A water flow control optimization method based on fishery environment dynamic monitoring

Abstract

The application discloses a water flow control optimization method based on fishery environment dynamic monitoring and belongs to the technical field of control optimization, mainly including the following steps: constructing a fishery water flow PID control system, adopting an improved gray butterfly brood parasitism optimization algorithm to set parameters, and finally applying the best parameters obtained by optimization to a flow controller. The application utilizes reverse learning initialization to improve initial solution set coverage, gives individuals the ability to jump out of local extreme values through a Levy flight strategy, adopts a vector difference strategy to correct physical dimension and accelerate convergence, and applies a reverse defense restart mechanism to crack iteration stagnation by using spatial symmetry. The algorithm is applied to water flow PID parameter setting, reduces the overshoot and steady-state error of the system step response, shortens the adjustment time, and realizes precise dynamic control of the fishery environment water flow.

Description

Technical Field

[0001] This invention belongs to the technical field of control optimization, and in particular relates to a method for optimizing water flow control based on dynamic monitoring of the fishery environment. Background Technology

[0002] Digital fisheries has become an inevitable path to achieving intensive, precise, and ecological aquaculture. Among its components, the dynamic monitoring system for the fishery environment, serving as the perception center and decision-making core, plays a crucial role in real-time data collection, status assessment, and feedback regulation of the aquaculture ecosystem. Within the numerous subsystems of this system, the water flow control system holds a pivotal position. It not only determines the mass transfer efficiency of dissolved oxygen and the dilution rate of harmful metabolites such as ammonia nitrogen and nitrite, but also directly affects the thermodynamic distribution and flow field uniformity of the aquaculture water. Therefore, the ability of the water flow control system to achieve high-precision, low-fluctuation control of water flow directly impacts the stress level, growth rate, and ultimately, the economic benefits of aquaculture. In practical engineering applications, the aquaculture water flow control object exhibits significant strong nonlinearity, large inertial pure time delay, and time-varying parameter characteristics. The system model parameters are in a state of dynamic change due to the combined effects of internal state variables such as the evolution of aquaculture density and biological growth cycles, as well as external random disturbances such as rainfall and evaporation. Furthermore, the dead zone characteristics and hysteresis effects of actuators such as variable frequency pumps and electric valves, and the measurement noise of sensors in complex water quality environments, further increase the control difficulty. Therefore, PID control is introduced. To achieve more efficient tuning and optimization, metaheuristic algorithms such as genetic algorithms and particle swarm optimization are generally used for parameter optimization. Although these algorithms improve control performance to some extent, they are prone to insufficient global exploration ability when facing complex optimizations such as water flow control. They are easily trapped in local optima in multi-peak functions, resulting in PID parameters that are not globally optimal. Moreover, the balance between local exploration ability and convergence speed is fragile, and "premature convergence" or search stagnation due to loss of population diversity can easily occur in the later stages of optimization. Furthermore, the algorithm itself is extremely sensitive to hyperparameters such as population size and crossover mutation rate, increasing the debugging cost for engineering implementation. Therefore, given the complex operating conditions faced by water flow control in fishery environmental dynamic monitoring systems, there is an urgent need to develop a metaheuristic optimization algorithm with stronger global traversal capabilities, faster convergence speed, and better robustness to adaptively and precisely tune the parameters of the water flow PID controller, thereby overcoming the limitations of existing control strategies and achieving more accurate dynamic regulation of fishery water flow. Summary of the Invention

[0003] To overcome the technical problems described in the background, this invention provides a method for optimizing water flow control based on dynamic monitoring of the fishery environment. It utilizes reverse learning initialization to improve the coverage of the initial solution set, employs a Lévy flight strategy to enable individuals to escape local extrema, uses a vector difference strategy to correct physical dimensions and accelerate convergence, and applies a reverse defense restart mechanism to overcome iterative stagnation using spatial symmetry. Applying this algorithm to PID parameter tuning for water flow reduces the overshoot and steady-state error of the system's step response, shortens the settling time, and achieves precise dynamic control of water flow in the fishery environment.

[0004] The technical solution of this invention is: a method for optimizing water body flow control based on dynamic monitoring of the fishery environment, comprising the following steps:

[0005] S1. Construct a dynamic monitoring system for water flow PID control of aquaculture environment in aquaculture recirculating aquaculture system, including a water flow error calculation module, a water flow PID controller module, an improved optimization algorithm module for brood parasitism of gray butterfly, a water flow regulation module based on water pump frequency converter, and a main circulation pipeline water flow monitoring module.

