An Optimization Method for Controlling the Gas Flow Inside a Burner

Abstract

The present invention discloses an optimization method for controlling the gas flow inside a burner, which relates to the field of control optimization and includes: Step 1, establishing a gas flow control system model inside the burner according to the roasting temperature requirements of different types of anode carbon blocks; Step 2, improving the umbrella lizard optimization algorithm by improving the individual motion parameter I and introducing the "simulated annealing" strategy; Step 3, selecting a suitable objective function, optimizing the PID controller parameters through the IFLO algorithm, simulating the natural behavior of the umbrella lizard, searching for the best PID control parameters in the search space, and obtaining the optimal PID control method, namely the IFLO-PID control method; Step 4, precisely controlling the gas flow inside the burner through the IFLO-PID control method, and realizing the closed-loop control of the gas flow through the IFLO-PID controller according to the temperature value of the flue area feedback in real time.

Description

Technical Field

[0001] The present invention relates to the field of internal gas flow control of burners, and particularly to an optimization method for internal gas flow control of burners. Background Art

[0002] During the baking process of anode carbon blocks, the optimized control of the gas flow inside the burner is not only a key factor to ensure the baking quality, but also an important means to improve energy efficiency and reduce production costs; due to differences in the structure, material properties, and dimensions of different types of anode carbon blocks, their requirements for baking temperature vary; if the temperature control is inaccurate, it may lead to a decline in the performance of the carbon blocks or an increase in energy consumption; therefore, precisely controlling the gas flow becomes the core means to achieve the target temperature regulation; by optimizing the gas flow, it is possible to ensure that the carbon blocks reach the optimal temperature curve during the baking process, and at the same time avoid energy waste, achieving the purpose of efficient production.

[0003] In the current burner flow control process, the PID control method is mainly adopted; the PID controller (Proportional-Integral-Derivative control) is one of the most common technologies in industrial process control. By adjusting the three parameters of proportional, integral, and derivative, the PID controller can precisely adjust the gas flow to maintain the target temperature in the flue; although the PID control method is widely used in burner flow control and has high stability and reliability, the performance of the PID controller highly depends on the settings of the Kp, Ki, and Kd parameters; different working conditions and system responses require different parameter values, and manually adjusting these parameters often requires a lot of experience and experiments. Sometimes, improper adjustment may lead to system instability or slow response.

[0004] The Frilled Lizard Optimization (FLO) algorithm is a new swarm intelligence optimization algorithm proposed in 2024. This algorithm divides the algorithm into exploration and exploitation stages by imitating the hunting behavior of frilled lizards and their retreat behavior after eating, and seeks the optimal solution to the problem through cooperation and information sharing among individuals; using the frilled lizard optimization algorithm to optimize the PID controller parameters can significantly improve the performance of burner flow control, especially when dealing with complex and changing external environments; although the frilled lizard optimization algorithm can well balance global exploration and local exploitation and has strong global search ability in the initial stage of the algorithm, however, when facing high-dimensional problems, FLO may show instability, and the local search ability of FLO is poor. When approaching the optimal solution, the fine search ability of FLO may be insufficient, resulting in low accuracy of the final solution. Summary of the Invention

[0005] The object of the present invention is: aiming at the deficiencies of the prior art, the present invention proposes an optimization method for controlling the gas flow inside a burner. By improving the Iguana Flock Optimization (IFLO) algorithm to optimize the parameters of the PID controller, precise control of the gas flow inside the burner is achieved, thereby ensuring more precise and reliable temperature control during the baking process of anode carbon blocks.

[0006] To achieve the above object, the present invention adopts the following technical solutions: an optimization method for controlling the gas flow inside a burner, including: improving the Iguana Flock Optimization (IFLO) algorithm and the positional PID control algorithm. The specific steps are as follows:

[0007] Step 1: According to the baking temperature requirements of different types of anode carbon blocks, establish a burner flow control system model, which includes a mathematical model and a simulation model.

[0008] Step 2: Improve the Iguana Flock Optimization algorithm by improving the individual movement parameter I and introducing the "simulated annealing" strategy. The improvement points are as follows:

[0009] S21: When the Iguana Flock Optimization algorithm is in the exploration stage, introduce an adaptive factor to improve the individual movement parameter , improving the self - adaptability of the algorithm;

[0010] S22: When the Iguana Flock Optimization algorithm is in the exploitation stage, introduce the "simulated annealing" strategy to improve the position update formula in the exploitation stage of the algorithm, and improve the local search ability of the algorithm.

[0011] Step 3: Select a suitable objective function, optimize the parameters of the PID controller through the IFLO algorithm, perform algorithm iteration and optimization. By simulating the natural behavior of iguanas, search for the best PID control parameters in the search space, and obtain the optimal PID control method, that is, the IFLO - PID control method.

