Power distribution cabinet temperature control optimization method based on improved dead leaf butterfly optimization algorithm

Abstract

The invention discloses a power distribution cabinet temperature control optimization method based on an improved dead leaf butterfly optimization algorithm, and the method comprises the steps: S1, building a to-be-optimized power distribution cabinet temperature control mathematical model based on the temperature control demands of a power distribution cabinet; s2, constructing a temperature PID control system of the power distribution cabinet; s3, introducing an improved dead leaf butterfly optimizer; s4, setting the in-cabinet temperature PID control parameters in the power distribution cabinet temperature PID control system by using the improved dead leaf butterfly optimizer, and obtaining the optimal control parameters through optimization; and S5, optimal control parameters obtained through optimization are set as parameters of a temperature PID controller in the power distribution cabinet temperature PID control system, and the temperature control effect is optimized. According to the invention, a dead leaf butterfly optimization algorithm is introduced and improved, so that the stability of the temperature control system of the power distribution cabinet can be improved, the temperature can be more quickly stabilized to be close to a temperature preset value, and related equipment or electronic components in the power distribution cabinet are prevented from being damaged by relatively large temperature fluctuation.

Description

Technical Field

[0001] The present invention belongs to the technical field of control optimization, and in particular relates to a temperature control optimization method for a power distribution cabinet based on an improved dead leaf butterfly optimization algorithm. Background Art

[0002] As a crucial piece of equipment in power systems, distribution cabinets are widely used in a variety of fields, including industrial control, building power, data centers, and new energy. Their primary function is to distribute, protect, control, and monitor electrical energy, ensuring the safe and stable operation of the entire electrical system. However, as the load density of distribution systems continues to increase, the number of electrical components integrated within distribution cabinets has increased significantly, leading to a corresponding increase in heat generation. Excessive temperatures within distribution cabinets can accelerate component aging, reduce equipment reliability, and even cause safety incidents such as fires. Therefore, temperature control of distribution cabinets requires paramount attention.

[0003] Currently, temperature control in power distribution cabinets typically relies on devices such as temperature sensors and fans, coupled with PID controllers for automatic adjustment. PID controllers are widely used in industrial temperature control due to their simple structure, strong real-time performance, and ease of implementation. However, power distribution cabinet temperature control systems often face complex operating conditions such as nonlinearity, time-varying performance, and external interference, making it difficult for traditional PID control to achieve ideal control results. In particular, improper initial parameter tuning can lead to problems such as slow system response, large overshoot, and large steady-state errors.

[0004] Traditional PID parameter tuning methods primarily include empirical methods, trial-and-error methods, and model-based analysis. Empirical methods rely on the engineer's experience and system response curves, are highly subjective, and are difficult to adapt to diverse operating conditions. Trial-and-error methods are time-consuming, inefficient, and susceptible to human error. Model-based analysis methods require high accuracy of the system model and struggle to cope with complex systems characterized by time lag, nonlinearity, and disturbances. Therefore, efficiently and accurately tuning PID controller parameters to improve the robustness and control performance of distribution cabinet temperature control systems has become a pressing technical challenge in current research and engineering practice. Summary of the invention

[0005] In order to overcome the technical problems described in the above background technology, the present invention provides a distribution cabinet temperature control optimization method based on an improved dead leaf butterfly optimization algorithm. The dead leaf butterfly optimization algorithm is introduced and improved, which can improve the stability of the distribution cabinet temperature control system, so that the temperature can be stabilized at the preset temperature value more quickly, avoiding large temperature fluctuations that may damage related equipment or electronic components in the distribution cabinet.