[0006] S2. An improved optimization algorithm for brood parasitism in gray butterflies is introduced. The specific improvement strategy is as follows:

[0007] S21. Introduce a population initialization strategy based on reverse learning. By generating reverse solutions and selectively retaining the best ones, the coverage of the solution space by the initial population is improved. That is, generate random solutions and their reverse solutions, and select the best ones based on fitness. Individual;

[0008] S22. Introduce a parasitic infiltration strategy based on Lévy flight. By introducing random step sizes of varying lengths, the algorithm's ability to escape local optima is enhanced. That is, the Mantegna algorithm is used to generate Lévy step sizes and combined with an adaptive learning rate to update the position of parasitic solutions.

[0009] S23. Introduce a host manipulation strategy based on vector difference. By using the vector difference between individuals to replace the scalar operation of the original algorithm, the physical dimensions are corrected and the guidance accuracy is improved. That is, the host solution is simultaneously pulled by the difference vector of the parasitic guided individual and the global optimal solution.

[0010] S24. Introduce a defense restart strategy based on reverse learning. By mapping inferior individuals to symmetric positions in the search space, the symmetry information of the solution space is utilized. That is, when the fitness of the parasitic solution is inferior to the average level of the host, its reverse solution about the boundary is calculated.

[0011] S3. The improved parasitic brooding algorithm of the gray butterfly is used to tune the PID control parameters of the water flow in the dynamic monitoring system of the fishery environment, and the optimal control parameters are obtained through optimization. , , ;

[0012] S4. The three optimal control parameters obtained by using the improved gray butterfly brooding parasitism optimization algorithm are set as the PID control parameters of the water body flow in the dynamic monitoring water body flow PID control system of the fishery environment. The water body flow PID controller module uses the water body flow PID control parameters and real-time error to calculate the control quantity and transmit it to the water body flow regulation module. The water body flow regulation module controls the water body circulation flow in the main circulation pipeline by dynamically adjusting the output frequency of the water pump frequency converter, thereby realizing the rapid tracking and regulation of the target water body flow and the maintenance of steady-state accuracy.

[0013] Furthermore, in the fishery environment dynamic monitoring water body flow PID control system constructed in step S1, the main circulation pipeline water body flow monitoring module collects the actual water body flow in the main circulation pipeline in real time and transmits it to the water body flow error calculation module. The water body flow error calculation module receives the set target water body flow and calculates the difference between the target water body flow and the actual water body flow, then outputs the real-time error to the water body flow PID controller module. The improved gray butterfly brooding parasitism optimization algorithm module uses the error integral performance index of the fishery environment dynamic monitoring water body flow control system as the fitness function and the proportional coefficient of the water body flow PID controller module. Integral coefficient and differential coefficients After optimization and tuning, the water flow PID controller module constructs a control law based on the tuned parameters and real-time error, and outputs a control signal to the water flow regulation module based on the water pump frequency converter. The water flow regulation module based on the water pump frequency converter changes the water circulation flow in the main circulation pipeline by adjusting the output frequency of the water pump frequency converter.

[0014] Furthermore, the mathematical expression of the transfer function model of the fishery environment dynamic monitoring water flow control system constructed in step S1 is as follows:

[0015] ,

[0016] In the formula This represents the open-loop transfer function of the system. Indicates process gain. Indicates the principal time constant. Indicates pure time delay. This represents the Laplace transform operator.

[0017] Furthermore, the reverse learning mechanism in step S21 includes the following steps:

[0018] First, a random initial population is generated within the search space. Then, its reverse solution population with respect to the boundary is calculated. ;

[0019] ,

[0020] In the formula It is a reverse population. Let the lower bound vector of the search space be . Let the upper bound vector of the search space be , It is a random population.

[0021] Furthermore, the parasitic infiltration strategy based on Levy's flight in step S22 includes the following steps:

[0022] Using the Mantegna algorithm to generate step sizes that follow a Lévy distribution Step length It includes frequent short jumps and occasional long jumps, parasitic solutions The position update incorporates an adaptive learning rate that decays with the number of iterations. With Levi's stride length;

[0023] ,

[0024] In the formula For the updated parasitic solution location, This is the current location of the parasitic solution. To follow the step size of the Lévy distribution, This is the current optimal host solution location.

[0025] Furthermore, in step S22, the adaptive learning rate The formula for calculating the decay with the number of iterations is as follows:

[0026] ,

[0027] In the formula For adaptive learning rate, Indicates the current iteration number. This indicates the maximum number of iterations.