[0012] Step 4: Precisely control the gas flow inside the burner through the IFLO - PID control method. According to the temperature value of the flue region fed back in real - time, calculate the deviation value e(t) between the target temperature and the actual temperature and input it into the IFLO - PID controller to obtain the control quantity , that is, the gas flow inside the burner, and realize the closed - loop IFLO - PID control of the gas flow.

[0013] Furthermore, the mathematical model of the burner flow control includes: First, according to the baking temperature requirements of different types of anode carbon blocks, establish a temperature demand model; Second, since the size of the gas flow inside the burner directly affects the flue temperature, it is necessary to establish a mathematical model between the gas flow and the heating temperature.

[0014] Furthermore, the temperature requirement model describes that during the reaction process of the burner, the temperature requirement curve includes three stages: the heating stage, the constant temperature stage, and the cooling stage. The temperature requirement curve is:

[0015] (1);

[0016] In Equation (1), is the temperature of the burner at time t, is the initial temperature, is the target temperature, and are the heating and cooling rates respectively, and t is the operation time of the burner.

[0017] Furthermore, the gas flow rate will directly affect the temperature change in the combustion furnace. Therefore, the mathematical model between the gas flow rate and the heating temperature is established as:

[0018] (2);

[0019] In Equation (2), is the temperature of the burner at time t, is the gas flow rate at time t, is the ambient temperature, is the function of the gas flow rate and temperature, is the influence of the ambient temperature and time on the temperature in the furnace, and the meanings of other parameters are the same as above.

[0020] Furthermore, the gas flow rate is adjusted by a PID controller to meet the temperature requirement. The controller dynamically adjusts the gas flow rate according to the error value between the target temperature and the actual temperature. The position-type PID controller model is:

[0021] (3);

[0022] In Equation (3), is the gas flow rate at time t, e(t) is the error value between the target temperature and the actual temperature, , Kp, Ki, and Kd are the proportional, integral, and differential coefficients of the PID controller, and the meanings of other parameters are the same as above.

[0023] Furthermore, to achieve the optimal control of the gas flow rate inside the burner, a suitable objective function needs to be selected. Considering the temperature requirement of the burner and the reaction speed of the gas flow rate control comprehensively, the selected objective function is:

[0024] (4);

[0025] In Equation (4), T is the system operation time, is the target temperature of the burner, is the actual temperature of the burner at time t, is a parameter for adjusting the sensitivity of the non-linear weight, and the meanings of other parameters are the same as above.

[0026] Furthermore, the umbrella lizard optimization algorithm is improved. More specifically, an adaptive factor is introduced into the mathematical model of position update in the exploration stage of the umbrella lizard optimization algorithm to improve the individual motion parameters , enhancing the self-adaptability of the algorithm. After improvement, the mathematical model is:

[0027] (5);

[0028] In formula (5), i = 1,..., nPop, where nPop is the population size, is a function that varies with the iteration number k, represents the attraction between the current individual position and the best individual position, is the best individual position, is the current individual position at the current iteration, represents the random intensity of individual position update, is a random number with a value in the range (0, 1], the mathematical model of is:

[0029] (6);

[0030] In formula (6), is the initial motion parameter intensity value. By introducing a sine function into the mathematical model, a periodic fluctuation is added to the adaptive factor to enhance the randomness of the algorithm, is the amplitude of the fluctuation, is the number of algorithm iterations, and MaxIter is the final number of algorithm iterations, is the difference between the fitness of the current individual position and the fitness of the best individual position, is the fitness of the best individual position, ensuring that the algorithm has a strong exploration ability in the initial stage, avoiding local optima, and ensuring the convergence of the algorithm by reducing randomness and increasing exploitation in the later stage.

[0031] Furthermore, the umbrella lizard optimization algorithm is improved. More specifically, a "simulated annealing" strategy is introduced to improve the mathematical model of position update in the exploitation stage of the umbrella lizard optimization algorithm, enhancing the local search ability of the algorithm. The improved mathematical model of position update in the exploitation stage is:

[0032] (7);

[0033] In formula (7), is the individual position updated by using the simulated annealing strategy in the exploitation stage, is the position of the individual in the current iteration, ub and lb are the upper and lower bounds of the algorithm search space, that is, row vectors with a dimension value of dim, is used to control the intensity of the simulated annealing part, is an individual randomly selected from the population, is the difference between the fitness of the current individual position and the fitness of the best individual position, is the annealing temperature that decreases with the number of iterations, is a random number with a value in the range (0, 1], and the meanings of other parameters are the same as above, The mathematical model of is:

[0034] (8);

[0035] In formula (8), is the initial temperature value of the simulated annealing, is the amplitude of temperature decrease, and the meanings of other parameters are the same as above.

