Grid-connected inverter parameter optimization method based on improved genetic particle swarm algorithm

Abstract

The invention discloses a grid-connected inverter parameter optimization method based on an improved genetic particle swarm algorithm. The grid-connected inverter parameter optimization method comprises the following steps: constructing a target function; performing population position initialization; calculating a fitness value of each particle at an initial moment, recording an optimal solution, and storing initial global optimal particle and individual optimal particle information; random disturbance is added to the speed and the position of the particles; updating the speed and the position of the particle; performing multi-point crossover and mutation operation; according to the current optimal solution, carrying out local domain search by adopting a fine search mode based on Laplacian distribution and carrying out global neighborhood search by adopting a search mode based on Cauchy distribution; and checking convergence conditions, if convergence occurs, outputting an optimal solution, and obtaining optimal parameters of the grid-connected inverter system. The method has the remarkable effects that by optimizing three key parameters in a control system and utilizing a nonlinear dynamic strategy to enhance the global search capability and the convergence speed of the algorithm, various defects in the prior art can be overcome.

Description

Technical Field

[0001] The invention relates to the technical field of parameter optimization of a grid-connected inverter control system, and in particular to a method for optimizing parameters of a grid-connected inverter based on an improved genetic particle swarm algorithm. Background Art

[0002] In the new energy fields such as power electronics, wind power, photovoltaics, etc., the performance of the grid-connected inverter control system will directly affect the dynamic response, stability and efficiency of the entire system. In these complex systems, the design of the control system usually involves the optimization of multiple key parameters, such as the inner loop control bandwidth, outer loop control bandwidth and phase-locked loop control bandwidth. The reasonable configuration of these parameters is crucial to improving the dynamic performance, steady-state error, robustness, etc. of the system.

[0003] However, the optimization process of control parameters usually faces complex problems such as nonlinearity, non-convexity, and multiple extreme values. Traditional optimization algorithms are difficult to balance global search and local fine optimization, and are prone to fall into local optimality, resulting in unsatisfactory optimization results. Particle swarm optimization (PSO) and genetic algorithm (GA) are classic global optimization algorithms and are widely used for tuning control parameters. However, standard PSO and GA have some limitations in the search process, such as insufficient population diversity, slow convergence speed, and easy to fall into local optimal solutions. In addition, with the expansion of the scale and complexity of the problem, a single optimization algorithm is difficult to meet the dual requirements of accuracy and speed in practical applications. Therefore, combining multiple optimization strategies to enhance the algorithm's global search capability, convergence speed, and optimization accuracy has become an important direction in current optimization algorithm research. For example:

[0004] The Chinese invention patent with application number CN202410956640.3 proposes a parameter optimization method for the temperature control system of a cement production decomposition furnace based on a genetic algorithm. The MPC model in the present invention adopts a state space model, and there are many designed control parameters, and the parameters interact with each other. Therefore, it is time-consuming and laborious to adopt the artificial experience trial and error method. The GA algorithm has good global search capabilities and can quickly search out all solutions in the solution space without falling into the trap of rapid descent of local optimal solutions; and using its inherent parallelism, it can easily perform distributed computing and speed up the solution. At the same time, the present invention designs a rolling time domain fitness function, which can improve the stability and accuracy of parameter optimization.

[0005] The Chinese invention patent with application number CN202410754002.3 proposes a deep cascade feedback control method and device for AUV based on genetic algorithm, which includes: initializing chromosome population, current number of iterations, maximum number of iterations and global best chromosome; if the fitness value of the current best individual is less than the fitness value of the global best chromosome, updating the global best chromosome to the current best individual; decoding the final global best chromosome to obtain the corresponding control parameters, and the cascade feedback controller determines the control information for controlling the AUV based on the control parameters and the real-time status data of the AUV. This method can adjust the controller parameters in real time according to different sea conditions to improve the control effect.

