Method for setting GPC parameters of wind tunnel directly-heated heater based on improved sparrow algorithm

Abstract

The invention belongs to the technical field of generalized predictive control, and discloses a wind tunnel directly-heated heater GPC parameter setting method based on an improved sparrow algorithm. The setting method comprises the steps that a sparrow population is initialized, and a to-be-set parameter search space is defined; defining the maximum number of iterations, and determining the proportion of a discoverer, a follower and a warning person in the sparrow population; defining a system expected output temperature; initial control parameters are input into the GPC controller, a mathematical model of the directly-heated heater of the wind tunnel is established, and the controller is used for simulating an operation control system; the real-time acquisition system outputs and calculates performance indexes, and individual fitness values of sparrows are calculated according to the performance indexes and sorted; fusing the improved sparrow algorithm with a sparrow position updating formula, performing iterative updating according to the improved position updating formula, updating an individual fitness value, and recording a current iterative optimal individual and an optimal fitness value; and outputting the GPC optimal parameter group. The setting method reduces the blowing test cost and has engineering practical value.

Description

Technical Field

[0001] This invention belongs to the field of generalized predictive control technology, specifically relating to an improved sparrow algorithm for tuning GPC parameters of a wind tunnel direct-heating heater. Background Technology

[0002] Direct-heating heaters are a common method for heating gas in wind tunnels. This method produces less pollution to the test airflow, but power consumption increases with heating time, raising the cost of wind tunnel tests. Furthermore, the heater's internal structure is complex, exhibiting dynamic characteristics such as time lag and nonlinearity, making it difficult to control the output airflow temperature during wind tunnel tests.

[0003] Generalized predictive control (GPC) exhibits strong robustness to time-delay systems and has proven to be an effective control scheme for temperature heating systems. However, its performance is highly dependent on the proper tuning of parameters such as the prediction time domain, control time domain, and weighting coefficients. Traditional parameter tuning relies on manual experience or Ziegler-Nichols empirical formulas, which suffers from drawbacks such as long tuning cycles and poor adaptability. Using intelligent optimization algorithms for automatic parameter tuning can improve tuning efficiency to some extent. Among these, the sparrow search algorithm, as a novel swarm intelligence algorithm, shows application potential due to its simple structure and efficient search. However, the standard sparrow algorithm is prone to getting trapped in local optima in multidimensional parameter optimization.

[0004] Currently, there is an urgent need to develop an improved method for tuning the GPC parameters of a wind tunnel direct-heating heater using the sparrow algorithm. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an improved sparrow algorithm for tuning the GPC parameters of a wind tunnel direct-heating heater, so as to overcome the defects of the prior art.

[0006] The improved sparrow algorithm-based method for tuning GPC parameters of a wind tunnel direct-heating heater according to the present invention includes the following steps: S1. Initialize the sparrow population and define the search space for the parameters to be tuned; S2. Define the maximum number of iterations. To determine the proportions of discoverers, followers, and vigilants in a sparrow population; S3. The overshoot of the system output. and rise time Defined as the output performance index of the heater temperature control system, and defined as the system's desired output temperature. ; S4. Input the initial control parameters into the GPC controller, establish a mathematical model of the wind tunnel direct-heating heater, and use the controller to simulate the operation of the control system; S5. Real-time acquisition of system output and calculation of performance indicators, calculation of individual sparrow fitness values ​​based on performance indicators and sorting of them; S6. Integrate the improved sparrow algorithm with the sparrow position update formula, perform iterative updates according to the improved position update formula, update the individual fitness value at the same time, and record the best individual and the best fitness value in the current iteration; The improved location update formulas include the improved discoverer location update formula and the improved vigilant location update formula; The improved discoverer location update formula is as follows: ; In the formula, All are uniformly distributed random numbers within the range [0,1]. and Each is an independent dynamic learning factor; This represents the number of iterations. For the first In the next iteration, the first of the discoverers Victor The historical best position of an individual sparrow; For the first In the next iteration, the first of the discoverers Victor The optimal position of a sparrow in its dynamic neighborhood; This is a nonlinear decaying inertia weighting factor; Indicates the first During the nth iteration, the 1st The sparrow was in the first Dimensional location information; This indicates the improved location of the discoverer; A random warning value within the range [0,1]. A random safety value within the range of [0.5, 1]. It is a random number that follows a standard normal distribution. It is a unit column vector; The improved alerter position update formula is as follows: ; In the formula, The optimal fitness value in the neighborhood; To obtain After 6 iterations The difference; For dynamic disturbance coefficients; The threshold for determining where the neighborhood stops; These are all step size adjustment parameters. , ; For the first The globally optimal position of the individual sparrow at the next iteration; and These are the current, globally best, and globally worst individual fitness values, respectively; to prevent the denominator from being zero, we take... It is the minimum constant; For the first The position of the worst-performing sparrow in the next iteration; S7. Determine if the current iteration count has reached the maximum iteration count. If not, continue executing S4~S6. If yes, output the optimal GPC parameter set.