[0036] Furthermore, when is less than 0, the algorithm unconditionally adopts the simulated annealing strategy to update the individual position in the exploitation stage. Otherwise, according to the magnitude of the acceptance probability P, it decides whether to adopt the simulated annealing strategy to update the individual position. As the number of iterations increases, the value of P gradually decreases. The mathematical model of P is:

[0037] (9);

[0038] In formula (9), the meanings of each parameter are the same as above. The mathematical model for choosing whether to adopt simulated annealing is:

[0039] (10);

[0040] In formula (10), is the individual position after update in the exploitation stage, has the same meaning as above, is the individual position using the mathematical model of updating with the original position in the exploitation stage, is a random number with a value in the range (0, 1], and the meanings of other parameters are the same as above. By introducing the acceptance mechanism of simulated annealing, the ability of the frill-necked lizard algorithm to jump out of the local optimum in the exploitation stage is enhanced, while ensuring the robustness of the overall optimization of the algorithm, effectively improving the convergence speed of the algorithm and the quality of the final solution.

[0041] Further, in step 4, the coefficients of the PID controller are the proportional coefficient Kp, the integral coefficient Ki, and the derivative coefficient Kd. Kp, Ki, and Kd are encoded as the search space dimension dim of the frill-necked lizard optimization algorithm. More specifically, the PID control parameters Kp, Ki, and Kd for optimizing the internal gas flow of the burner are encoded as the latitude values of nPop three-dimensional vectors. These three-dimensional vectors form a search space with upper and lower bounds of [ub, lb]. Each three-dimensional vector establishes a three-dimensional mapping with the individual position, and the values of the Kp, Ki, and Kd parameters are updated by updating the individual position.

[0042] Further, in step 3, through iterative optimization, the natural behavior of the frill-necked lizard is simulated to search for the optimal PID control parameters in the search space and obtain the optimal PID control method. The specific steps are as follows:

[0043] S31. Initialize the population size nPop of the improved frill-necked lizard algorithm (IFLO), the final number of iterations MaxIter, the upper and lower bounds ub and lb of the search space, and the dimension dim of the search space;

[0044] S32. Generate the initial positions of the population individuals by the pseudo-random number method, calculate and save the fitness of the initial positions of each individual. The mathematical model of the pseudo-random method is:

[0045] (11);

[0046] In formula (11), i = 1,..., nPop, is the initial position of the individual generated by the pseudo-random number, r is a random number with a value in the range (0, 1], and the meanings of other parameters are the same as above;

[0047] S33. The algorithm enters the exploration stage, simulating the hunting behavior of the frill-necked lizard to update the positions of the population individuals. First, determine a prey candidate set for each individual , and the fitness of the positions of all prey in the prey candidate set is less than the fitness of the individual's position. Secondly, the mathematical model for updating the position by simulating the hunting behavior of the frill-necked lizard in the algorithm is:

[0048] (12);

[0049] In formula (12), is the updated individual position by simulating the hunting behavior of the frill-necked lizard, is the individual position in the current iteration, is a prey position randomly selected from the candidate prey set, is the individual motion parameter improved by introducing the adaptive factor, has the same mathematical model as above, is a random number with a value in the range of (0, 1];

[0050] S34. Calculate the fitness of the updated individual position, and determine whether the fitness of the updated individual position is less than the fitness of the original individual position through greedy selection. The mathematical model of greedy selection is:

[0051] (13);

[0052] In formula (13), is the position after update of the individual in the exploration stage of the algorithm, is the fitness of the position, is the fitness of the position, and the meanings of other parameters are the same as above;

[0053] S35. The algorithm enters the exploitation stage, and simulates the retreat behavior of the frilled lizard after eating to update the positions of the population individuals. The original position update mathematical model of the algorithm when simulating the retreat behavior of the frilled lizard after eating is:

[0054] (14);

[0055] In formula (14), is the position of the individual updated by simulating the retreat behavior of the frilled lizard after eating. As described in formula (10), by determining whether to adopt the simulated annealing strategy to update the individual position according to the size of the acceptance probability P. When adopting the simulated annealing strategy to update the individual position, the position update formula of the algorithm in the exploitation stage is as described in formula (7). When not adopting the simulated annealing strategy to update the individual position, the position update formula of the algorithm in the exploitation stage is as described in formula (14);

[0056] S36. Calculate the fitness of the updated individual position, and determine whether the fitness of the updated individual position is less than the original position of the individual through greedy selection. The mathematical model of greedy selection is the same as above;

[0057] S37. Save the best individual position and the best fitness obtained in the current iteration, and compare them with the fitness of the best individual in the previous iteration. Select the individual with the smaller fitness as the final best individual, and the fitness of this individual is the final best fitness;

[0058] S38. Determine whether the current iteration number is greater than the final iteration number. If so, output the best individual position, and decode the latitude value of the individual position into the Kp, Ki, Kd parameters. If not, return to step S33 to continue the optimization.

