Thermal power plant DCS performance analysis and optimization system based on artificial intelligence

Abstract

The invention relates to the technical field of power plant control systems, in particular to a thermal power plant DCS performance analysis and optimization system based on artificial intelligence, and the system comprises a data collection module which collects the operation data of each control point of a thermal power plant in real time; the data preprocessing module is used for cleaning and standardizing the collected data so as to filter noise and abnormal values; the intelligent analysis module is used for automatically identifying unstable factors and precision deviation in process control by applying a time-delay dynamic modeling method, and dynamically adjusting and optimizing a control strategy through genetic algorithm optimization so as to improve the stability and precision of the process control; and the execution and feedback module is used for applying the optimized control strategy to the DCS system in real time, monitoring the execution effect and feeding back and continuously adjusting according to the system performance. According to the invention, the intelligent control strategy optimization not only improves the stability and precision of the system, but also reduces the energy consumption and cost, and brings significant economic benefits and environmental benefits to the operation management of the thermal power plant.

Description

Technical Field

[0001] The present invention relates to the technical field of power plant control systems, and in particular to a DCS performance analysis and optimization system for a thermal power plant based on artificial intelligence. Background Art

[0002] In the operation and management of modern thermal power plants, distributed control systems (DCS) play a vital role. However, traditional DCS systems have certain limitations in processing large amounts of real-time data, identifying unstable factors and accuracy deviations, and optimizing control strategies. Traditional DCS systems face the following problems:

[0003] Traditional DCS systems have certain difficulties in identifying unstable factors and precision deviations in process control. They lack intelligent analysis methods and algorithms and are unable to discover and solve problems in a timely and accurate manner. The control strategy optimization of traditional DCS systems may be affected by factors such as constraints and inflexible parameter adjustment, making it difficult to dynamically adjust and optimize control strategies, affecting the stability and precision of the system. The optimization algorithms commonly used in traditional DCS systems have limitations such as low search efficiency and local optimal solution problems, which affect the reliability of optimization results and the improvement of system performance. Summary of the invention

[0004] Based on the above objectives, the present invention provides a thermal power plant DCS performance analysis and optimization system based on artificial intelligence.

[0005] The artificial intelligence-based DCS performance analysis and optimization system for thermal power plants includes:

[0006] Data acquisition module: real-time collection of operating data of each control point in the thermal power plant, including temperature, pressure, flow rate, and power output parameters;

[0007] Data preprocessing module: cleans and standardizes the collected data to filter out noise and outliers;

[0008] Intelligent analysis module: Apply the time-delay dynamic modeling method to conduct in-depth analysis of pre-processed data, automatically identify unstable factors and precision deviations in process control, and dynamically adjust and optimize the control strategy through genetic algorithm optimization to improve the stability and precision of process control;

[0009] Execution and feedback module: Apply the optimized control strategy to the DCS system in real time, monitor the execution effect, and make continuous adjustments based on system performance feedback.

[0010] Furthermore, the data acquisition module includes a plurality of sensors distributed at various control points of the thermal power plant for real-time monitoring and recording of temperature, pressure, flow rate and power output parameters, including:

[0011] Temperature sensors: distributed inside the boiler, at the turbine outlet, on the condenser and heat exchange equipment;

[0012] Pressure sensor: installed in the boiler, water supply pipeline, steam pipeline and the inlet and outlet of the steam turbine;

[0013] Flow rate sensor: placed in the main path of water vapor flow, including the feed water pipe, main steam pipe and cooling water system;

[0014] Power output sensor: located at the generator outlet, used to monitor power output parameters, including voltage, current and frequency;

[0015] Vibration sensors: installed on steam turbines, generators and large pumps;

[0016] Emission monitoring sensor: installed in the flue gas emission system to monitor the emission of SO2 and NOx gases.

[0017] Furthermore, the data preprocessing module uses sliding window technology to perform real-time processing on the collected data stream. The data in each window is regarded as a data segment for subsequent processing and analysis. The moving average filter technology is used to smooth the data in the window to eliminate the noise caused by sensor measurement errors. The standardization processing includes unifying the data collected by different sensors into the same dimension and range, and the standardization processing uses the minimum-maximum standardization method.