[0006] The technical solution of the present invention is: a distribution cabinet temperature control optimization method based on an improved dead leaf butterfly optimization algorithm, comprising the following steps: S1. Based on the temperature control requirements of the power distribution cabinet, a mathematical model for the temperature control of the power distribution cabinet to be optimized is constructed; S2. Build a power distribution cabinet temperature PID control system, including a cabinet temperature threshold setting module, a cabinet temperature PID controller module, an improved dead leaf butterfly optimization algorithm module, a cabinet temperature adjustment module, and a cabinet temperature monitoring module; S3, introduce the improved dead leaf butterfly optimization algorithm, the specific improvement strategy is: S31, introduce three update strategies, namely adaptive collective color change strategy, differential imitation color change strategy and adaptive recall color change strategy, to improve the exploration phase of the dead leaf butterfly optimization algorithm; S32. Introduce the elite-guided differential migration strategy to improve the migration mechanism of the dead leaf butterfly optimization algorithm, and select the top population with the best fitness value from the current population. % of individuals, forming the elite population , by performing differential mutation on the elite population to generate migration targets , for ordinary individuals selected for migration, For the individuals in the selected elite population, a perturbation strategy of Gaussian mutation is used; S4. Use the improved dead leaf butterfly optimization algorithm to adjust the cabinet temperature PID control parameters in the distribution cabinet temperature PID control system, and obtain the best control parameters by searching for the best , , ; S5. The three optimal control parameters obtained by optimizing the improved dead leaf butterfly optimization algorithm are set as the parameters of the temperature PID controller in the temperature PID control system of the distribution cabinet to optimize the temperature control effect.

[0007] Furthermore, in step S1, the temperature control mathematical model of the power distribution cabinet is: , in, Indicates the final change in the temperature inside the power distribution cabinet under the action of the unit control quantity, which is used to reflect the sensitivity of the system to the cooling adjustment input. and It is the process time constant that reflects the inertia of the distribution cabinet during the heating and cooling process. It is the pure delay time that represents the time lag between the action of the regulating device and the detection of a significant temperature change by the temperature sensor. is the independent variable of the Laplace transform.

[0008] Furthermore, in the distribution cabinet temperature PID control system constructed in step S2, the cabinet temperature threshold is obtained through the cabinet temperature threshold setting module, and the current temperature difference is obtained based on the actual cabinet temperature monitored by the temperature monitoring module and the cabinet temperature threshold. The current temperature difference is input into the PID controller module, and the parameters of the PID controller module are optimized through the improved dead leaf butterfly optimization algorithm module. The optimized PID controller outputs the control quantity, and the control quantity is input into the cabinet temperature adjustment module for adjustment.

[0009] Furthermore, in step S31, the exploration phase of the improved leaf butterfly optimization algorithm includes the following steps: When randomly selecting an update from the adaptive collective color-changing strategy, differential imitation color-changing strategy, and adaptive recall color-changing strategy, the specific steps are: C1, the adaptive collective color change strategy includes the current individual based on its own and the global optimal individual The corresponding formula is: , in, is the updated position, is the individual position of the current iteration, is the adaptive weight, and is the random disturbance term; C2, the differential imitation color change strategy involves generating a new position by the differential vectors of three different individuals randomly selected from the population. The corresponding formula is: , in, is the updated position, 、 、 are the positions of three different individuals selected randomly, and F is the dynamic scaling factor; C3, the adaptive recall color change strategy includes the current individual based on its own and the individual's historical optimal position The corresponding formula is: , in, is the updated position, is the individual position of the current iteration, is the adaptive weight, and is a random disturbance term.

[0010] Furthermore, the adaptive weight and the probability of choosing the exploration phase With the number of iterations The square of the ratio of the maximum number of iterations T is nonlinearly attenuated, and the corresponding formulas are: , , Where T is the maximum number of iterations, and are the maximum and minimum values of the adaptive weight respectively.

[0011] Furthermore, the migration mechanism of the improved leaf butterfly optimization algorithm in step S32 includes the following steps: S321, when continuous When the global optimal solution is not improved after the first iteration, the migration mechanism is triggered, and the top p% of individuals are selected according to the fitness value to form an elite population, where is the preset number of stagnation tolerances; S322, by three random individuals in the elite population 、 and Perform differential mutation to generate an elite-guided migration target , ; S323: For non-elite migrants, guide them to migrate towards elite-guided destinations Perform random step moves, , where D is the dimension of the solution space; S324. Apply a Gaussian mutation perturbation that decays with the number of iterations to the individuals in the elite population to perform local fine search.