[0059] The beneficial effects of the present invention are:

[0060] In the present invention, an optimization method for controlling the gas flow inside a burner is proposed. The parameters of a PID controller are optimized by improving the Iguana Flock Optimization (IFLO) algorithm. First, the individual motion parameters are improved by introducing an adaptive factor , enhancing the adaptability of the algorithm. Second, by introducing a "simulated annealing" strategy, the position update formula of the algorithm in the exploration stage is improved, enhancing the local search ability of the algorithm. Further, the accuracy of the PID controller is improved, as well as its stability in the face of complex and variable external environments, thereby achieving precise control of the gas flow in the burner, reducing the energy consumption of the burner, and more importantly, ensuring more precise and reliable temperature control during the baking process of anode carbon blocks. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 FIG. is a flow chart for optimizing the internal flow control of a burner.

[0062] Figure 2 FIG. is a flow chart for optimizing a PID controller using the improved Iguana Flock Optimization algorithm.

[0063] Figure 3 FIG. is a comparison chart of the fitness value curves of the improved Iguana Flock Optimization algorithm and the standard Iguana Flock Optimization algorithm for optimizing the internal flow PID controller of a burner.

[0064] Figure 4 FIG. is a curve chart for optimizing the Kp, Ki, and Kd parameters of the internal flow PID controller of a burner using the improved Iguana Flock Optimization algorithm.

[0065] Figure 5 FIG. is a comparison chart of the effects of the improved Iguana Flock Optimization algorithm and the standard Iguana Flock Optimization algorithm for optimizing the internal flow PID controller of a burner. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0067] An optimization method for controlling the gas flow inside a burner provided by the present invention includes: the improved Iguana Flock Optimization (IFLO) algorithm and the positional PID control algorithm, as Figure 1 shown. The specific steps are as follows:

[0068] Step 1: According to the baking temperature requirements of different types of anode carbon blocks, establish a gas flow control system model inside the burner, and the model includes a mathematical model and a simulation model.

[0069] Further, the mathematical model for controlling the gas flow inside the burner includes: First, a temperature demand model is established according to the roasting temperature requirements of different types of anode carbon blocks. Second, since the size of the gas flow inside the burner directly affects the flue temperature, a mathematical model between the gas flow and the heating temperature needs to be established.

[0070] Further, the temperature demand model describes that during the reaction process of the burner, the temperature demand curve includes three stages: the heating stage, the constant temperature stage, and the cooling stage. The temperature demand curve is:

[0071] (1);

[0072] In formula (1), is the temperature of the burner at time t, is the initial temperature, is the target temperature, and are the heating and cooling rates respectively, and t is the operating time of the burner.

[0073] Further, the gas flow will directly affect the temperature change in the combustion furnace. Therefore, the relationship model between the gas flow and the temperature change is established as:

[0074] (2);

[0075] In formula (2), is the temperature of the burner at time t, is the gas flow, is the ambient temperature, is the function of the gas flow and the temperature, is the influence of the ambient temperature and time on the temperature inside the furnace, and the meanings of other parameters are the same as above.

[0076] Further, the gas flow is adjusted by a PID controller to meet the temperature demand. The controller dynamically adjusts the gas flow according to the difference between the temperature demand and the actual temperature. The position-type PID controller model is:

[0077] (3);

[0078] In formula (3), is the gas flow at time t, e(t) is the error value between the actual temperature and the target temperature, , Kp, Ki, and Kd are the proportional, integral, and derivative coefficients of the PID controller, and the meanings of other parameters are the same as above.

[0079] Furthermore, the burner flow control simulation model includes: a PID controller module, an improved lizard optimization algorithm module, a burner dynamic characteristic module, a temperature signal acquisition module, and an error calculation module.

[0080] Step 2: Improve the optimization algorithm of the frilled lizard by improving the individual motion parameter I and introducing the "simulated annealing" strategy. The improvements are as follows:

[0081] S21. When the optimization algorithm for the lizard is in the exploratory stage, an adaptive factor is introduced to improve the individual motion parameters. , improve the adaptability of the algorithm;

[0082] S22. When the lizard optimization algorithm is in the development stage, the "simulated annealing" strategy is introduced to improve the position update formula in the algorithm development stage and enhance the local search capability of the algorithm.

[0083] Furthermore, an adaptive factor is introduced into the mathematical model of position update in the exploration phase of the lizard optimization algorithm to improve the individual motion parameters. , improve the adaptability of the algorithm, after improvement The mathematical model is:

[0084] (5);

[0085] In formula (5), i=1,…,nPop, nPop is the population size of the algorithm, is a function of the number of iterations The function of change, represents the attraction between the current individual position and the best individual position, set to 0.9, is the best individual position, is the individual position of the current iteration, represents the random intensity of individual position updates, set to 2, is a random number between (0,1], The mathematical model is:

[0086] (6);

[0087] In formula (6), is the initial motion parameter intensity value. By introducing the sine function into the mathematical model, periodic fluctuations are added to the adaptive factor to enhance the randomness of the algorithm. is the fluctuation amplitude, set to 0.5, is the number of algorithm iterations, MaxIter is the final number of algorithm iterations, is the difference between the current individual position fitness and the best individual position fitness, is the fitness of the best individual position.