[0006] The Chinese invention patent with application number CN202410752394.X proposes a first-order inverted pendulum fuzzy PID control method and system based on genetic algorithm optimization, including: establishing a system model of a first-order linear inverted pendulum system. Based on the system model, a genetic algorithm is used to optimize the fuzzy control rules and membership functions to generate the optimal fuzzy controller output parameters. The fuzzy controller output parameters are used to adjust the P, I, and D parameters of the PID controller in real time. Through the adjusted PID controller, the control signal is calculated and applied to achieve stable control of the system. The advantages of the present invention are: not only the nonlinear processing capability and robustness of the control system are improved, but also the parameter adjustment process is simplified, and it performs well in terms of response speed, stability and anti-disturbance capability, significantly reducing the steady-state error, and is suitable for the control of complex dynamic systems.

[0007] However, the above existing methods have the following shortcomings:

[0008] 1. Most existing technologies use random initialization methods, which may cause particles to be concentrated in certain areas of the search space, lack global diversity, and easily fall into local optimal solutions.

[0009] 2. Fixed control parameters cannot adapt to the search requirements at different stages. The scope of exploration cannot be expanded in the early stage of the search, and the convergence may be too slow in the later stage of the search, which reduces the overall efficiency of the algorithm.

[0010] 3. Traditional genetic algorithms or particle swarm algorithms lack dynamic perturbations during the iteration process, which easily leads to the search being too dependent on the current optimal solution and unable to effectively escape from the local optimum.

[0011] 4. The existing speed update mechanism often adopts linear rules and cannot make adaptive adjustments according to the dynamic search process of the population, which easily leads to slow convergence or premature convergence.

[0012] 5. The existing crossover and mutation operations are usually relatively simple and cannot fully guarantee the diversity of the population, resulting in individual convergence in the population and falling into local optimality during the optimization process.

[0013] 6. The existing technology lacks an effective balance between local fine search and global search, and it is often difficult to take into account both the accuracy of the solution and the ability to escape the local optimum, resulting in low solution quality. Summary of the invention

[0014] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a method for optimizing the parameters of a grid-connected inverter based on an improved genetic particle swarm algorithm. The method optimizes three key parameters in the control system based on a genetic particle swarm optimization algorithm based on the chaotic radioactivity principle, and can overcome the various deficiencies in the prior art on the basis of the traditional genetic particle swarm optimization algorithm.

[0015] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0016] The present invention provides a method for optimizing parameters of a grid-connected inverter based on an improved genetic particle swarm algorithm, the key of which is that it comprises the following steps:

[0017] Step 1: construct an objective function of the grid-connected inverter system to be optimized;

[0018] Step 2: Use the chaotic mapping Logistic mapping algorithm to initialize the population position and set the hyperparameters of the algorithm;

[0019] Step 3: Calculate the fitness value of each particle at the initial moment through the objective function, record the optimal solution, and save the initial global optimal particle and individual optimal particle information;

[0020] Step 4: Add random perturbations to the particle's velocity and position based on the principle of radioactive decay;

[0021] Step 5: Update the velocity and position of the particle;

[0022] Step 6: Perform multi-point crossover and mutation operations based on chaotic sequences;

[0023] Step 7: Update the particle fitness value, and use a Laplace distribution-based fine search method to perform a local area search and a Cauchy distribution-based search method to perform a global neighborhood search according to the current optimal solution, and update the individual optimal position and the global optimal position;

[0024] Step 8: Check the convergence conditions. If the maximum number of iterations is met or the convergence index is less than the preset threshold, terminate the iteration, output the optimal solution, and obtain the optimal parameters of the grid-connected inverter system. Otherwise, return to step 4 for an iterative cycle.

[0025] Furthermore, in step 1, the grid-connected inverter system is modeled and the specific steps of constructing the objective function are as follows:

[0026] Step 1.1, modeling the grid-connected inverter system to be optimized to obtain a linearized impedance model;

[0027] Step 1.2: Establish a power grid model in the dq coordinate system;

[0028] Step 1.3: Based on the established impedance model and grid model, a grid-connected model of the grid-connected inverter system is constructed, and an objective function of the grid-connected inverter system to be optimized is obtained;

[0029] The objective function is expressed as: F(ω cc ,ω dvc ,ω pll ),

[0030] Among them, ω cc is the inner loop control bandwidth; ω dvc is the outer loop control bandwidth; ω pll is the phase-locked loop control bandwidth.