[0012] Furthermore, in step S324, the Gaussian mutation applied to the elite population individuals has a standard deviation of With the number of iterations Linearly decreasing, the mathematical expression is: , Among them, the standard deviation of the Gaussian distribution With the number of iterations linearly decreasing, and , is the current number of iterations, and T is the maximum number of iterations.

[0013] Furthermore, in step S4, the improved dead leaf butterfly optimization algorithm is used to adjust the cabinet temperature PID control parameters in the distribution cabinet temperature PID control system, including the following steps: S41, set the ratio ,integral ,differential The range of parameter values constitutes the search space for optimization; S42. Initialize the population of the improved dead leaf butterfly optimization algorithm, where each individual represents a set of PID parameters; S43. Take the integral criterion of the time product absolute error of the power distribution cabinet temperature PID control system as the fitness function, and iteratively execute the population update and migration steps of the improved dead leaf butterfly optimization algorithm; S44. When the preset maximum number of iterations or the fitness value converges, the algorithm terminates, and the current optimal set of PID parameters is output as the final result.

[0014] By adopting the above technical solutions, the beneficial effects of the present invention are as follows: The dead leaf butterfly optimization algorithm is improved. The exploration stage of the dead leaf butterfly optimization algorithm is improved by introducing differential mutation and adaptive weight strategies. The migration mechanism of the dead leaf butterfly optimization algorithm is improved by introducing an elite-guided differential migration strategy, which improves the overall optimization performance of the algorithm. In the mimicry color change in the exploration stage, the current individual learns from a random individual, but the learning direction and step size are determined by the difference vector of two other random individuals. This method can generate more diverse new positions and greatly enhance the global exploration ability. In the collective color change and recall color change in the exploration stage, an adaptive inertia weight that changes with the iteration number t is introduced. In the initial stage, it explores more towards the global optimum, and in the later stage, it conducts fine searches more near the individual historical optimum. And if the stagnation condition is triggered and migration behavior needs to occur, the migration is no longer a blind random jump, but is guided by the collective wisdom of the elite population. This increases the probability of finding a better solution area. At the same time, different migration strategies are adopted for ordinary individuals and elite individuals, which not only ensures the global exploration ability of the algorithm, the bold jumps of ordinary individuals, but also protects the high-quality solutions that have been found, the careful fine-tuning of elite individuals, and realizes the dynamic balance between exploration and exploitation. Using the improved dead leaf butterfly optimization algorithm for the optimization of the power distribution cabinet temperature PID parameters can improve the stability of the power distribution cabinet temperature control system, enable the temperature to stabilize faster near the preset temperature value, and avoid large temperature fluctuations from damaging the relevant equipment or electronic components in the power distribution cabinet. Description of the Drawings

[0015] Figure 1 is the flow schematic diagram of the present invention.

[0016] Figure 2 is the model diagram of the power distribution cabinet temperature PID control system of the present invention.

[0017] Figure 3 is the comparison chart of the optimal ITAE between the standard dead leaf butterfly optimization algorithm and the improved dead leaf butterfly optimization algorithm.

[0018] Figure 4 is the optimization parameter values of the standard dead leaf butterfly optimization algorithm and the improved dead leaf butterfly optimization algorithm Comparison diagram

[0019] Figure 5 They are the optimization parameter values of the standard dead leaf butterfly optimization algorithm and the improved dead leaf butterfly optimization algorithm Comparison diagram

[0020] Figure 6 They are the optimization parameter values of the standard dead leaf butterfly optimization algorithm and the improved dead leaf butterfly optimization algorithm Comparison diagram

[0021] Figure 7 They are the comparison diagrams of the optimization PID control effects of the standard dead leaf butterfly optimization algorithm and the improved dead leaf butterfly optimization algorithm Specific implementation method

[0022] Example 1: As Figures 1 - 7 shown, the present invention provides an optimized method for controlling the temperature of a power distribution cabinet based on an improved dead leaf butterfly optimization algorithm, including steps S1 to S5