[0018] Furthermore, the time-delay dynamic modeling method is used to model the operation data of each control point, taking into account the influence of the time-delay factor on stability and accuracy, and specifically adopting a discrete time delay model to model the relationship between input and output as a dynamic system with time delay;

[0019] The parameter estimation algorithm based on the least squares method estimates the parameters of the time-delay dynamic model, including the time-delay parameters and the state space parameters of the dynamic system, by fitting the model and actual data;

[0020] Using the estimated time-lag dynamic model, the actual data is predicted and compared with the actual data to detect abnormal points and irregular changes, so as to automatically identify unstable factors and accuracy deviations;

[0021] Based on the differences between the model prediction results and the actual data, anomaly detection algorithms are used to identify unstable factors and accuracy deviations, and provide optimization suggestions or control strategies.

[0022] Further, the discrete time delay model is expressed as:

[0023] y(t)=f(y(t-1),y(t-2),...,y(td),u(t-1),u(t-2),...,u(td),θ), where y(t) is the output of the system, u(t) is the input of the system, θ is the parameter of the system, d is the length of the time lag, y(t) is represented by the operating data of each control point of the thermal power plant, u(t) is represented by the control signal, and θ is the parameter of the discrete time delay that needs to be estimated.

[0024] Furthermore, the least squares-based parameter estimation algorithm is used to estimate the parameters in the discrete time delay model. There are N groups of observation data (y i ,u i ), the corresponding predicted value is The goal of the least squares method is to minimize the residual sum of squares:

[0025] in:

[0026] By solving the minimization problem, we get the estimated value of the parameter θ.

[0027] Furthermore, the dynamic adjustment and optimization of the control strategy by genetic algorithm optimization specifically includes:

[0028] Initialization of population: The intelligent initialization method is used to generate the initial population. When generating the initial population, the introduction of constraints is considered to ensure the feasibility and rationality of the individual control strategies;

[0029] Coding design: Design a special chromosome coding scheme to adapt to the control strategy of thermal power plants, divide the chromosome into multiple regions, each region corresponds to a control parameter, and consider the coding method of time lag characteristics at the same time, so that the coding is more in line with the actual control needs;

[0030] Fitness evaluation: When evaluating the fitness of individuals in a population, consider the introduction of constraints to ensure that the fitness evaluation not only considers the excellence of the optimization goal, but also considers whether the individual meets the constraints;

[0031] Selection operation: Perform selection operation according to fitness value to ensure that the selected individuals meet the constraints;

[0032] Crossover operation: When performing crossover operation on selected individuals, the particularity of the dedicated coding scheme is taken into account to ensure that the new individuals generated by the crossover operation also meet the constraints;

[0033] Mutation operation: When performing mutation operation on individuals after crossover, it is also necessary to consider the constraints to ensure that the mutated individuals are still feasible solutions;

[0034] Replacement operation: replace individuals in the parent population with newly generated individuals to form a new population;

[0035] Iterative optimization: Repeat the encoding design to replace operations until the set stopping condition is reached.

[0036] Furthermore, the intelligent initialization method generates the initial population by: assuming that the population size is N, the chromosome length is L, the control parameter range is [a, b], and the chromosome of each individual i is represented as

[0037] x i =[x i1 ,x i2 ,...,x iL ], where x ij is the jth gene of the ith individual, satisfying a≤x ij ≤ b;

[0038] The dedicated chromosome encoding scheme is real number encoding, and for time-delay dynamic parameter optimization, real number encoding is used to represent the control parameter value;

[0039] The fitness evaluation also includes defining a fitness function f(x) according to the analysis results of the time-delay dynamic modeling, and considering the constraint condition g(x), the fitness function is defined as:

[0040] f(x)=objective function-λ·constraint function, where λ is the constraint weight, expressed as: f(x)=objective function-λ·|constraint function|.

[0041] Furthermore, the selection operation calculates the selection probability P of each individual i , the selection probability calculation formula is:

[0042]

[0043] The crossover operation formula is:

[0044] x new =α·x parent1 +(1-α)·x parent2 , where α is the crossover parameter, and its value is [0,1];

[0045] The mutation operation formula is:

[0046] x mutated =x old +β·(x max -x min ), where β is the variation parameter, and its value is [-1,1], x max and x min are the maximum and minimum values ​​of the control parameters respectively.