[0028] Furthermore, the host manipulation strategy based on vector difference in step S23 includes the following steps:

[0029] Each host solution During the update process, it is not only affected by randomly selected parasitic guide individuals. The influence is also affected by the global optimal solution. traction,

[0030] ,

[0031] In the formula To determine the location of the updated host, Solution for the host, For randomly selected parasitic guide individuals, This is the globally optimal solution. and It is a random vector.

[0032] Furthermore, the reverse learning-based defensive restart strategy in step S24 includes the following steps:

[0033] When parasitic solution fitness value Inferior to the average fitness of the host population At this point, random initialization is no longer performed; instead, the reverse position of the solution with respect to the boundary of the search space is calculated.

[0034] ,

[0035] In the formula To defend against the location of parasitic solutions after reboot, For parasitic solution, Let the lower bound vector of the search space be . Let the upper bound vector of the search space be denoted as .

[0036] If the fitness of the reverse solution is better than that of the original solution, then the reverse solution is replaced.

[0037] Furthermore, the fitness value in step S21 The fitness function formula used is as follows:

[0038] ,

[0039] In the formula Indicates the simulation time. This represents the real-time error between the target water flow rate and the actual water flow rate at time t. Indicates the actual settling time of the system. Indicates the target adjustment time. This indicates the adjustment of the time penalty weight. This represents the actual overshoot of the system. Indicates the target overshoot. This indicates the overshoot penalty weight.

[0040] Furthermore, the three optimal control parameters obtained in step S4 using the improved gray butterfly brood parasitism optimization algorithm are set as the parameters of the water flow PID controller in the fishery environment dynamic monitoring water flow PID control system. , , Subsequently, the mathematical expression for the PID control law of the water flow PID controller module, which outputs the control quantity, is as follows:

[0041] ,

[0042] In the formula This indicates the control signal output by the PID controller to the water flow regulation module. This represents the optimized scaling factor. This represents the optimized integral coefficient. Represents the optimized differential coefficients. This represents the real-time error between the target water flow rate and the actual water flow rate at time t.

[0043] The beneficial effects resulting from the adoption of the above technology are as follows:

[0044] 1. The present invention adopts a population initialization strategy based on reverse learning. The population initialization strategy based on reverse learning increases the coverage of the initial population in the solution space by calculating the reverse solution population of the random initial population with respect to the search space boundary and performing merging and selection.

[0045] 2. This invention adopts a parasitic penetration strategy based on Lévy flight. The parasitic penetration strategy based on Lévy flight uses the Mantegna algorithm to generate a random step size that follows the Lévy distribution. The parasitic penetration strategy based on Lévy flight uses the random step size that follows the Lévy distribution to update the position of the parasitic solution, so that the parasitic solution performs long-distance jumps and jumps out of the local extreme value region during the iteration process.

[0046] 3. The present invention adopts a host manipulation strategy based on vector difference. The host manipulation strategy based on vector difference uses the difference between the position vector of the parasitic guide individual and the position vector of the host solution, as well as the difference between the position vector of the global optimal solution and the position vector of the host solution, to calculate the update amount of the host solution. The host manipulation strategy based on vector difference maintains the consistency of physical dimensions in the position update operation and drives the host population to converge toward the global optimal region.

[0047] 4. The present invention adopts a defense restart strategy based on reverse learning. When the fitness value of the parasitic solution is inferior to the average fitness value of the host population, the defense restart strategy based on reverse learning calculates the symmetrical reverse position of the parasitic solution that triggers the defense mechanism with respect to the boundary of the search space. The defense restart strategy based on reverse learning utilizes the symmetry of the solution space to map individuals in the inferior region to the symmetrical position of the search space.

[0048] 5. This invention utilizes an improved parasitic brooding algorithm for gray butterflies to tune the proportional coefficient, integral coefficient, and derivative coefficient of the PID control system for water flow dynamic monitoring in fisheries environments. It is equipped with a PID controller for water flow with optimized control parameters, which reduces the overshoot in the step response process of water flow, shortens the adjustment time required for water flow to reach steady state, and reduces the steady-state error in the water flow regulation process. Attached Figure Description

[0049] Figure 1 This is a flowchart of the improved strategy introduced in the improved parasitic brooding optimization algorithm for gray butterflies of the present invention.

[0050] Figure 2 This is a comparison of the optimal fitness curves of the improved brood parasitism optimization algorithm for blue butterflies and the brood parasitism optimization algorithm for blue butterflies according to the present invention.

[0051] Figure 3 This is a comparison of the step response curves of the improved brood parasitism optimization algorithm for blue butterflies and the brood parasitism optimization algorithm for blue butterflies according to the present invention.