[0007] Furthermore, the mathematical model of the wind tunnel direct-heating heater... for: ; In the formula, Indicates the system output quantity. This is the transfer function structure for the power part of the model. The discrete structure form is as follows: ; This represents the nonlinear structure of the power component of the model. The structural form is: ; In the formula, This is the output power signal; Input the transfer function structure for the temperature perturbation part into the model. The discrete structure form is as follows: ; Input the nonlinear structure of the temperature perturbation component into the model. The structural form is: ; In the formula, To output a disturbance signal.

[0008] Furthermore, the GPC controller selects the controlled autoregressive integral moving average (CARIMA) model as the prediction model to describe the dynamic characteristics of the controlled object and reflect the dynamic causal relationship between the input and output of the controlled object; and constructs the optimization objective function of the GPC controller. The specific form is as follows: ; In the formula, For the maximum prediction time domain; This represents the current state of the controller. The number of steps the controller runs; In order to be in The Future of Time Step output value; In order to be in The Future of Time The expected output sequence value of the step, Set the temperature for THTCS; To control the time domain, In order to be in The Future of Time The expected control sequence value of the step; The control weighting coefficient is set manually and ranges from 0 to 1. A feedback correction mechanism is constructed. After each control cycle, the actual measured value of the controlled variable is collected and compared with the model prediction value at the same time. The difference between the actual value and the prediction value is calculated, the magnitude of the deviation is analyzed, and deviation correction is performed.

[0009] Furthermore, the fitness value is: ; In the formula, For the controller coefficient group, the upper right corner of the vector group is... It is the transpose symbol. The overshoot of the system output. This refers to the temperature rise time output by the system. The maximum allowable temperature rise time is [value missing]. The maximum allowable overshoot is , To adjust the weighting coefficient of time, This is the weighting coefficient for overshoot.

[0010] The improved sparrow algorithm-based GPC parameter tuning method for wind tunnel direct-heated heaters of this invention can find the optimal generalized predictive controller parameters when the controlled object is under different operating conditions. This effectively improves the efficiency of controller parameter tuning, enhances the control accuracy of the outlet temperature of the wind tunnel direct-heated heater, shortens the rise time of the output temperature, reduces the power loss caused by wind tunnel blowing tests, and lowers the cost of blowing tests. It has practical engineering value. Attached Figure Description

[0011] Figure 1 The flowchart shows the GPC parameter tuning method for a wind tunnel direct-heating heater based on the improved sparrow algorithm of the present invention. Figure 2 The output comparison curves obtained by calculating and tuning the controller parameters using different algorithms are shown in the example. Detailed Implementation