[0023] S1. Construction of the mathematical model for temperature control of the power distribution cabinet. The temperature change of the power distribution cabinet mainly involves three aspects: internal heat generation, heat retention, and heat dissipation (including passive heat dissipation and active cooling), which can be equivalent to a second-order plus delay model with inertia and pure delay characteristics. Therefore, the mathematical model for temperature control of the power distribution cabinet is: , wherein, represents the amplitude of the final temperature change inside the power distribution cabinet under the action of a unit control quantity, and is used to reflect the sensitivity of the system to the input of cooling regulation, and are the process time constants reflecting the inertia of the heating and cooling processes of the power distribution cabinet, is the pure delay time representing the time lag between the action of the regulating device and the obvious temperature change detected by the temperature sensor, is the independent variable of the Laplace transform

[0024] Therefore, for an industrial power distribution cabinet with an internal frequency converter and forced air cooling using a fan, take = , representing the thermal inertia of the core components = 180 seconds, representing the thermal inertia of the air circulation inside the cabinet = 50 seconds, representing the combined delay of air flow and sensor = 20 seconds, and the transfer function formula is: .

[0025] S2. Construct a temperature PID control system for the power distribution cabinet and optimization objectives, including an in-cabinet temperature threshold setting module, an in-cabinet temperature PID controller module, an improved dead leaf butterfly optimization algorithm module, an in-cabinet temperature regulation module, and an in-cabinet temperature monitoring module. The output u(t) of the PID controller consists of three parts: proportional, integral, and derivative: , where the optimization variables are the proportional gain , the integral time , and the derivative time , which have a one-to-one correspondence with , , . Therefore, this optimization process is to find the optimal PID control parameters.

[0026] At the same time, the optimization objective is to minimize the integral of time-weighted absolute error ITAE when the system is subjected to a step disturbance (such as a sudden increase in the external environmental temperature or a sudden increase in the internal device load), which can impose a heavier penalty on long-existing errors and can effectively obtain a control effect with fast response and small overshoot. The corresponding fitness function formula is: , where is the set temperature threshold, and is the current temperature inside the power distribution cabinet.

[0027] S3. Introduce an improved dead leaf butterfly optimization algorithm, and its specific improvement strategies and implementation steps are S31 and S32.

[0028] S31. Hybrid update strategy for exploration and exploitation stages. In each iteration, for each individual in the population, first calculate an exploration probability that nonlinearly decays with the iteration number t , where T is the maximum number of iterations.

[0029] Then define an adaptive inertia weight , whose value nonlinearly decays from 0.9 to 0.2, that is: .

[0030] Also define a dynamic disturbance amplitude scaling factor , whose value linearly decays from 1 to 0.2, that is: .

[0031] The basic perturbation vector used in all update strategies is defined as: , where represents a three-dimensional random vector whose components follow a standard normal distribution, and I is the identity matrix.

[0032] Then a random number R that randomly takes values within [0, 1] is generated. If , it enters the exploration stage and randomly selects one of the following three strategies C1, C2, and C3 to execute.

[0033] C1. The adaptive collective color change strategy includes the current individual moving according to the position difference between itself and the global optimal individual . The corresponding formula is: , where is the updated position, is the position of the individual at the current iteration, is the adaptive weight, is a random number that follows a standard normal distribution, is the basic perturbation vector.

[0034] C2. The differential imitation color change strategy includes generating a new position through the differential vectors of three different individuals randomly selected from the population. The corresponding formula is: , where is the updated position, , , are three different individuals randomly selected respectively, F is the dynamic scaling factor, such that the step size of differential mutation also adapts to the iteration process.

[0035] C3. The adaptive recall color change strategy includes the current individual moving according to the difference between itself and the individual historical optimal position . The corresponding formula is: , where is the updated position, is the position of the individual at the current iteration, is the adaptive weight, is a random number that follows a standard normal distribution, is the basic perturbation vector.

[0036] If , it enters the exploitation stage and executes a strategy that combines the global optimal 、Individual historical optimum and weighted average update strategy of neighboring individual information: , where is the individual with the closest Euclidean distance to itself among the population except itself, , , are three independent random weights uniformly distributed in [0, 1].