[0088] Furthermore, the frill-necked lizard optimization algorithm is improved. More specifically, during the development stage of the frill-necked lizard optimization algorithm, a "simulated annealing" strategy is introduced into the position update mathematical model to improve the position update formula of the algorithm and enhance the local search ability of the algorithm. The improved position update mathematical model during the development stage is as follows:

[0089] (7);

[0090] In formula (7), is the individual position updated by the simulated annealing strategy during the development stage, is the current iteration individual position, ub and lb are the upper and lower bounds of the algorithm search space, that is, row vectors with a dimension value of dim, is used to control the intensity of the simulated annealing part, is an individual randomly selected from the population, is the difference between the fitness of the current individual position and the fitness of the best individual position, is the annealing temperature that decreases with the number of iterations, r4 is a random number with a value in the range (0, 1], and the meanings of other parameters are the same as above, The mathematical model of is as follows:

[0091] (8);

[0092] In formula (8), is the initial temperature value of the simulated annealing, is the amplitude of temperature reduction, set to 0.25, and the meanings of other parameters are the same as above.

[0093] More specifically, when is less than 0, the algorithm unconditionally adopts the simulated annealing strategy to update the individual position during the development stage. Otherwise, according to the size of the acceptance probability P, it decides whether to adopt the simulated annealing strategy to update the individual position. As the number of iterations increases, the value of P gradually decreases. The mathematical model of P is:

[0094] (9);

[0095] In formula (9), the meanings of each parameter are the same as above. The mathematical model for choosing whether to adopt simulated annealing is:

[0096] (10);

[0097] In formula (10), is the individual position updated during the development stage, has the same meaning as above, is the individual position using the original position update mathematical model during the development stage, is a random number with a value in the range (0, 1], and the meanings of other parameters are the same as above.

[0098] Step 3: Select an appropriate objective function and optimize the PID controller parameters through the IFLO algorithm. The algorithm iteratively searches for the optimal solution. By simulating the natural behavior of the frilled lizard, it searches for the best PID control parameters in the search space and obtains the optimal PID control method, namely the IFLO-PID control method. The specific steps are as follows:

[0099] S31: Initialize the population size nPop of the improved frilled lizard algorithm (IFLO) to 30, the final number of iterations MaxIter to 20, the upper and lower bounds ub and lb of the search space to [0 0 0] and [10 10 10], respectively, and the dimension dim of the search space to 3;

[0100] S32: Select the objective function as:

[0101] (4);

[0102] In Equation (4), T is the system running time, is the target temperature of the burner, is the actual temperature of the burner at time t, is the parameter for adjusting the sensitivity of the non-linear weight, and the meanings of other parameters are the same as above;

[0103] S33: Generate the initial positions of the population individuals through the pseudo-random number method, calculate and save the fitness of the initial positions of each individual. The mathematical model of the pseudo-random method is:

[0104] (11);

[0105] In Equation (11), i = 1,..., nPop, is the initial position of the individual generated through pseudo-random numbers, r is a random number with a value in the range (0, 1], and the meanings of other parameters are the same as above;

[0106] S34: The algorithm enters the exploration stage. Simulate the hunting behavior of the frilled lizard to update the positions of the population individuals. First, determine a prey candidate set for each individual. The fitness of the positions of all prey in the prey candidate set is less than the fitness of the position of this individual. Secondly, the mathematical model for updating the positions in simulating the hunting behavior of the frilled lizard is:

[0107] (12);

[0108] In Equation (12), is the position of the individual updated by simulating the hunting behavior of the frilled lizard, is the position of the individual in the current iteration, is from A prey position randomly selected from the candidate prey set is the individual movement parameter improved by introducing an adaptive factor The mathematical model is the same as above

[0109] S35. Calculate the fitness of the updated individual position, and judge whether the fitness of the updated individual position is less than the original position of the individual through greedy selection. The mathematical model of greedy selection is as follows:

[0110] (13);

[0111] In formula (13), is the updated position in the exploration stage of the algorithm is the fitness of the position is the fitness of the position, and the meanings of other parameters are the same as above

[0112] S36. The algorithm enters the exploitation stage, and simulates the retreat behavior of the frilled lizard after feeding to update the positions of the population individuals. The mathematical model of the position update of the algorithm in simulating the retreat behavior of the frilled lizard after feeding is as follows:

[0113] (14);

[0114] In formula (14), is the individual position updated by simulating the retreat behavior of the frilled lizard after feeding. As described in formula (10), by judging the magnitude of the acceptance probability P, it is decided whether to adopt the simulated annealing strategy to update the individual position. When adopting the simulated annealing strategy to update the individual position, the position update formula of the algorithm in the exploitation stage is as described in formula (7). When not adopting the simulated annealing strategy to update the individual position, the position update formula of the algorithm in the exploitation stage is as described in formula (14);

[0115] S37. Calculate the fitness of the updated individual position, and judge whether the fitness of the updated individual position is less than the original position of the individual through greedy selection. The mathematical model of greedy selection is the same as above;

[0116] S38. Save the best individual position and the best fitness obtained in the current iteration, and compare them with the fitness of the best individual in the previous iteration. Select the individual with the smaller fitness as the final best individual, and the fitness of this individual is the final best fitness;

[0117] S39. Judge whether the current iteration number is greater than the final iteration number. If so, output the best individual position, and decode the latitude value of the individual position into the Kp, Ki, Kd parameters. If not, return to step S33 to continue the optimization.

[0118] Step 4: Precisely control the gas flow inside the burner through the IFLO-PID control method. Calculate the deviation value e(t) between the target temperature and the actual temperature based on the temperature value of the flue gas zone feedback in real time, and input it into the IFLO-PID controller to obtain the control quantity , that is, the gas flow inside the burner, to achieve closed-loop IFLO-PID control of the gas flow. The specific operation steps of the system are as follows:

[0119] S41: Use Matlab and Simulink to establish a simulation model for the gas flow control system inside the burner. Build a system simulation model through Simulink, including a positional PID controller model, a temperature demand model, and an objective function model;

[0120] S41: Convert the problem of controlling the gas flow inside the burner into a mathematical model optimization problem. Design a third-order transfer function to simulate the influence of the gas flow on the temperature during the operation of the burner. The transfer function is:

[0121] ;

[0122] In the formula, G is the transfer function value, and s is the complex frequency domain variable after Laplace transform;

[0123] S42: Set the target temperature of the burner to 1200 °C, set the running time of the simulation model to 50 s, and calculate the error value e(t) between the actual temperature and the target temperature through the error calculation module, ;

[0124] S43: Input the error value e(t) into the PID controller module to dynamically adjust the gas flow inside the burner , as Figure 3 shown. Through the iterative optimization of the improved frill-necked lizard optimization algorithm module, the optimal Kp, Ki, and Kd parameters of the PID controller are obtained as 2.076, 1.002, and 0.243;

[0125] S44: Input the gas flow into the burner dynamic characteristic module, that is, the third-order transfer function, and output the actual temperature ;

[0126] S45: The temperature signal acquisition module acquires the actual temperature and outputs it to the error calculation module to achieve closed-loop IFLO-PID control of the combustion flow inside the burner.

[0127] As Figure 4As shown, by comparing and analyzing the fitness curves of the improved umbrella lizard optimization algorithm and the standard umbrella lizard optimization algorithm for optimizing the internal flow PID controller of the burner, it can be seen that in the first 10 iterations, the fitness value of the improved umbrella lizard optimization algorithm (IFLO) drops rapidly, and its convergence speed is significantly faster than that of the standard umbrella lizard optimization algorithm (FLO). When the iteration reaches the 12th time, the convergence speed of the improved umbrella lizard optimization algorithm begins to level off, indicating that the algorithm has fallen into a local optimum, but it is still converging slowly. The final fitness value approaches 244.447. While the standard umbrella lizard optimization algorithm falls into a local optimum after the 9th iteration and cannot jump out, and the final fitness value approaches 298.379. According to the principle that the smaller the fitness value, the better the solution found by the algorithm, it can be concluded that the improved umbrella lizard optimization algorithm (IFLO) is significantly stronger than the standard umbrella lizard optimization algorithm (FLO) in terms of optimization performance. This shows that when the improved umbrella lizard optimization algorithm is applied to optimize the internal flow PID controller of the burner, it has obvious performance advantages and can significantly improve the performance and stability of the control system.

[0128] As Figure 5 shown, by comparing and analyzing the response curves of the improved umbrella lizard optimization algorithm and the standard umbrella lizard optimization algorithm for optimizing the internal flow PID controller of the burner, and comparing them in terms of indicators such as response speed, overshoot, and dynamic response time, it can be seen that the response speed of the improved algorithm is basically the same as that of the standard algorithm, and both reach the target temperature of 1200°C at 2s. However, the overshoot of the improved algorithm is significantly smaller than that of the standard algorithm, and the stabilization time is much faster than that of the ordinary algorithm. Analyzing the dynamic response time of the PID controller, the improved umbrella lizard optimization algorithm for optimizing the internal flow PID controller of the burner reaches a stable state at 15s, while the ordinary algorithm reaches a stable state at 24s. This shows that it is effective to optimize the parameters of the internal flow PID controller of the burner through the improved umbrella lizard optimization algorithm.