[0047] Furthermore, the genetic algorithm also includes an adaptive parameter adjustment algorithm, specifically including:

[0048] Crossover probability adjustment: According to real-time feedback information, the crossover probability is dynamically adjusted through an adaptive algorithm. The formula for dynamic adjustment of the crossover probability is:

[0049] Among them, P c is the crossover probability after dynamic adjustment, and are the maximum and minimum values ​​of the crossover probability, respectively, and f best is the fitness value of the best individual in the population, f avg is the average fitness value of the population, f worst It is the fitness value of the worst individual in the population. The crossover probability is dynamically adjusted according to the fitness distribution of the population. When the fitness difference is large, the crossover probability is reduced to maintain the diversity of the population; when the fitness difference is small, the crossover probability is increased to accelerate the search process.

[0050] Mutation probability adjustment: According to real-time feedback information, the mutation probability is dynamically adjusted through an adaptive algorithm. The mutation probability dynamic adjustment formula is:

[0051] Among them, P m is the mutation probability after dynamic adjustment, and are the maximum and minimum values ​​of the mutation probability, respectively, and f best is the fitness value of the best individual in the population, f avg is the average fitness value of the population, f worst It is the fitness value of the worst individual in the population. The mutation probability is dynamically adjusted according to the fitness distribution of the population. When the fitness difference is large, the mutation probability increases, increasing the breadth of the search; when the fitness difference is small, the mutation probability decreases, accelerating the search process.

[0052] Beneficial effects of the present invention:

[0053] The present invention can automatically identify unstable factors and precision deviations in process control by real-time collection of operating data of each control point and intelligent analysis, discover problems in a timely manner and optimize control strategies, thereby improving the stability and precision of process control and ensuring the efficiency and safety of thermal power plant operation. The optimized control strategy can more accurately adjust the operating parameters of thermal power plant equipment, so that the system operates in the best state, effectively reducing energy consumption and operating costs, improving resource utilization, and maximizing economic benefits.

[0054] The present invention adopts a time-delay dynamic modeling method for in-depth analysis, takes into account the influence of time-delay factors on stability and accuracy, can simulate the actual control process more accurately, and can accurately identify unstable factors and accuracy deviations by modeling and estimating parameters of actual data, providing a reliable basis for subsequent optimization and adjustment. It introduces dynamic adjustment and optimization of genetic algorithm optimization control strategies, and the control strategy can realize adaptive adjustment and dynamically adjust algorithm parameters according to real-time feedback information to achieve more efficient process control. Intelligent control strategy optimization not only improves the stability and accuracy of the system, but also reduces energy consumption and cost, bringing significant economic and environmental benefits to the operation and management of thermal power plants. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] 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.

[0056] Figure 1 A schematic diagram of the functional modules of the optimization system according to an embodiment of the present invention;

[0057] Figure 2 Schematic diagram of genetic algorithm optimization according to an embodiment of the present invention. DETAILED DESCRIPTION

[0058] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0059] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0060] like Figure 1-Figure 2 As shown in the figure, the DCS performance analysis and optimization system of thermal power plants based on artificial intelligence includes:

[0061] Data acquisition module: real-time collection of operating data of each control point in the thermal power plant, including temperature, pressure, flow rate, and power output parameters;

[0062] Data preprocessing module: cleans and standardizes the collected data to filter out noise and outliers;

[0063] Intelligent analysis module: Apply the time-delay dynamic modeling method to conduct in-depth analysis of pre-processed data, automatically identify unstable factors and precision deviations in process control, and dynamically adjust and optimize the control strategy through genetic algorithm optimization to improve the stability and precision of process control;

[0064] Execution and Feedback Module: Apply the optimized control strategy to the DCS system in real time, monitor the execution effect, and make continuous adjustments based on system performance feedback, providing a user interface that allows operators to monitor system status, adjust parameters, and receive optimization suggestions generated by the system.