[0031] Furthermore, the formula for initializing the population position using the chaotic mapping Logistic mapping algorithm in step 2 is:

[0032] x n+1 =μ x ·x n ·(1-x n )

[0033] Among them, x n is the position of the particle in the nth chaotic sequence; is the chaos control parameter; φ0 is the initial phase; int(t) is the current iteration number; T max is the maximum number of iterations.

[0034] Furthermore, in the process of initializing the population position using the chaotic mapping Logistic mapping algorithm in step 2, the formula for initializing each parameter ω to be optimized is:

[0035]

[0036] in, is the initial value of the i-th group of parameters to be optimized; min is the minimum value of the parameter to be optimized; max is the maximum value of the parameter ω to be optimized; x0 is the initial value generated by the Logistic mapping; is the initial speed of the parameter ω to be optimized; is the maximum speed of the parameter ω to be optimized.

[0037] Furthermore, the calculation formula for calculating the fitness value of each particle at the initial moment through the objective function in step 3 is:

[0038]

[0039] Among them, f i is the fitness value of the i-th particle; is the inner loop control bandwidth parameter value corresponding to the i-th particle; is the outer loop control bandwidth parameter value corresponding to the i-th particle; is the phase-locked loop control bandwidth corresponding to the i-th particle; φ() is the activation function; F() is the objective function; K is the number of parameters to be optimized; ψ k is the weight coefficient, satisfying represents the value of the kth parameter to be optimized corresponding to the i-th particle; is the target value of the kth parameter to be optimized.

[0040] Furthermore, in step 4, the expression for adding random perturbations to the particle velocity is:

[0041]

[0042] in, is the velocity of the i-th particle at the t-th iteration; ← is the parameter update symbol; δ is the perturbation factor of the particle; is the disturbance term, and its calculation formula is is the first random variable that follows a standard normal distribution; is the velocity disturbance intensity, is the initial velocity disturbance intensity; cre is the velocity disturbance attenuation exponent; r t is the radioactive decay coefficient of the current iteration; int(t) is the number of current iterations; T max is the maximum number of iterations;

[0043] The expression for adding random perturbations to the particle's position is:

[0044]

[0045] in, is the position of the i-th particle at the t-th iteration; is the perturbation term of the particle’s position; is the second random variable that follows the standard normal distribution; is the position disturbance intensity; is the initial position disturbance intensity; σ is the position disturbance attenuation exponent.

[0046] Furthermore, when the diversity of the particle population is less than the diversity threshold, the velocity perturbation intensity Adjust according to the following formula:

[0047]

[0048] in, is the disturbance gain coefficient, D threshold To adjust the diversity threshold of disturbance intensity, D t For the diversity of populations.

[0049] Furthermore, in step 5, the formula for updating the particle velocity is:

[0050]

[0051] in, is the particle velocity at the t+1th iteration; is the particle velocity at the tth iteration; is the nonlinear acceleration factor; α max is the maximum nonlinear acceleration factor value; α min is the minimum nonlinear acceleration factor value; β is the nonlinear acceleration factor index; int(t) represents the current iteration number; T max is the maximum number of iterations; c1 is the first speed learning factor; c2 is the second speed learning factor; and All are random numbers in the interval [0, 1]; is the optimal position of the individual; G best is the global optimal position; is the particle swarm position of the tth iteration; The velocity disturbance term of the particle added in step 4;

[0052] The update formula of particle position is:

[0053]

[0054] in, is the particle position at the t+1th iteration; is the particle position at the tth iteration; is the particle velocity at the t+1th iteration; The position disturbance term of the particle added in step 4;

[0055] In the process of updating the particle position, the mirror reflection method is used for boundary processing, as shown below:

[0056]

[0057] Among them, ωmin is the minimum value of the parameter to be optimized; max is the maximum value of the parameter ω to be optimized.