[0012] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0013] Example: Figure 1 As shown, the improved sparrow algorithm-based method for tuning GPC parameters of a wind tunnel direct-heating heater in this embodiment includes: S1. Initialize the sparrow population and define the search space for the parameters to be tuned; Set the initial population size Set the vector group of controller coefficients to be tuned in GPC The search scope is specifically defined as: setting the prediction time domain. The search range is [1, 100]; control time domain The search range is [1, 50]; the weighting coefficient is controlled. With the softening coefficient The search range is [0,1); step coefficient The search range is [-1.2, 1.2]; S2. Define the maximum number of iterations. To determine the proportions of discoverers, followers, and vigilants in a sparrow population; Define the maximum number of iterations Number of discoverers Number of followers Number of vigilant individuals Among them, the sparrow population ; S3. The overshoot of the system output. and rise time Defined as the output performance index of the heater temperature control system, and defined as the system's desired output temperature. ; The overshoot of the system output and rise time As the fitness of the algorithm, the relationship between fitness and performance metrics is as follows: ; In the formula, The fitness function; This is a set of controller coefficient vectors, where the upper right corner of the vector set is... It is the transpose symbol; The overshoot of the system output; This refers to the temperature rise time output by the system; the maximum allowable temperature rise time is... The maximum allowable overshoot is: ; The weighting coefficient for adjusting time; This is the weighting coefficient for overshoot; S4. Input the initial control parameters into the GPC controller, establish a mathematical model of the wind tunnel direct-heating heater, and use the controller to simulate the operation of the control system; Specifically, inputting the initial controller coefficient vector set in the GPC controller involves setting: , , , , A mathematical model of the wind tunnel direct-heating heater is established and a controller is used, specifically: the system input power is collected. Input disturbance temperature A Hammerstein model framework is established to describe the nonlinear characteristics of the controlled object. The dynamic linear part of the model is described using a transfer function structure, as follows: ; In the formula, This is the transfer function structure for the power part of the model; Input the nonlinear structure of the temperature perturbation component into the model; Output polynomial coefficients for the transfer function; The transfer function controls the input polynomial coefficients; The perturbation input polynomial coefficients are used to the transfer function; and These are the coefficients to be identified, where From 1 to From 1 to From 1 to 0 natural numbers; , and The order corresponding to the coefficients to be identified. is the discrete factor.

[0014] use The nonlinear part of the model is described by a polynomial of order 1, with the following structure: ; In the formula, This represents the nonlinear structure of the power component of the model; This represents the nonlinear structure of the perturbation part of the model; This is an intermediate function for the nonlinear structure of the power part of the model; This is an intermediate function for the nonlinear structure of the perturbation part of the model; These are the coefficients to be identified for the nonlinear structure of the power component of the model. These are the coefficients to be identified for the nonlinear structure of the perturbation part of the model. The order of the coefficients to be identified corresponding to the power control quantity. The order of the coefficients to be identified corresponding to the disturbance control quantity; Construct a linear regression form for the input and output: ; In the formula, It is a parameter vector containing the coefficients to be identified. ; This is an information vector containing system input and output data; This is the disturbance deviation; The recursive least squares parameter estimation algorithm is used to identify the model parameters, and the final mathematical model of the controlled object under the condition of disturbance is as follows: ; In the formula, Output temperature for the model; This is the transfer function structure for the power part of the model; Input the nonlinear structure of the temperature perturbation component into the model; For discrete factors, It is a continuous factor; Input the nonlinear structure of the temperature perturbation component into the model; This represents the nonlinear structure of the power component of the model; For input power signal; For input disturbance signals; Design a generalized predictive controller and run the control system on the Simulink simulation platform of MATLAB. Simulate the temperature output effect of a wind tunnel direct-heating heater after different optimization algorithms are applied to the controller parameters. Set the simulation heating time to 140 seconds and the sampling rate to 3Hz. S5. Real-time acquisition of system output and calculation of performance indicators, calculation of individual sparrow fitness values ​​based on performance indicators and sorting of them; S6. Integrate the improved sparrow algorithm with the sparrow position update formula, perform iterative updates according to the improved position update formula, update the individual fitness value at the same time, and record the best individual and the best fitness value in the current iteration; The improved location update formulas include the improved discoverer location update formula and the improved vigilant location update formula; The improved discoverer location update formula is as follows: ; In the formula, All are uniformly distributed random numbers within the range [0,1]. and Each is an independent dynamic learning factor; This represents the number of iterations. For the first In the next iteration, the first of the discoverers Victor The historical best position of an individual sparrow; For the first In the next iteration, the first of the discoverers Victor The optimal position of a sparrow in its dynamic neighborhood; This is a nonlinear decaying inertia weighting factor; Indicates the first During the nth iteration, the 1st The sparrow was in the first Dimensional location information; Indicates the first The discoverer's location after the next iteration; A random warning value within the range [0,1]. A random safety value within the range of [0.5, 1]. It is a random number that follows a standard normal distribution. for The unit column vector of the column; Specifically, the improved discoverer location update formula will be applied in... Under these conditions, an individual cognitive term is introduced into the improved discoverer location update formula. Collaboration with neighboring areas ,in This indicates the discoverer's historical best position; This represents the optimal location in the dynamic neighborhood of the discoverer. Indicates the current location of the discoverer; for The solution is obtained by taking the current sparrow individual as the center and calculating the solution within the radius of the domain. The fitness minimum value is determined within the defined range, expressed as follows: ; In the formula, Let the initial radius be set as follows: This is equal to 20% of the number of sparrows in the colony; Maximum number of iterations; Set the decay index. The goal is to focus on local development by continuously increasing the neighborhood radius with the number of iterations; The following expression is used for selecting the dynamic learning factor: ; In the formula, Set the initial learning factor to the initial value of the learning factor. ; and The final value of the learning factor satisfies... as well as By changing the dynamic learning factor, it is ensured that the discoverer can develop from the unknown areas of individual cognitive items in the group; The design of the nonlinear decaying inertia weighting factor is as follows: ; Compared with the weight factors in the standard SSA algorithm, the weight factors designed in this implementation show a trend of decay rate from slow to fast, thereby ensuring that the discoverer's exploration ability in the early stage of iteration is no less than that of the standard SSA algorithm. As the number of iterations increases, the discoverer's exploration mechanism smoothly transitions to individual cognition terms and neighborhood cooperation terms.