[0065] The data acquisition module includes multiple sensors distributed at various control points in the thermal power plant to monitor and record temperature, pressure, flow rate and power output parameters in real time, including:

[0066] Temperature sensors: Distributed inside the boiler, at the turbine outlet, on the condenser and on the heat exchange equipment, these sensors monitor the temperature of each part to ensure that the system operates within a safe and optimal temperature range;

[0067] Pressure sensor: installed in the boiler, water supply pipeline, steam pipeline and the inlet and outlet of the steam turbine. The pressure data at these locations is crucial to the overall pressure balance and operational safety of the monitoring system;

[0068] Flow rate sensors: Placed in the main steam flow path, including feedwater pipes and main steam pipes and cooling water systems, these sensors help monitor the speed and amount of fluid flow, which is very important for adjusting boiler loads and optimizing water chemical treatment processes;

[0069] Power output sensor: located at the generator outlet, used to monitor power output parameters, including voltage, current and frequency;

[0070] Vibration sensors: Installed on steam turbines, generators and large pumps, these sensors are used to monitor the operating status of equipment and detect early signs of mechanical failure in a timely manner;

[0071] Emission monitoring sensor: installed in the flue gas emission system to monitor the emission of SO2 and NOx gases.

[0072] The data preprocessing module uses sliding window technology to process the collected data stream in real time. The data in each window is regarded as a data segment for subsequent processing and analysis. The moving average filter technology is used to smooth the data in the window to eliminate the noise caused by sensor measurement errors. The standardization processing includes unifying the data collected by different sensors into the same dimension and range. The standardization processing uses the minimum-maximum standardization method.

[0073] The time-delay dynamic modeling method is used to model the operation data of each control point, considering the influence of time-delay factors on stability and accuracy. Specifically, a discrete time delay model is used to model the relationship between input and output as a dynamic system with time delay.

[0074] The parameter estimation algorithm based on the least squares method estimates the parameters of the time-delay dynamic model, including the time-delay parameters and the state space parameters of the dynamic system, by fitting the model and actual data;

[0075] Using the estimated time-lag dynamic model, the actual data is predicted and compared with the actual data to detect abnormal points and irregular changes, so as to automatically identify unstable factors and accuracy deviations;

[0076] Based on the differences between the model prediction results and the actual data, anomaly detection algorithms are used to identify unstable factors and accuracy deviations, and provide optimization suggestions or control strategies.

[0077] The discrete time delay model is expressed as:

[0078] y(t)=f(y(t-1),y(t-2),...,y(td),u(t-1),u(t-2),...,u(td),θ), where y(t) is the output of the system, u(t) is the input of the system, θ is the parameter of the system, d is the length of the time lag, y(t) represents the operating data (temperature, pressure, flow rate, power output parameters) of each control point of the thermal power plant, u(t) represents the control signal, and θ is the parameter of the discrete time delay that needs to be estimated.

[0079] The parameter estimation algorithm based on the least squares method is used to estimate the parameters in the discrete time delay model. There are N sets of observation data (y i ,u i ), the corresponding predicted value is The goal of the least squares method is to minimize the residual sum of squares:

[0080] in:

[0081] By solving the minimization problem, we get the estimated value of the parameter θ.

[0082] After discrete time delay modeling and parameter estimation, the predicted output is obtained and the actual observed output y i The goal of anomaly detection is to identify anomalies by comparing the difference between the predicted output and the actual observed output. The anomaly detection method is based on the residual method, which calculates the residual Then determine whether the residual exceeds a certain threshold to determine whether there is an anomaly, calculated as:

[0083] If |e i |>threshold, it is considered abnormal.

[0084] Residual calculation: The residual is the difference between the actual observed value and the predicted value, which represents the deviation between the model prediction and the actual observation. By calculating the residual, the prediction accuracy of the model can be quantified.

[0085] Abnormality judgment: By setting a threshold, the size of the residual is used to determine whether there is an abnormal situation. When the residual exceeds the threshold, it means that the difference between the actual observation value and the model prediction value is large, indicating that unstable factors or accuracy deviations have occurred.