[0129] In summary, the present invention provides an adaptive airway pressure control method for a ventilator. This method optimizes the parameters of the PID controller of the ventilator airway pressure control system through the improved secretary bird optimization algorithm, improves the dynamic response performance and stability of the PID controller, enables the controller to quickly respond to airway pressure changes, reduces the overshoot and stabilization time of the controller, thereby ensuring the performance and robustness of the ventilator airway pressure control system, and ensuring that the patient receives a stable and appropriate airway pressure. Through this optimization method, the overall performance of the ventilator airway pressure control system has been significantly improved.

Claims

1. A method for optimizing the internal gas flow control of a burner, comprising: Improved Fringed Lizard Optimization Algorithm (IFLO) and Position PID Control Algorithm, the specific steps are: Step 1: Establish a burner flow control system model according to the roasting temperature requirements of different types of anode carbon blocks, wherein the model includes a mathematical model and a simulation model; Step 2: Improve the optimization algorithm of the frilled lizard by improving the individual motion parameter I and introducing the "simulated annealing" strategy. The improvements are as follows: S21, when the lizard optimization algorithm is in the exploration stage, an adaptive factor is introduced to improve the individual motion parameter I to improve the adaptability of the algorithm; S22. When the lizard optimization algorithm is in the development stage, the "simulated annealing" strategy is introduced to improve the position update formula in the algorithm development stage and enhance the local search capability of the algorithm. Step 3, select a suitable objective function, optimize the PID controller parameters by using the IFLO algorithm, iterate the algorithm to find the best, simulate the natural behavior of the parasol lizard, search for the best PID control parameters in the search space, and obtain the optimal PID control method, namely the IFLO-PID control method; Step 4: The internal gas flow of the burner is precisely controlled by the IFLO-PID control method. According to the temperature value of the fire channel area fed back in real time, the deviation value e(t) between the target temperature and the actual temperature is calculated and input into the IFLO-PID controller to obtain the control quantity Q(t), i.e., the internal gas flow of the burner, thereby realizing closed-loop IFLO-PID control of the gas flow.

2. A burner internal gas flow control optimization method according to claim 1, characterized in that: Taking into account the burner temperature requirements and the gas flow control reaction speed, overshoot, and dynamic response time, the selected objective function is: In formula (4), T is the system running time, T ref is the target temperature of the burner, T(t) is the actual temperature of the burner at time t, γ is the parameter for adjusting the sensitivity of the nonlinear weight, and the other parameters have the same meanings as above.

3. A burner internal gas flow control optimization method according to claim 2, characterized in that: The improved frill lizard optimization algorithm includes introducing an adaptive factor into the mathematical model of position update in the exploration phase of the frill lizard optimization algorithm to improve the individual motion parameter I, thereby improving the adaptability of the algorithm. The improved mathematical model of I is: I=A(κ)×[α×(X best -X i (k))+β×r3] (5); In formula (5), i = 1, ..., nPop, nPop is the population size, A(κ) is a function that changes with the number of iterations k, α represents the attraction between the current individual position and the best individual position, X best is the best individual position, X i (κ) is the individual position of the current iteration, β represents the random intensity of the individual position update, r3 is a random number between (0, 1], and the mathematical model of A(κ) is: In formula (6), A0 is the initial motion parameter strength value. By introducing the sine function in the mathematical model, periodic fluctuations are added to the adaptive factor to enhance the randomness of the algorithm. γ is the fluctuation amplitude, κ is the number of algorithm iterations, MaxIter is the final number of algorithm iterations, Δf is the difference between the current individual position fitness and the best individual position fitness, and f best The fitness of the best individual position ensures that the algorithm has a strong exploration ability in the early stage, while avoiding local optimality, and ensures the convergence of the algorithm in the later stage by reducing randomness and increasing development.

