: used to set multi-objective optimization targets, including minimizing NOx emissions, maximizing combustion efficiency and minimizing energy consumption, and constructing a comprehensive optimization objective function through a linear weighted method;

[0202] Hybrid optimization algorithm module: It uses a hybrid optimization algorithm based on a combination of particle swarm optimization algorithm and genetic algorithm to perform global optimization on multi-objective optimization targets and determine the optimal flue gas recirculation flow rate and recirculation ratio;

[0203] Adaptive control module: Based on the optimal control parameters obtained by the optimization algorithm, the PID controller is used and the operating parameters of the flue gas recirculation system are adjusted in real time through the adaptive control strategy to ensure the optimal control effect under different boiler loads. The adaptive control module includes a parameter self-learning unit. The parameter self-learning unit adjusts the parameters of the PID controller in real time through the parameter self-learning mechanism, optimizes the performance of the control system, and ensures the robustness and accuracy of the system under different working conditions.

[0204] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0205] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An optimization control method for a flue gas recirculation and boiler coupling system, characterized in that: The following steps are involved: S1. Data collection: Multiple sensors are installed at key locations of the boiler system to collect real-time boiler operation data, including flue gas composition, flow rate, temperature and boiler combustion status data. The sensors include flue gas analyzers, flow meters, temperature sensors and oxygen content detectors; S2. Data preprocessing: preprocess the collected data, use data cleaning algorithm to remove noise, use interpolation algorithm to fill in the missing parts of the data, and perform data standardization; S3. Model establishment: Based on the data after data preprocessing, a multivariable dynamic model of the flue gas recirculation and boiler coupling system is constructed using a machine learning algorithm. The multivariable dynamic model includes a boiler combustion efficiency prediction model, a NOx emission prediction model, and a system energy efficiency prediction model; S4. Optimization target setting: According to the operation requirements of the boiler system, set multi-objective optimization targets, including minimizing NOx emissions, maximizing combustion efficiency and minimizing energy consumption, and solve the comprehensive optimization objective function through linear weighted solution; S5. Optimization algorithm design: A hybrid optimization algorithm based on a combination of particle swarm optimization algorithm and genetic algorithm is used to globally optimize the model and find the optimal flue gas recirculation flow rate and recirculation ratio; S6. Real-time control: Based on the optimal control parameters obtained by the optimization algorithm, the operating parameters of the flue gas recirculation system are adjusted in real time through an adaptive control strategy to ensure that the optimal control effect can be achieved under different boiler loads.

2. The optimization control method of the flue gas recirculation and boiler coupling system according to claim 1 is characterized in that: The S1 specifically includes: S11, sensor placement: multiple sensors are placed at key locations of the boiler system, including the boiler furnace inlet and outlet, the recirculation flue gas introduction point, the boiler combustion chamber, the economizer outlet, the rear of the air preheater and the flue gas duct. The sensors include flue gas analyzers, flow meters, temperature sensors and oxygen content detectors; S12, sensor calibration: calibrating the multiple sensors, including calibrating the flue gas analyzer using standard gas, calibrating the flow meter using a known flow standard, and calibrating the temperature sensor using a temperature calibrator; S13, data acquisition frequency setting: set the data acquisition frequency of the sensor according to the boiler operating conditions.

3. The optimization control method of the flue gas recirculation and boiler coupling system according to claim 2 is characterized in that: The S2 specifically includes: S21, data cleaning: removing noise from the collected data; S22, data interpolation: use linear interpolation algorithm to fill in the missing data; S23, data standardization: standardize the cleaned and interpolated data.

4. The optimization control method of the flue gas recirculation and boiler coupling system according to claim 3 is characterized in that: The S3 specifically includes: S31, data set preparation: based on the data preprocessed in step S2, construct a data set containing multiple variables, and divide the data set into a training set and a test set; S32, multivariable dynamic model construction: Multilayer perceptron neural network is used as the machine learning algorithm to construct a multivariable dynamic model of the flue gas recirculation and boiler coupling system; S33, boiler combustion efficiency prediction model construction: Under the framework of multivariable dynamic model, the boiler combustion efficiency prediction model is constructed using support vector regression algorithm; S34, NOx emission prediction model construction: Under the multivariate dynamic model framework, the random forest regression algorithm is used to construct the NOx emission prediction model; S35, system energy efficiency prediction model construction: Under the multivariable dynamic model framework, a long short-term memory network is used to construct a system energy efficiency prediction model.

5. The optimization control method of the flue gas recirculation and boiler coupling system according to claim 4, characterized in that: The S4 specifically includes: S41, multi-objective optimization objective function setting: based on the operation requirements of the boiler system, multi-objective optimization objectives are set, wherein the optimization objectives include minimizing NOx emissions, maximizing boiler combustion efficiency, and minimizing system energy consumption; S42, linear weighted combination of objective functions: using a linear weighted method, the optimization objectives are combined to form a comprehensive optimization objective function, which is expressed as: F(w)=w1·f1+w2·f2-w3·f3; Among them, F(w) is the comprehensive optimization objective function, w1, w2 and w3 are the weight coefficients of NOx emission, combustion efficiency and energy consumption, respectively, and satisfy w1+w2+w3=1; S43, determination of weight coefficients: determining weight coefficients of various objective functions by analyzing boiler system operation data and evaluating the importance of different optimization objectives; S44, solving the comprehensive optimization objective function: after setting the weight coefficients of each objective function, the comprehensive optimization objective function F(w) is solved by linear weighting to determine the optimal combination of flue gas recirculation parameters.