[0086] Genetic algorithm is an evolutionary algorithm suitable for global search and parameter optimization of optimization problems. By optimizing the control strategy through genetic algorithm, the control parameters can be adjusted to achieve the stability and accuracy of process control. The dynamic adjustment and optimization of the control strategy through genetic algorithm optimization specifically includes:

[0087] Initialization of population: The intelligent initialization method is used to generate the initial population. When generating the initial population, the introduction of constraints is considered to ensure the feasibility and rationality of the individual control strategies;

[0088] Coding design: Design a special chromosome coding scheme to adapt to the control strategy of thermal power plants, divide the chromosome into multiple regions, each region corresponds to a control parameter, and consider the coding method of time lag characteristics at the same time, so that the coding is more in line with the actual control needs;

[0089] Fitness evaluation: When evaluating the fitness of individuals in a population, consider the introduction of constraints to ensure that the fitness evaluation not only considers the excellence of the optimization goal, but also considers whether the individual meets the constraints;

[0090] Selection operation: Perform selection operation according to fitness value to ensure that the selected individuals meet the constraints;

[0091] Crossover operation: When performing crossover operation on selected individuals, the particularity of the dedicated coding scheme is taken into account to ensure that the new individuals generated by the crossover operation also meet the constraints;

[0092] Mutation operation: When performing mutation operation on individuals after crossover, it is also necessary to consider the constraints to ensure that the mutated individuals are still feasible solutions;

[0093] Replacement operation: replace individuals in the parent population with newly generated individuals to form a new population;

[0094] Iterative optimization: Repeat the encoding design to replacement operation until the set stopping condition (such as the number of iterations, fitness threshold, etc.) is reached.

[0095] The intelligent initialization method generates the initial population by: assuming that the population size is N, the chromosome length is L, the control parameter range is [a, b], and the chromosome of each individual i is represented by x i =[x i1 ,x i2 ,...,x iL ], where x ij is the jth gene of the ith individual, satisfying a≤x ij ≤ b;

[0096] The dedicated chromosome coding scheme is real number coding. For time-delay dynamic parameter optimization, real number coding is used to represent the control parameter values;

[0097] Fitness evaluation also includes defining the fitness function f(x) based on the analysis results of time-delay dynamic modeling, considering the constraint condition g(x), and defining the fitness function as:

[0098] f(x)=objective function-λ·constraint function, where λ is the constraint weight, expressed as: f(x)=objective function-λ·|constraint function|.

[0099] The selection operation calculates the selection probability P of each individual i , the selection probability calculation formula is:

[0100] The crossover operation formula is:

[0101] x new =α·x parent1 +(1-α)·x parent2 , where α is the crossover parameter, and its value is [0,1];

[0102] The mutation operation formula is:

[0103] x mutated =x old +β·(x max -x min ), where β is the variation parameter, and its value is [-1,1], x max and x min are the maximum and minimum values ​​of the control parameters respectively.

[0104] Genetic algorithms also include adaptive parameter adjustment algorithms, including:

[0105] Crossover probability adjustment: According to real-time feedback information, the crossover probability is dynamically adjusted through an adaptive algorithm. The formula for dynamic adjustment of the crossover probability is:

[0106] Among them, P c is the crossover probability after dynamic adjustment, and are the maximum and minimum values ​​of the crossover probability, respectively, and f best is the fitness value of the best individual in the population, f avg is the average fitness value of the population, f worst It is the fitness value of the worst individual in the population. The crossover probability is dynamically adjusted according to the fitness distribution of the population. When the fitness difference is large, the crossover probability is reduced to maintain the diversity of the population; when the fitness difference is small, the crossover probability is increased to accelerate the search process.

[0107] Mutation probability adjustment: According to real-time feedback information, the mutation probability is dynamically adjusted through an adaptive algorithm. The mutation probability dynamic adjustment formula is:

[0108] Among them, P m is the mutation probability after dynamic adjustment, and are the maximum and minimum values ​​of the mutation probability, respectively, and f best is the fitness value of the best individual in the population, f avg is the average fitness value of the population, f worst It is the fitness value of the worst individual in the population. The mutation probability is dynamically adjusted according to the fitness distribution of the population. When the fitness difference is large, the mutation probability increases, increasing the breadth of the search; when the fitness difference is small, the mutation probability decreases, accelerating the search process.