4. A burner internal gas flow control optimization method according to claim 3, characterized in that: The "simulated annealing" strategy is introduced to improve the mathematical model of position update in the development phase of the lizard optimization algorithm. The improved mathematical model of position update in the development phase is: In formula (7), is the individual position updated by the simulated annealing strategy in the development phase, X i (κ) is the individual position of the current iteration, ub and lb are the upper and lower bounds of the algorithm search space, that is, the row vector with the latitude value dim, η is used to control the intensity of the simulated annealing part, and X′ i (κ) is a randomly selected individual in the population, Δf is the difference between the current individual position fitness and the best individual position fitness, T(κ) is the annealing temperature that decreases with the number of iterations, r4 is a random number between (0, 1], and the other parameters have the same meanings as above. The mathematical model of T(κ) is: T(κ)=T0×exp(-ρ×κ) (8); In formula (8), T0 is the initial temperature value of simulated annealing, ρ is the amplitude of temperature reduction, and the other parameters have the same meaning as above; Furthermore, when Δf is less than 0, the algorithm unconditionally adopts the simulated annealing strategy to update the individual position in the development stage. Otherwise, it decides whether to adopt the simulated annealing strategy to update the individual position according to the size of the acceptance probability P. As the number of iterations increases, the value of P gradually decreases. The mathematical model of P is: In formula (9), the meanings of the parameters are the same as above. The mathematical model for selecting whether to use simulated annealing is: In formula (10), X i (κ+1) is the individual position after update in the development phase, The meaning is the same as above. is the individual position of the mathematical model updated using the original position in the development phase, r5 is a random number between (0, 1], and the other parameters have the same meanings as above.

5. A burner internal gas flow control optimization method according to claim 4, characterized in that: In the step 4, the coefficients of the PID controller are respectively the proportional coefficient Kp, the integral coefficient Ki and the differential coefficient Kd. Kp, Ki and Kd are encoded as the search space dimension dim of the lizard optimization algorithm. More deeply, the PID control parameters Kp, Ki and Kd used to optimize the internal gas flow of the burner are encoded as the latitude values ​​of nPop three-dimensional vectors, where nPop is the population size. These three-dimensional vectors constitute a search space with upper and lower bounds [ub, lb]. Ub and lb are the upper and lower bounds of the algorithm search space. A three-dimensional mapping is established between each three-dimensional vector and the individual position, and the parameter values ​​of Kp, Ki and Kd are updated by updating the individual position.

6. A burner internal gas flow control optimization method according to claim 5, characterized in that: In step 3, the optimal PID control parameters are searched in the search space by iterative optimization to simulate the natural behavior of the frilled lizard, and the optimal PID control method is obtained. The specific steps are as follows: S31, initialize the population size nPop of the improved fringed lizard algorithm (IFLO), the final number of iterations MaxIter, the upper and lower bounds ub and lb of the search space, and the dimension dim of the search space; S32, generating the initial positions of the individuals in the population by a pseudo-random number method, calculating and saving the fitness of the initial positions of each individual, the mathematical model of the pseudo-random method is: X i =lb+r×(ub-lb) (11); In formula (11), i=1,…,nPop,X i is the individual initial position generated by pseudo-random numbers, r is a random number between (0, 1], and the other parameters have the same meanings as above; S33, the algorithm enters the exploration phase, simulating the hunting behavior of the frilled lizard to update the position of the population individuals. First, a prey candidate set CP is determined for each individual. i , the fitness of the positions of all prey in the prey candidate set is less than the fitness of the individual position. Secondly, the mathematical model of the algorithm in simulating the hunting behavior of frilled lizards is updated as follows: In formula (12), To simulate the hunting behavior of frilled lizards, the updated individual positions, X i (κ) is the individual position in the current iteration, SP i From X i (κ) is a prey position randomly selected from the candidate prey set, I is the individual motion parameter after the introduction of the adaptive factor, the mathematical model of I is the same as above, r6 is a random number between (0, 1]; S34, calculate the fitness of the updated individual position, and judge whether the fitness of the updated individual position is less than the fitness of the original individual position through greedy selection. The mathematical model of greedy selection is: In formula (13), X i (κ+1) is the updated position of the individual in the exploration phase of the algorithm, ƒ i P1 for The fitness of the position, ƒ i For X i (κ) fitness of the position, other parameters have the same meaning as above; S35, the algorithm enters the development stage, simulating the retreat behavior of the frilled lizard after eating to update the position of the population individuals. The algorithm simulates the original position update mathematical model of the frilled lizard after eating: In formula (14), To simulate the retreat behavior of the frilled lizard after eating, the updated individual position is as described in formula (10). According to the size of the acceptance probability P, it is decided whether to use the simulated annealing strategy to update the individual position. When the simulated annealing strategy is used to update the individual position, the position update formula of the algorithm in the development stage is as described in formula (7). When the simulated annealing strategy is not used to update the individual position, the position update formula of the algorithm in the development stage is as described in formula (14); S36, calculating the fitness of the updated individual position, and judging whether the fitness of the updated individual position is less than the original individual position through greedy selection, the mathematical model of greedy selection is the same as above; S37, saving the best individual position and best fitness obtained in the current iteration, and comparing them with the fitness of the best individual in the previous iteration, selecting the individual with smaller fitness as the final best individual, and the fitness of the individual is the final best fitness; S38. Determine whether the current number of iterations is greater than the final number of iterations. If so, output the best individual position and decode the latitude value of the individual position into Kp, Ki, and Kd parameters. If not, return to step S33 to continue optimizing.

Images