[0037] Generate a new position and perform boundary checking to ensure it is within the search space [lb, ub], then calculate its fitness value . If it is better than the current individual, update the individual position and the individual historical optimum position, and if it is better than the global optimum, update the global optimum solution and the optimal fitness value .

[0038] S32. Introduce an elite-guided differential migration mechanism, including steps S321 to S324.

[0039] S321. Set a stagnation counter. If the global optimum solution has not been improved for = 3 consecutive times, trigger the migration mechanism. At this time, sort all individuals in the current population according to their fitness values, and select the top 10% and no less than 3 individuals to form an elite population .

[0040] S322. Randomly select three different individuals from the elite population , and , and generate an elite-guided migration target through differential mutation.

[0041] .

[0042] S323. For non-elite individual migration, randomly select = 20% of the individuals who have not been selected as elites for migration. These selected individuals move towards the migration target : , where rand(1, D) is a D-dimensional random vector with each component uniformly distributed in [0, 1], D is the dimension of the optimization problem, and the current value is 3. Indicates the new position vector calculated after performing the migration operation towards the migration target The new position vector calculated after performing the migration operation towards the migration target is the updated state of this individual and will be used for subsequent fitness evaluation and population update Indicates the individual The original position vector before performing the migration operation is the starting state of this individual in the current iteration step

[0043] S324. To enhance the local exploration ability of the elite population, a multiplicative Gaussian mutation perturbation that decays with the number of iterations is applied to each individual in the elite population , where the standard deviation of the Gaussian distribution decreases linearly with the number of iterations and , where T is the maximum number of iterations

[0044] All new individuals generated through migration and perturbation also need to be subjected to boundary checking and fitness evaluation, and the population, individual best, and global best information are updated accordingly. After the migration mechanism is triggered, the stagnation counter is reset to 0

[0045] S4. Use the improved dead leaf butterfly optimization algorithm to tune the in-cabinet temperature PID control parameters in the distribution cabinet temperature PID control system, and obtain the optimal control parameters through optimization , , , including steps S401 to S404

[0046] S401. Parameter initialization, that is, setting the algorithm parameters, including population size N = 50, optimization dimension D = 3, maximum number of iterations T = 50, setting the search space for PID parameters, for example, lower limit lb = [−80, −80, −40], upper limit ub = [0 0 0], setting the stagnation tolerance number to 3, migration individual ratio to 0.2, and elite population ratio to 0.1

[0047] S402. Population initialization, that is, randomly generating N individuals within the search space set in step S401, and each individual represents a set of [Kp, Ki, Kd] parameters

[0048] S403. Iterative optimization: Using the ITAE integral as the fitness function, in the loop from t = 1 to T, execute all the update steps of the improved dead leaf butterfly optimization algorithm described in S3, including the hybrid update strategy and the migration mechanism triggered when conditions are met

[0049] S404. Termination and output: After reaching the maximum number of iterations T, the algorithm terminates and outputs the globally optimal solution recorded at this time , which are the optimal PID parameters obtained , , .

[0050] S5. Set the three optimal control parameters obtained by optimizing with the improved dead leaf butterfly optimization algorithm as the parameters of the temperature PID controller in the power distribution cabinet temperature PID control system to optimize the temperature control effect.

[0051] In the power distribution cabinet temperature PID control system constructed in step S2, obtain the temperature threshold inside the cabinet through the temperature threshold setting module inside the cabinet, and obtain the current temperature difference based on the actual temperature inside the cabinet monitored by the temperature monitoring module and the temperature threshold inside the cabinet. Input the current temperature difference into the PID controller module, and optimize the parameters of the PID controller module through the improved dead leaf butterfly optimization algorithm module. The optimized PID controller outputs a control quantity, and input the control quantity into the temperature adjustment module inside the cabinet for adjustment.