[0052] Figure 4 This is a comparison of the optimization curves of the water flow PID control parameters between the improved brood parasitism optimization algorithm of the present invention and the brood parasitism optimization algorithm of the present invention. Detailed Implementation

[0053] Example 1: As Figure 1 As shown, this invention provides a method for optimizing water body flow control based on dynamic monitoring of the fishery environment, comprising the following steps:

[0054] S1. Construct a dynamic monitoring PID control system for water flow in aquaculture recirculating aquaculture systems, including a water flow error calculation module, a water flow PID controller module, an improved optimization algorithm module for butterfly brood parasitism, a water flow regulation module based on a pump frequency converter, and a main circulation pipeline water flow monitoring module. The main circulation pipeline water flow monitoring module collects the actual water flow in the main circulation pipeline in real time and transmits it to the water flow error calculation module. The water flow error calculation module receives the set target water flow and calculates the difference between the target water flow and the actual water flow, then outputs the real-time error to the water flow PID controller module. The improved butterfly brood parasitism optimization algorithm module uses the error integral performance index of the dynamic monitoring water flow control system as the fitness function to adjust the proportional coefficient of the water flow PID controller module. Integral coefficient and differential coefficients The water flow PID controller module performs optimization tuning. Based on the tuned parameters and real-time error, it constructs a control law and outputs a control signal to the water flow regulation module based on the water pump frequency converter. The water flow regulation module based on the water pump frequency converter changes the water circulation flow in the main circulation pipeline by adjusting the output frequency of the water pump frequency converter.

[0055] S2. An improved optimization algorithm for brood parasitism in gray butterflies is introduced. The specific improvement strategy is as follows:

[0056] S21. Introduce a population initialization strategy based on reverse learning. By generating reverse solutions and selectively retaining the best ones, the coverage of the solution space by the initial population is improved. That is, generate random solutions and their reverse solutions, and select the best ones based on fitness. Individual;

[0057] S22. Introduce a parasitic infiltration strategy based on Lévy flight. By introducing random step sizes of varying lengths, the algorithm's ability to escape local optima is enhanced. That is, the Mantegna algorithm is used to generate Lévy step sizes and combined with an adaptive learning rate to update the position of parasitic solutions.

[0058] S23. Introduce a host manipulation strategy based on vector difference. By using the vector difference between individuals to replace the scalar operation of the original algorithm, the physical dimensions are corrected and the guidance accuracy is improved. That is, the host solution is simultaneously pulled by the difference vector of the parasitic guided individual and the global optimal solution.

[0059] S24. Introduce a defense restart strategy based on reverse learning. By mapping inferior individuals to symmetric positions in the search space, the symmetry information of the solution space is utilized. That is, when the fitness of the parasitic solution is inferior to the average level of the host, its reverse solution about the boundary is calculated.

[0060] S3. The improved parasitic brooding algorithm of the gray butterfly is used to tune the PID control parameters of the water flow in the dynamic monitoring system of the fishery environment, and the optimal control parameters are obtained through optimization. , , ;

[0061] S4. The three optimal control parameters obtained by using the improved gray butterfly brooding parasitism optimization algorithm are set as the PID control parameters of the water body flow in the dynamic monitoring water body flow PID control system of the fishery environment. The water body flow PID controller module uses the water body flow PID control parameters and real-time error to calculate the control quantity and transmit it to the water body flow regulation module. The water body flow regulation module controls the water body circulation flow in the main circulation pipeline by dynamically adjusting the output frequency of the water pump frequency converter, thereby realizing the rapid tracking and regulation of the target water body flow and the maintenance of steady-state accuracy.

[0062] Among them, the three optimal control parameters obtained by using the improved gray butterfly brood parasitism optimization algorithm are set as the parameters of the water flow PID controller in the dynamic monitoring water flow PID control system for fishery environment. , , Subsequently, the mathematical expression for the PID control law of the water flow PID controller module, which outputs the control quantity, is as follows:

[0063] ,

[0064] In the formula This indicates the control signal output by the PID controller to the water flow regulation module. This represents the optimized scaling factor. This represents the optimized integral coefficient. Represents the optimized differential coefficients. This represents the real-time error between the target water flow rate and the actual water flow rate at time t.

[0065] The mathematical expression for the transfer function model of the fishery environment dynamic monitoring water flow control system constructed in step S1 is as follows:

[0066] ,

[0067] In the formula This represents the open-loop transfer function of the system. Indicates process gain. Indicates the principal time constant. Indicates pure time delay. This represents the Laplace transform operator.