6. The optimization control method of the flue gas recirculation and boiler coupling system according to claim 5, characterized in that: The S5 specifically includes: S51, selection and initialization of optimization algorithm: a hybrid optimization algorithm based on the combination of particle swarm optimization algorithm and genetic algorithm is used for global optimization; S52, Particle Swarm Optimization Algorithm: Use the particle swarm optimization algorithm to update the speed and position of the particles to gradually approach the optimal solution; S53, Genetic Algorithm Optimization: Improve the diversity of solutions and avoid local optimality through the selection, crossover and mutation operations of genetic algorithms; S54, selection and convergence of the optimal solution: evaluate and select the optimal solution through an iterative optimization process until convergence and the optimal flue gas recirculation parameters are found.

7. The optimization control method of the flue gas recirculation and boiler coupling system according to claim 6, characterized in that: The S6 specifically includes: S61, initialization and application of control parameters: input the obtained optimal flue gas recirculation parameters as initial control parameters into the boiler control system to start the preliminary operation of the system; S62, Adaptive control strategy design: Adopt adaptive control strategy to adjust the operating parameters of flue gas recirculation system in real time through PID controller; S63, real-time monitoring and parameter adjustment: real-time monitoring of boiler operating status, including flue gas composition, flow, temperature and combustion efficiency, and dynamic adjustment of PID controller parameters based on monitoring data; S64, real-time implementation and feedback of control strategy: based on the adjustment results of the PID controller, the flue gas recirculation flow rate and recirculation ratio are adjusted in real time.

8. The optimization control method of the flue gas recirculation and boiler coupling system according to claim 7, characterized in that: The step S63 dynamically adjusts the parameters of the PID controller based on the monitoring data, including: S631, parameter self-learning mechanism design: using the real-time collected boiler operation data, combined with the machine learning algorithm, the parameters of the PID controller are self-learned and optimized, and the parameters of the PID controller are dynamically updated through the recursive least squares method or adaptive gradient algorithm. The parameter update formula is expressed as: Among them, θ(t) is the PID controller parameter vector at time t, including the proportional coefficient K p , integral coefficient K i and the differential coefficient K d , η is the learning rate, J(θ) is the target loss function; S632, real-time parameter adjustment and feedback implementation: according to the results of parameter self-learning, the parameter value K of the PID controller is adjusted dynamically in real time p , K i and K d .

9. The optimization control method of the flue gas recirculation and boiler coupling system according to claim 8, characterized in that: The target loss function in the S631 parameter self-learning mechanism includes: Construction of loss function: The target loss function consists of the weighted sum of the control error and the regularization term of the parameter, and the expression is: Among them, e i (t) represents the control error in the i-th control cycle, that is, the difference between the set value and the actual value, N is the total number of control cycles, θ j (t) represents the jth PID controller parameter vector at time t, including the proportional coefficient K p , integral coefficient K i Or differential coefficient K d , λ is the regularization coefficient, which is used to control the size of the parameter to avoid overfitting; Loss function optimization: By minimizing the target loss function, the parameters θ of the PID controller are dynamically adjusted during the parameter self-learning process.

10. An optimization control system for a flue gas recirculation and boiler coupling system, used to implement the optimization control method for a flue gas recirculation and boiler coupling system according to any one of claims 1 to 9, characterized in that: Includes the following modules: Data acquisition module: used to arrange multiple sensors at key locations of the boiler system to collect boiler operation data in real time, including flue gas composition, flow, temperature and boiler combustion status data; Data preprocessing module: used to preprocess the collected data, including data cleaning, interpolation and standardization; Multivariable dynamic modeling module: Based on the preprocessed data, a multivariable dynamic model of the flue gas recirculation and boiler coupling system is constructed using a machine learning algorithm. The model includes a boiler combustion efficiency prediction model, a NOx emission prediction model, and a system energy efficiency prediction model. Optimization target setting module: used to set multi-objective optimization targets, including minimizing NOx emissions, maximizing combustion efficiency and minimizing energy consumption, and constructing a comprehensive optimization objective function through a linear weighted method; Hybrid optimization algorithm module: It uses a hybrid optimization algorithm based on a combination of particle swarm optimization algorithm and genetic algorithm to perform global optimization on multi-objective optimization targets and determine the optimal flue gas recirculation flow rate and recirculation ratio; Adaptive control module: Based on the optimal control parameters obtained by the optimization algorithm, the PID controller is used to adjust the operating parameters of the flue gas recirculation system in real time through an adaptive control strategy. The adaptive control module includes a parameter self-learning unit, which adjusts the parameters of the PID controller in real time through a parameter self-learning mechanism to optimize the performance of the control system.

Images