[0052] By converting the above technical solution into project code running in Matlab, establish a corresponding power distribution cabinet temperature PID control system model in Simulink, initialize the parameters of the improved dead leaf butterfly optimization algorithm, including the population size N = 50 of the algorithm, the problem dimension d = 3, the maximum number of iterations 50, the upper limit ub = [0 0 0] of the search space, and the lower limit lb = [-80 -80 -40]. Run the project code to obtain the optimal ITAE comparison curve graph of the dead leaf butterfly optimization algorithm and the improved dead leaf butterfly optimization algorithm, as Figure 3 shown. The improved dead leaf butterfly optimization algorithm reaches approximately at the 24th iteration . Compared with the standard dead leaf butterfly optimization algorithm, the optimization accuracy is better and the convergence speed is faster, which can better improve the control performance of the power distribution cabinet temperature PID control system, thus achieving a better temperature control effect inside the power distribution cabinet.

[0053] And as Figures 4 - 6 shown, the PID controller in the power distribution cabinet temperature PID control system is respectively parameter-tuned by the dead leaf butterfly optimization algorithm and the improved dead leaf butterfly optimization algorithm. The optimal PID control parameters obtained by the dead leaf butterfly optimization algorithm are = -11.068, = -0.026, = -21.539. The optimal PID control parameters obtained by the improved dead leaf butterfly optimization algorithm are = -3.476, = -0.016, = -7.064. Then, input the optimal control parameters into the PID controller of the power distribution cabinet temperature PID control system respectively to obtain the corresponding optimal control effect, asFigure 7 As shown in the figure, it is a comparison chart of the optimization effects of the dead leaf butterfly optimization algorithm and the improved dead leaf butterfly optimization algorithm on PID. The target input value is a step signal of -1, which is used to simulate the temperature drop process in the temperature control of the distribution cabinet. Among them, the response speed of the response curve of the improved dead leaf butterfly optimization algorithm for optimizing the PID control effect is relatively fast and the overshoot is relatively low, so that the temperature inside the distribution cabinet can reach a steady state faster after adjustment, and the control effect on the temperature inside the cabinet is better.

Claims

1. An optimized method for temperature control of a power distribution cabinet based on an improved kallima inachus optimization algorithm, characterized in that: The steps include: S1. Based on the temperature control requirements of the power distribution cabinet, a mathematical model for the temperature control of the power distribution cabinet to be optimized is constructed; S2. Build a power distribution cabinet temperature PID control system, including a cabinet temperature threshold setting module, a cabinet temperature PID controller module, an improved dead leaf butterfly optimization algorithm module, a cabinet temperature adjustment module, and a cabinet temperature monitoring module; S3, introduce the improved dead leaf butterfly optimization algorithm, the specific improvement strategy is: S31, introduce three update strategies, namely adaptive collective color change strategy, differential imitation color change strategy and adaptive recall color change strategy, to improve the exploration phase of the dead leaf butterfly optimization algorithm; S32. Introduce the elite-guided differential migration strategy to improve the migration mechanism of the dead leaf butterfly optimization algorithm, and select the top population with the best fitness value from the current population. % of individuals, forming the elite population , by performing differential mutation on the elite population to generate migration targets , for ordinary individuals selected for migration, For the individuals in the selected elite population, a perturbation strategy of Gaussian mutation is used; S4. Use the improved Kallima inachus optimization algorithm to tune the cabinet temperature PID control parameters in the power distribution cabinet temperature PID control system, and obtain the optimal control parameters through optimization , , ; S5. The three optimal control parameters obtained by optimizing the improved dead leaf butterfly optimization algorithm are set as the parameters of the temperature PID controller in the temperature PID control system of the distribution cabinet to optimize the temperature control effect.

2. The optimized method for controlling the temperature of a power distribution cabinet based on an improved dead leaf butterfly optimization algorithm according to claim 1, characterized in that: In step S1, the temperature control mathematical model of the power distribution cabinet is: ， in, Indicates the final change in the temperature inside the power distribution cabinet under the action of the unit control quantity, which is used to reflect the sensitivity of the system to the cooling adjustment input. and It is the process time constant that reflects the inertia of the distribution cabinet during the heating and cooling process. It is the pure delay time that represents the time lag between the action of the regulating device and the detection of a significant temperature change by the temperature sensor. is the independent variable of the Laplace transform.