[0068] The reverse learning mechanism in step S21 includes the following steps:

[0069] First, a random initial population is generated within the search space. Then, its reverse solution population with respect to the boundary is calculated. ;

[0070] ,

[0071] In the formula It is a reverse population. Let the lower bound vector of the search space be . Let the upper bound vector of the search space be , For a random population; thus, by and Calculate fitness value after merging And select the top ones based on fitness values. The optimal individuals are used as the initial population, thus achieving a more comprehensive coverage of the solution space in the early stages of the algorithm's operation.

[0072] The parasitic infiltration strategy based on Levy's flight in step S22 includes the following steps:

[0073] Using the Mantegna algorithm to generate step sizes that follow a Lévy distribution Step length It includes frequent short jumps and occasional long jumps, parasitic solutions The position update incorporates an adaptive learning rate that decays with the number of iterations. With Levi's stride length;

[0074] ,

[0075] In the formula For the updated parasitic solution location, This is the current location of the parasitic solution. To follow the step size of the Lévy distribution, The current optimal host solution position is used, which enables the algorithm to have a wide range of exploration capabilities in the early stage and to perform fine-grained search in the later stage. Boundary reflection is used to handle out-of-bounds solutions to maintain population diversity.

[0076] Adaptive learning rate The formula for calculating the decay with the number of iterations is as follows:

[0077] ,

[0078] In the formula For adaptive learning rate, Indicates the current iteration number. This indicates the maximum number of iterations.

[0079] The host manipulation strategy based on vector difference in step S23 includes the following steps:

[0080] Each host solution During the update process, it is not only affected by randomly selected parasitic guide individuals. The influence is also affected by the global optimal solution. traction,

[0081] ,

[0082] In the formula To determine the location of the updated host, Solution for the host, For randomly selected parasitic guide individuals, This is the globally optimal solution. and Since the vectors are random, the physical meaning of the update quantities is ensured by the difference operation between the position vectors, and the convergence speed of the host population is accelerated by the guiding role of the global optimal solution.

[0083] The reverse learning-based defense restart strategy in step S24 includes the following steps:

[0084] When parasitic solution fitness value Inferior to the average fitness of the host population At this point, random initialization is no longer performed; instead, the reverse position of the solution with respect to the boundary of the search space is calculated.

[0085] ,

[0086] In the formula To defend against the location of parasitic solutions after reboot, For parasitic solution, Let the lower bound vector of the search space be . Let the upper bound vector of the search space be denoted as .

[0087] If the fitness of the reverse solution is better than that of the original solution, it is replaced. This utilizes the central symmetry of the solution space, allowing individuals trapped in local inferior regions to quickly jump to the other side of the search space, thereby improving the efficiency of the algorithm in escaping stagnation.

[0088] In step S3, the tuning of the PID control parameters for water flow in the dynamic monitoring system of fishery environment using the improved gray butterfly brood parasitism optimization algorithm includes the following steps:

[0089] Step 1: Initialize algorithm parameters;

[0090] For the aquaculture recirculating aquaculture system, a subsystem of the fishery environmental dynamic monitoring system, the water circulation flow rate in the main circulation pipeline is controlled by adjusting the output frequency of the water pump inverter. The control objective is to achieve rapid and stable adjustment under aeration or sewage discharge conditions when the flow rate setpoint is stepped from 50 cubic meters per hour to 80 cubic meters per hour. The proportional coefficient of the PID control system for the water flow rate of the fishery environmental dynamic monitoring system is set accordingly. Integral coefficient and differential coefficients Set the variable to be optimized, and make the variable dimension... The value is 3, which sets the upper bound vector of the search space. Values Lower bound vector Values Set the total population size The value is 50, representing the maximum number of iterations. The value is set to 200, and the fitness function is defined. To determine the error integral performance index of the fishery environment dynamic monitoring water body flow control system, a transfer function model of the fishery environment dynamic monitoring water body flow control system is constructed through system identification.

[0091] ,

[0092] In the formula This represents the open-loop transfer function of the system. This indicates the process gain and its value is 0.8. This represents the principal time constant and has a value of 15.0 seconds. This represents the pure time delay and has a value of 2.0 seconds. This represents the Laplace transform operator.