[0109] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention is limited to these examples. Under the concept of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0110] The present invention is intended to cover all such substitutions, modifications and variations that fall within the broad scope of the claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. The DCS performance analysis and optimization system for thermal power plants based on artificial intelligence is characterized by: include: Data acquisition module: real-time collection of operating data of each control point in the thermal power plant, including temperature, pressure, flow rate, and power output parameters; Data preprocessing module: cleans and standardizes the collected data to filter out noise and outliers; Intelligent analysis module: Apply the time-delay dynamic modeling method to conduct in-depth analysis of pre-processed data, automatically identify unstable factors and precision deviations in process control, and dynamically adjust and optimize the control strategy through genetic algorithm optimization to improve the stability and precision of process control; Execution and feedback module: Apply the optimized control strategy to the DCS system in real time, monitor the execution effect, and make continuous adjustments based on system performance feedback.

2. The artificial intelligence-based DCS performance analysis and optimization system for thermal power plants according to claim 1 is characterized in that: The data acquisition module includes multiple sensors distributed at various control points of the thermal power plant, which are used to monitor and record temperature, pressure, flow rate and power output parameters in real time, including: Temperature sensors: distributed inside the boiler, at the turbine outlet, on the condenser and heat exchange equipment; Pressure sensor: installed in the boiler, water supply pipeline, steam pipeline and the inlet and outlet of the steam turbine; Flow rate sensor: placed in the main path of water vapor flow, including the feed water pipe, main steam pipe and cooling water system; Power output sensor: located at the generator outlet, used to monitor power output parameters, including voltage, current and frequency; Vibration sensors: installed on steam turbines, generators and large pumps; Emission monitoring sensor: installed in the flue gas emission system to monitor the emission of SO2 and NOx gases.

3. The artificial intelligence-based DCS performance analysis and optimization system for thermal power plants according to claim 1 is characterized in that: The data preprocessing module uses sliding window technology to process the collected data stream in real time. The data in each window is regarded as a data segment for subsequent processing and analysis. The moving average filter technology is used to smooth the data in the window to eliminate the noise caused by sensor measurement errors. The standardization processing includes unifying the data collected by different sensors into the same dimension and range. The standardization processing uses the minimum-maximum standardization method.

4. The artificial intelligence-based DCS performance analysis and optimization system for thermal power plants according to claim 1 is characterized in that: The time-delay dynamic modeling method is used to model the operation data of each control point, taking into account the influence of the time-delay factor on stability and accuracy, and specifically adopts a discrete time delay model to model the relationship between input and output as a dynamic system with time delay; The parameter estimation algorithm based on the least squares method estimates the parameters of the time-delay dynamic model, including the time-delay parameters and the state space parameters of the dynamic system, by fitting the model and actual data; Using the estimated time-lag dynamic model, the actual data is predicted and compared with the actual data to detect abnormal points and irregular changes, so as to automatically identify unstable factors and accuracy deviations; Based on the differences between the model prediction results and the actual data, anomaly detection algorithms are used to identify unstable factors and accuracy deviations, and provide optimization suggestions or control strategies.

5. The artificial intelligence-based DCS performance analysis and optimization system for thermal power plants according to claim 4 is characterized in that: The discrete time delay model is expressed as: y(t)=f(y(t-1),y(t-2),...,y(td),u(t-1),u(t-2),...,u(td),θ), where y(t) is the output of the system, u(t) is the input of the system, θ is the parameter of the system, d is the length of the time lag, y(t) is represented by the operating data of each control point of the thermal power plant, u(t) is represented by the control signal, and θ is the parameter of the discrete time delay that needs to be estimated.

6. The artificial intelligence-based DCS performance analysis and optimization system for thermal power plants according to claim 5 is characterized in that: The least squares-based parameter estimation algorithm is used to estimate the parameters in the discrete time delay model. There are N groups of observation data (y i ,u i ), the corresponding predicted value is The goal of the least squares method is to minimize the residual sum of squares: in: By solving the minimization problem, we get the estimated value of the parameter θ.