[0015] The improved alerter position update formula is as follows: ; In the formula, The optimal fitness value in the neighborhood; To obtain After 6 iterations The difference; For dynamic disturbance coefficients; The threshold for determining where the neighborhood stops; These are all step size adjustment parameters. , ; For the first The globally optimal position of the individual sparrow at the next iteration; and These are the current, globally best, and globally worst individual fitness values, respectively; to prevent the denominator from being zero, Pick ; For the first The position of the worst-performing sparrow in the next iteration; S7. Determine if the current iteration count has reached the maximum iteration count. If not, continue executing S4~S6. If yes, output the optimal GPC parameter set.

[0016] according to The heater temperature control system model under the given conditions was compared, and the impact of parameter tuning methods based on four algorithms (PSO, BA, SSA, and ISSA) on the controller performance was analyzed. The final control parameter tuning results of the different algorithms are shown in Table 1. The system output responses were compared as follows: Figure 2 As shown in Table 2, the performance indicators of the test results are as follows.

[0017] Table 1 Control parameter tuning results under different algorithms

[0018] Table 2 Controller performance metrics under different algorithms

[0019] Table 1 shows the performance of the four algorithms in the prediction time domain. and control time domain The tuning results were almost identical, with the main differences lying in the remaining control parameters. In Table 2, the fitness value of the ISSA algorithm was only 0.44, lower than the fitness values ​​obtained from the other three algorithms. Therefore, the ISSA algorithm has a better tuning effect and is more likely to find the global optimum of the controller parameters. Figure 2 It can be seen that the controller parameters tuned by the improved sparrow algorithm have the ability to suppress system output overshoot and coordinate the optimization of temperature rise time, and have excellent anti-disturbance performance.