[0093] Step 2: Perform population initialization based on reverse learning;

[0094] In search space The size of the internal random generation is initial population ,calculate Reverse solution population within the search space ;

[0095]

[0096] In the formula Represents the reverse population matrix. Represents the lower bound vector of the search space. Represents the upper bound vector of the search space. This indicates random initialization of the population matrix;

[0097] right Elements that exceed the boundary are truncated so that their values ​​are within the specified range. and Between; constructing by and The candidate solution set is composed of individuals, and the fitness value of each individual in the candidate solution set is calculated using the fitness function. ;

[0098] ,

[0099] In the formula Represents the fitness value. This represents the simulation time and has a value of 100.0 seconds. This represents the real-time error between the target water flow rate and the actual water flow rate at time t. Indicates the actual settling time of the system. This indicates the target adjustment time and its value is 40.0 seconds. This indicates that the time penalty weight is adjusted and has a value of 100.0. This represents the actual overshoot of the system. This indicates the target overshoot and its value is 0.05. This indicates the overshoot penalty weight and its value is 200.0;

[0100] Sort the candidate solutions in ascending order of fitness value and select the top solution with the smallest fitness value. The current population consists of individuals; the individuals with the lowest fitness value in the current population are selected. Each individual is divided into a host solution set. ,in And the value is 45; the remaining population in the current population Each individual is divided into a parasitic solution set. ,in And its value is 5; record the global optimal solution in the current population. .

[0101] Step 3: Update the adaptive learning rate;

[0102] Start the iteration loop, if the current iteration number is... Less than Then proceed with the following steps: calculate the adaptive learning rate for the current iteration. ;

[0103] ,

[0104] In the formula Indicates the first The learning rate for the next iteration. Indicates the current iteration number. This indicates the maximum number of iterations.

[0105] Step 4: Execute the Lévy flight position update for the parasitic solution set;

[0106] parasitic disassembly Each body in Generate Levi's flight stride Among them, the Levi Flight Index And its value is 1.5;

[0107] Using Levi's flight stride and the current host set The optimal individual renew Location;

[0108] ,

[0109] In the formula Indicates the position of the updated candidate parasitic solution. Indicates the current number The location of a parasitic solution. This represents the Lévy flight step size vector generated based on the Mantegna algorithm. This indicates the location of the individual with the lowest fitness value in the current host solution set.

[0110] Step 5: Perform boundary reflection and selection of the parasitic solution set;

[0111] examine For each dimension component, if any component exceeds the boundary, calculate the reflection value of that component with respect to the boundary and remap it into the search space;

[0112] calculate The fitness value, if The fitness value is less than The fitness value is then used. replace ,like The fitness value is less than the global optimum. The fitness value is then updated. for .

[0113] Step 6: Execute the vector difference position update of the host solution set;

[0114] Host decomposition Each body in From Parasites One individual is randomly selected as the guide individual. Combined with the global optimal solution right Update;

[0115] ,

[0116] In the formula Indicates the updated candidate host solution location. Indicates the current number The location of the host solution. and This represents a random vector whose component values ​​are between 0 and 1. This indicates the location of a randomly selected parasitic guide individual. Indicates the current position of the global optimal solution;

[0117] right Perform boundary truncation; calculate The fitness value; if The fitness value is less than The fitness value is then used replace ;like The fitness value is less than The fitness value is then updated. for .

[0118] Step 7: Perform defense operations based on reverse learning;

[0119] Computational host solution set Average fitness value traversing the parasitic solution set Each body in ,like The fitness value is greater than Then calculate its inverse solution.

[0120] ,

[0121] In the formula Indicates the location of the generated inverse parasitic solution. Represents the lower bound vector of the search space. Represents the upper bound vector of the search space. Indicates the location of the parasitic solution that triggers the defense mechanism;

[0122] right Perform boundary truncation and calculate The fitness value; if The fitness value is less than The fitness value is then used replace ;like The fitness value is less than The fitness value is then updated. for .

[0123] Step 8: Terminate the iteration and output the results;

[0124] Determine the current iteration number Has it been achieved? If it is not achieved, then... Increment by 1 and return to Step 3; if the target has been reached, terminate the algorithm; output the global optimal solution. and will The three components are respectively determined as the proportional coefficients of the PID control system for water body flow in dynamic monitoring of the fishery environment. =7.4039, integral coefficient =0.4681 and differential coefficient =5.7247.