7. The artificial intelligence-based DCS performance analysis and optimization system for thermal power plants according to claim 1 is characterized in that: The dynamic adjustment and optimization of the control strategy by genetic algorithm optimization specifically includes: Initialization of population: The intelligent initialization method is used to generate the initial population. When generating the initial population, the introduction of constraints is considered to ensure the feasibility and rationality of the individual control strategies; Coding design: Design a special chromosome coding scheme to adapt to the control strategy of thermal power plants, divide the chromosome into multiple regions, each region corresponds to a control parameter, and consider the coding method of time lag characteristics at the same time, so that the coding is more in line with the actual control needs; Fitness evaluation: When evaluating the fitness of individuals in a population, consider the introduction of constraints to ensure that the fitness evaluation not only considers the excellence of the optimization goal, but also considers whether the individual meets the constraints; Selection operation: Perform selection operation according to fitness value to ensure that the selected individuals meet the constraints; Crossover operation: When performing crossover operation on selected individuals, the particularity of the dedicated coding scheme is taken into account to ensure that the new individuals generated by the crossover operation also meet the constraints; Mutation operation: When performing mutation operation on individuals after crossover, it is also necessary to consider the constraints to ensure that the mutated individuals are still feasible solutions; Replacement operation: replace individuals in the parent population with newly generated individuals to form a new population; Iterative optimization: Repeat the coding design to replace operations until the set stopping condition is reached.

8. The artificial intelligence-based DCS performance analysis and optimization system for thermal power plants according to claim 7 is characterized in that: The intelligent initialization method for generating an initial population includes: assuming that the population size is N, the chromosome length is L, the control parameter range is [a, b], and the chromosome of each individual i is represented by x i =[x i1 ,x i2 ,...,x iL ], where x ij is the jth gene of the ith individual, satisfying a≤x ij ≤ b; The dedicated chromosome encoding scheme is real number encoding, and for time-delay dynamic parameter optimization, real number encoding is used to represent the control parameter value; The fitness evaluation also includes defining a fitness function f(x) according to the analysis results of the time-delay dynamic modeling, and considering the constraint condition g(x), the fitness function is defined as: f(x)=objective function-λ·constraint function, where λ is the constraint weight, expressed as: f(x)=objective function-λ·|constraint function|.

9. The artificial intelligence-based DCS performance analysis and optimization system for thermal power plants according to claim 8 is characterized in that: The selection operation calculates the selection probability P of each individual i , the selection probability calculation formula is: The crossover operation formula is: x new =α·x parent1 +(1-α)·x parent2 , where α is the crossover parameter, and its value is [0,1]; The mutation operation formula is: x mutated =x old +β·(x max -x min ), where β is the variation parameter, and its value is [-1,1], x max and x min are the maximum and minimum values ​​of the control parameters respectively.

10. The artificial intelligence-based DCS performance analysis and optimization system for thermal power plants according to claim 9 is characterized in that ,The genetic algorithm also includes an adaptive parameter adjustment algorithm, specifically including: Crossover probability adjustment: According to real-time feedback information, the crossover probability is dynamically adjusted through an adaptive algorithm. The formula for dynamic adjustment of the crossover probability is: Among them, P c is the crossover probability after dynamic adjustment, and are the maximum and minimum values ​​of the crossover probability, respectively, and f best is the fitness value of the best individual in the population, f avg is the average fitness value of the population, f worst It is the fitness value of the worst individual in the population. The crossover probability is dynamically adjusted according to the fitness distribution of the population. When the fitness difference is large, the crossover probability is reduced to maintain the diversity of the population; when the fitness difference is small, the crossover probability is increased to accelerate the search process. Mutation probability adjustment: According to real-time feedback information, the mutation probability is dynamically adjusted through an adaptive algorithm. The mutation probability dynamic adjustment formula is: Among them, P m is the mutation probability after dynamic adjustment, and are the maximum and minimum values ​​of the mutation probability, respectively, and f best is the fitness value of the best individual in the population, f avg is the average fitness value of the population, f worst It is the fitness value of the worst individual in the population. The mutation probability is dynamically adjusted according to the fitness distribution of the population. When the fitness difference is large, the mutation probability increases, increasing the breadth of the search; when the fitness difference is small, the mutation probability decreases, accelerating the search process.

Images