[0020] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. For those skilled in the art, all features disclosed in the present invention, or all steps in all methods or processes disclosed, except for mutually exclusive features and / or steps, can be combined in any way without departing from the principles of the present invention. The present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A wind tunnel direct-fired heater GPC parameter tuning method that improves the sparrow search algorithm, characterized in that, The method comprises the following steps: S1. initializing a sparrow population and defining a search space of parameters to be set; S2. Define the maximum number of iterations determining the ratio of discoverers, followers and alarmers in the sparrow population; S3. Overshoot of the system output and rise time defined as the heater temperature control system output performance indicator, defining the system desired output temperature ; S4. inputting initial control parameters in a GPC controller, establishing a mathematical model of a direct heating heater of a wind tunnel and simulating and running the control system by using the controller; S5. collecting system outputs in real time and calculating performance indexes, calculating sparrow individual fitness values according to the performance indexes and sorting the individual fitness values; S6. fusing the improved sparrow algorithm and a sparrow position updating formula, iteratively updating according to the improved position updating formula, updating individual fitness values simultaneously, recording a current iteration optimal individual and an optimal fitness value; The improved position updating formula comprises an improved scout position updating formula and an improved sentinel position updating formula; The improved scout position updating formula is as follows: ； wherein, are uniformly distributed random numbers in the range of [0, 1]; are independent dynamic learning factors; are independent dynamic learning factors; is the number of iterations; is the historical optimal position of the jth sparrow individual in the dth dimension of the discoverer at the ith iteration; is the historical optimal position of the jth sparrow individual in the dth dimension of the discoverer at the ith iteration; is the historical optimal position of the jth sparrow individual in the dth dimension of the discoverer at the ith iteration; is the optimal position of the dynamic neighborhood of the jth sparrow individual in the dth dimension of the discoverer at the ith iteration; is the optimal position of the dynamic neighborhood of the jth sparrow individual in the dth dimension of the discoverer at the ith iteration; is the optimal position of the dynamic neighborhood of the jth sparrow individual in the dth dimension of the discoverer at the ith iteration; is the optimal position of the dynamic neighborhood of the jth sparrow individual in the dth dimension of the discoverer at the ith iteration; is a nonlinear decay inertia weight factor; represents the position information of the jth sparrow in the dth dimension at the ith iteration; represents the position information of the jth sparrow in the dth dimension at the ith iteration; represents the position information of the jth sparrow in the dth dimension at the ith iteration; represents the position information of the jth sparrow in the dth dimension at the ith iteration; represents the position information of the jth sparrow in the dth dimension at the ith iteration; represents the position information of the jth sparrow in the dth dimension at the ith iteration; represents the position information of the jth sparrow in the dth dimension at the ith iteration; is a random alert value in the range of [0, 1], is a random safety value in the range of [0.5, 1]; is a random number subject to a standard normal distribution, is a unit column vector of the jth column; is a unit column vector of the jth column; The improved sentinel position updating formula is as follows: ； In the formula, The optimal fitness value in the neighborhood; To obtain After 6 iterations The difference; For dynamic disturbance coefficients; The threshold for determining where the neighborhood stops; These are all step size adjustment parameters. , ; For the first The globally optimal position of the individual sparrow in the next iteration; and These are the current, globally best, and globally worst individual fitness values, respectively; to prevent the denominator from being zero, Pick ; For the first The position of the worst-performing sparrow in the next iteration; S7. judging whether a current iteration number reaches a maximum iteration number, if not, continuing to execute S4-S6, and if yes, outputting a GPC optimal parameter group.

2. The improved sparrow search algorithm based GPC parameter tuning method for wind tunnel direct-heating heater according to claim 1, wherein, The wind tunnel direct heating heater mathematical model Is: ； wherein represents the system output quantity, is a transfer function structure of the model power portion, The discrete structure form of is ； for the non-linear structure of the model power part, The structural form is: ； inputting a temperature disturbance part of a nonlinear structure to the model, The structural form is: ； In the formula, is an input disturbance signal.

3. The improved sparrow search algorithm-based GPC parameter tuning method for a direct-fired heater in a wind tunnel, according to claim 2, wherein the GPC controller selects a controlled autoregressive integrated moving average model (CARIMA) as a controller prediction model to describe the dynamic characteristics of the controlled object and reflect the dynamic causal relationship between the input and output of the controlled object. And build the GPC controller optimization objective function The specific form is: ； wherein, is the maximum prediction horizon; is the current time instant of the controller; is the number of steps of the controller run; is the output value of the future step at the time instant ; is the expected output sequence value of the future step at the time instant ; is the set temperature of the THTCS; is the control horizon, is the expected control sequence difference value of the future step at the time instant ; is the control weighting coefficient in the range [0, 1]; A feedback correction mechanism is constructed, actual measured values of controlled variables are collected after each control cycle ends, the actual measured values are compared with model predicted values at the same time, a difference between the actual measured values and the predicted values is calculated, a deviation is analyzed and a deviation correction treatment is performed.

4. The improved sparrow search algorithm based GPC parameter tuning method for wind tunnel direct-heating heater according to claim 3, wherein, The fitness value is as follows: ； wherein is the fitness function; is the controller coefficient vector set, where the upper right corner of the vector set is the transpose symbol; is the overshoot of the system output; is the warm-up time of the system output; the maximum value of the warm-up time is ; the maximum value of the overshoot is ; is the weight coefficient of the adjustment time; is the weight coefficient of the overshoot.

Images