[0125] To verify the relative effectiveness of the improved Lycaenidae brood parasitism optimization algorithm for tuning the PID control parameters of water flow in a dynamic monitoring system for fishery environmental conditions, two processes were implemented in Matlab: one using the Lycaenidae brood parasitism optimization algorithm to tune the PID control parameters of water flow in the dynamic monitoring system for fishery environmental conditions, and the other using the improved Lycaenidae brood parasitism optimization algorithm. The same configuration parameters were used, and the results are shown below. Figures 2-4 As shown, it can be seen that the improved brood parasitism optimization algorithm for the blue butterfly is significantly better than the blue butterfly brood parasitism optimization algorithm in the tuning of PID control parameters for water flow in the dynamic monitoring water flow PID control system for fishery environment. For example, Figure 2 As shown, the optimal fitness value of the improved brood parasitism optimization algorithm for the blue butterfly is 132.437351, while that of the improved brood parasitism optimization algorithm is 385.143738. Furthermore, the improved brood parasitism optimization algorithm gets stuck in a local optimum around the 10th iteration, with its fitness value stagnating at a high level and no longer decreasing. In contrast, the improved brood parasitism optimization algorithm continues to show a step-like decreasing trend after the 10th iteration, reaching a lower stable fitness value plateau around the 50th iteration. This indicates that the improved brood parasitism optimization algorithm exhibits earlier convergence characteristics and relatively lacks the ability to escape local optima. The improved brood parasitism optimization algorithm, however, maintains its exploratory ability in the later stages of iteration and can obtain relatively better solutions. Figure 3 As shown, the system controlled by the optimized algorithm for brood parasitism in the blue butterfly exhibits significant overshoot and noticeable oscillations after the peak. The system controlled by the improved optimized algorithm has very small overshoot, a smooth rise, minimal oscillations, and enters steady state in approximately 10 seconds, significantly improving the quality of closed-loop control. Figure 4 As shown, the three parameters of the optimized brood parasitism algorithm for the gray butterfly stop changing before the 20th iteration and become linear. In contrast, the three parameters of the improved optimized brood parasitism algorithm for the gray butterfly are continuously adjusted during the iteration process and reach stability in the later stage of the iteration, thus obtaining a better combination of PID parameters and achieving a better balance between response speed and stability.

Claims

1. A method for optimizing water body flow control based on dynamic monitoring of the fishery environment, characterized in that: Includes the following steps: S1. Construct a dynamic monitoring system for water flow PID control of aquaculture environment in aquaculture recirculating aquaculture system, including a water flow error calculation module, a water flow PID controller module, an improved optimization algorithm module for brood parasitism of gray butterfly, a water flow regulation module based on water pump frequency converter, and a main circulation pipeline water flow monitoring module. S2. An improved optimization algorithm for brood parasitism in gray butterflies is introduced. The specific improvement strategy is as follows: S21. Introduce a population initialization strategy based on reverse learning. By generating reverse solutions and selectively retaining the best ones, the coverage of the solution space by the initial population is improved. That is, generate random solutions and their reverse solutions, and select the best ones based on fitness. Individual; S22. Introduce a parasitic infiltration strategy based on Lévy flight. By introducing random step sizes of varying lengths, the algorithm's ability to escape local optima is enhanced. That is, the Mantegna algorithm is used to generate Lévy step sizes and combined with an adaptive learning rate to update the position of parasitic solutions. S23. Introduce a host manipulation strategy based on vector difference. By using the vector difference between individuals to replace the scalar operation of the original algorithm, the physical dimensions are corrected and the guidance accuracy is improved. That is, the host solution is simultaneously pulled by the difference vector of the parasitic guided individual and the global optimal solution. S24. Introduce a defense restart strategy based on reverse learning. By mapping inferior individuals to symmetric positions in the search space, the symmetry information of the solution space is utilized. That is, when the fitness of the parasitic solution is inferior to the average level of the host, its reverse solution about the boundary is calculated. S3. The improved parasitic brooding algorithm of the gray butterfly is used to tune the PID control parameters of the water flow in the dynamic monitoring system of the fishery environment, and the optimal control parameters are obtained through optimization. , , ; S4. The three optimal control parameters obtained by using the improved gray butterfly brooding parasitism optimization algorithm are set as the PID control parameters of the water body flow in the dynamic monitoring water body flow PID control system of the fishery environment. The water body flow PID controller module uses the water body flow PID control parameters and real-time error to calculate the control quantity and transmit it to the water body flow regulation module. The water body flow regulation module controls the water body circulation flow in the main circulation pipeline by dynamically adjusting the output frequency of the water pump frequency converter, thereby realizing the rapid tracking and regulation of the target water body flow and the maintenance of steady-state accuracy.