3. An optimized method for temperature control of a power distribution cabinet based on an improved Kallima inachus optimization algorithm according to claim 2, characterized in that: In the distribution cabinet temperature PID control system constructed in step S2, the cabinet temperature threshold is obtained through the cabinet temperature threshold setting module, and the current temperature difference is obtained based on the actual cabinet temperature monitored by the temperature monitoring module and the cabinet temperature threshold. The current temperature difference is input into the PID controller module, and the parameters of the PID controller module are optimized through the improved dead leaf butterfly optimization algorithm module. The optimized PID controller outputs the control quantity, and the control quantity is input into the cabinet temperature adjustment module for adjustment.

4. An optimized method for temperature control of a power distribution cabinet based on an improved dead leaf butterfly optimization algorithm according to claim 3, characterized in that: In the exploration phase of the improved Kallima inachus optimization algorithm in step S31, when the generated random number is less than , randomly select one of the adaptive collective color change strategy, differential imitation color change strategy, and adaptive recall color change strategy to perform an update. Specifically: C1. The adaptive collective color-changing strategy includes that the current individual moves according to the position difference between itself and the globally optimal individual and the corresponding formula is: , Among them, is the updated position, is the individual position of the current iteration, is the adaptive weight, and is the random perturbation term; C2, the differential imitation color change strategy involves generating a new position by the differential vectors of three different individuals randomly selected from the population. The corresponding formula is: ， Among them, is the updated position, , , are the positions of three randomly selected different individuals respectively, and F is the dynamic scaling factor; C3, the adaptive recall color change strategy includes the current individual based on its own and the individual's historical optimal position The corresponding formula is: ， Among them, is the updated position, is the individual position of the current iteration, is the adaptive weight, and is the random perturbation term.

5. The optimized method for controlling the temperature of a power distribution cabinet based on an improved Kallima inachus optimization algorithm according to claim 4, characterized in that: Adaptive weight and the probability of the selection and exploration stage both adopt non-linear attenuation according to the square of the ratio of the number of iterations and the maximum number of iterations T, and the corresponding formulas are respectively as follows: ， ， Where T is the maximum number of iterations, and are the maximum and minimum values of the adaptive weight respectively.

6. The method for optimizing temperature control of a power distribution cabinet based on an improved dead leaf butterfly optimization algorithm according to claim 5, characterized in that: The migration mechanism of the improved leaf butterfly optimization algorithm in step S32 includes the following steps: S321. When the global optimal solution has not been improved after consecutive iterations, trigger the migration mechanism, and select the top p% of individuals with the best fitness values to form an elite population, where is the preset stagnation tolerance count; S322. Generate an elite-guided migration target by performing differential mutation on three random individuals in the elite population , and . ;​​​​​​​​​​ S323. For non-elite migrating individuals, direct them towards the migration goal guided by the elite to perform a random step movement, , where D is the dimension of the solution space; S324. Apply a Gaussian mutation perturbation that decays with the number of iterations to the individuals in the elite population to perform local fine search.

7. An optimization method for the temperature control of a power distribution cabinet based on an improved Kallima inachus optimization algorithm, characterized in that: In step S324, for the individuals in the elite population, the standard deviation of the Gaussian mutation decreases linearly with the number of iterations and its mathematical expression is: ， Among them, the standard deviation of the Gaussian distribution decreases linearly with the number of iterations , and , where \(i\) is the current number of iterations and \(T\) is the maximum number of iterations.

8. An optimized method for temperature control of a power distribution cabinet based on an improved dead leaf butterfly optimization algorithm according to claim 7, characterized in that: In step S4, the improved dead leaf butterfly optimization algorithm is used to adjust the cabinet temperature PID control parameters in the distribution cabinet temperature PID control system, including the following steps: S41. Set the ratio . Integral . Differential . The value range of the parameters constitutes the search space for optimization; S42, initializing a population of the improved dead leaf butterfly optimization algorithm, wherein each individual represents a set of PID parameters; S43, using the time product absolute error integral criterion of the distribution cabinet temperature PID control system as the fitness function, iteratively executing the population update and migration steps of the improved dead leaf butterfly optimization algorithm; S44. When the preset maximum number of iterations is reached or the fitness value converges, the algorithm terminates and outputs the current optimal set of PID parameters as the final result.

Images