2. The method for optimizing water body flow control based on dynamic monitoring of the fishery environment according to claim 1, characterized in that: In the fishery environment dynamic monitoring water body flow PID control system constructed in step S1, the main circulation pipeline water body flow monitoring module collects the actual water body flow in the main circulation pipeline in real time and transmits it to the water body flow error calculation module. The water body flow error calculation module receives the set target water body flow and calculates the difference between the target water body flow and the actual water body flow, then outputs the real-time error to the water body flow PID controller module. The improved gray butterfly brooding parasitism optimization algorithm module uses the error integral performance index of the fishery environment dynamic monitoring water body flow control system as the fitness function and the proportional coefficient of the water body flow PID controller module. Integral coefficient and differential coefficients After optimization and tuning, the water flow PID controller module constructs a control law based on the tuned parameters and real-time error, and outputs a control signal to the water flow regulation module based on the water pump frequency converter. The water flow regulation module based on the water pump frequency converter changes the water circulation flow in the main circulation pipeline by adjusting the output frequency of the water pump frequency converter.

3. The method for optimizing water body flow control based on dynamic monitoring of the fishery environment according to claim 2, characterized in that: The mathematical expression of the transfer function model of the fishery environment dynamic monitoring water flow control system constructed in step S1 is as follows: ， In the formula This represents the open-loop transfer function of the system. Indicates process gain. Indicates the principal time constant. Indicates pure time delay. This represents the Laplace transform operator.

4. The method for optimizing water body flow control based on dynamic monitoring of the fishery environment according to claim 1, characterized in that: The reverse learning mechanism in step S21 includes the following steps: First, a random initial population is generated within the search space. Then, its reverse solution population with respect to the boundary is calculated. ; ， In the formula It is a reverse population. Let the lower bound vector of the search space be . Let the upper bound vector of the search space be , It is a random population.

5. The method for optimizing water body flow control based on dynamic monitoring of the fishery environment according to claim 1, characterized in that: The parasitic infiltration strategy based on Levy's flight in step S22 includes the following steps: Using the Mantegna algorithm to generate step sizes that follow a Lévy distribution Step length It includes frequent short jumps and occasional long jumps, parasitic solutions The position update incorporates an adaptive learning rate that decays with the number of iterations. With Levi's stride length; ， In the formula For the updated parasitic solution location, This is the current location of the parasitic solution. To follow the step size of the Lévy distribution, This is the current optimal host solution location.

6. The method for optimizing water body flow control based on dynamic monitoring of the fishery environment according to claim 5, characterized in that: Adaptive learning rate in step S22 The formula for calculating the decay with the number of iterations is as follows: ， In the formula For adaptive learning rate, Indicates the current iteration number. This indicates the maximum number of iterations.

7. The method for optimizing water body flow control based on dynamic monitoring of the fishery environment according to claim 1, characterized in that: The host manipulation strategy based on vector difference in step S23 includes the following steps: Each host solution During the update process, it is not only affected by randomly selected parasitic guide individuals. The influence is also affected by the global optimal solution. traction, ， In the formula To determine the location of the updated host, Solution for the host, For randomly selected parasitic guide individuals, This is the globally optimal solution. and It is a random vector.

8. The method for optimizing water body flow control based on dynamic monitoring of the fishery environment according to claim 1, characterized in that: The reverse learning-based defensive restart strategy in step S24 includes the following steps: When parasitic solution fitness value Inferior to the average fitness of the host population At this point, random initialization is no longer performed; instead, the reverse position of the solution with respect to the boundary of the search space is calculated. ， In the formula To defend against the location of parasitic solutions after reboot, For parasitic solution, Let the lower bound vector of the search space be . Let the upper bound vector of the search space be denoted as . If the fitness of the reverse solution is better than that of the original solution, then the reverse solution is replaced.

9. The method for optimizing water body flow control based on dynamic monitoring of the fishery environment according to claim 1, characterized in that: fitness value in step S21 The fitness function formula used is as follows: ， In the formula Indicates the simulation time. This represents the real-time error between the actual water flow rate of the system and the target water flow rate at time t. Indicates the actual settling time of the system. Indicates the target adjustment time. This indicates the adjustment of the time penalty weight. This represents the actual overshoot of the system. Indicates the target overshoot. This indicates the overshoot penalty weight.

10. The method for optimizing water body flow control based on dynamic monitoring of the fishery environment according to claim 1, characterized in that: In step S4, the three optimal control parameters obtained by using the improved gray butterfly brood parasitism optimization algorithm are set as the parameters of the water flow PID controller in the fishery environment dynamic monitoring water flow PID control system. , , Subsequently, the mathematical expression for the PID control law of the water flow PID controller module, which outputs the control quantity, is as follows: ， In the formula This indicates the control signal output by the PID controller to the water flow regulation module. This represents the optimized scaling factor. This represents the optimized integral coefficient. Represents the optimized differential coefficients. This represents the real-time error between the target water flow rate and the actual water flow rate at time t.

Images