[0058] Furthermore, the specific steps of performing the multi-point crossover operation based on the chaotic sequence in step 6 are as follows:

[0059] Step A1: update the chaotic sequence. The update method is expressed as:

[0060] θ n+1 =μ θ ·θ n ·(1-θ n )

[0061] Among them, θ n+1 is the n+1th chaotic sequence; θ n is the nth chaotic sequence; μ θ is the multi-point cross-weight term;

[0062] Step A2: perform a multi-point crossover operation. The expression of the multi-point crossover operation is:

[0063]

[0064] in, is the position of the ith particle after the multi-point crossover operation, for The kth parameter index of X i is the position of the ith particle, X j is the position of the jth particle, (X i ) k For X i The kth parameter index, (X j ) k For X j The kth parameter index of ; is the mixing factor of the crossover operation; is a random number in the interval [0, 1];

[0065] The specific steps of performing the mutation operation based on the chaotic sequence in step 6 are as follows:

[0066] Step B1: According to the number of iterations, the variation function is used to perform non-uniform variation on the parameters of the particles. The expression of the variation function is:

[0067]

[0068] Among them, Δ(int(t), y) is the variation function; r is a random number in the interval [0, 1], γ Δ is the variation index; y represents the input of the variation function; is the adaptive scaling factor, σ κ is the scaling amplitude; T κ is the scaling period; φ κ is the scaling phase; int(t) is the current iteration number; T max is the maximum number of iterations;

[0069] Step B2: Perform mutation operation based on the mutation function, the expression is:

[0070]

[0071] in, is the position of the ith particle after the mutation operation, for The kth parameter index of The kth parameter index of r k is a random number in the interval [0, 1]; Δ() is the variation function, ω min is the minimum value of the parameter to be optimized; max is the maximum value of the parameter ω to be optimized.

[0072] Furthermore, in step 7, the formula for local area search using a Laplace distribution-based fine search method is:

[0073] X local =G best +σ local ·Laplace(0,1)+sgn(u)·ln(1-2|u|)

[0074] Among them, X local is the particle position after local neighborhood search; G best is the global optimal position; σ local is the local search range; Laplace(0,1) represents the Laplace distribution in the interval from 0 to 1; u is a uniform random number in the interval [-0.5, 0.5]; sgn() is the sign function;

[0075] The formula for global neighborhood search using the search method based on Cauchy distribution is:

[0076]

[0077] Among them, X global is the particle position after global neighborhood search; X min Indicates the minimum value of the global particle position, X min represents the maximum value of the global particle position; Cauchy(0,1) represents the Cauchy distribution in the range of 0 to 1; v rew is a uniform random number in the interval [0, 1].

[0078] The remarkable effects of the present invention are:

[0079] 1. The present invention uses Logistic chaotic mapping to generate the initial position and velocity of particles, thereby ensuring the uniform distribution of the population in the search space, enhancing the diversity and globality of the search, thereby avoiding the concentration of the population in a certain local area in the initial stage, and improving the ability of the algorithm to jump out of the local optimal solution.

[0080] 2. The present invention dynamically adjusts the chaos control parameters according to the number of iterations, thereby enhancing the diversity of the initial population and gradually guiding the search convergence, thereby expanding the search space in the early stage of the search, avoiding the algorithm from converging to the local optimum too early, and accelerating the convergence to the global optimum in the later stage.

[0081] 3. The present invention perturbs the velocity and position of the particle based on the principle of radioactive decay, and the perturbation amplitude decreases with the number of iterations, thereby increasing the exploration capability in the early stage of iteration and making fine adjustments in the later stage of iteration, which helps to improve the accuracy of the global optimal solution and further avoid falling into the local optimum.

[0082] 4. The present invention adjusts the disturbance intensity of particle velocity through a fractional-order attenuation strategy, thereby improving the adaptability of the algorithm, thereby adopting appropriate disturbance intensity at different stages, enhancing the flexibility of the algorithm at different search stages, and further accelerating the convergence speed.

[0083] 5. The present invention adopts a nonlinear acceleration factor to replace the traditional linear speed update rule, so that the algorithm can more flexibly combine individual optimal and global optimal information, thereby improving the search efficiency of the population and enhancing the convergence speed of the algorithm, especially when searching for a solution close to the global optimal solution, it can converge more quickly and stably.

[0084] 6. The present invention adopts multi-point crossover and non-uniform mutation operations based on chaotic sequences to improve the diversity of the population during the optimization process, thereby avoiding the population from falling into a local optimal solution too early and ensuring the exploration of the global optimal solution.

[0085] 7. The present invention combines Laplace distribution and Cauchy distribution to perform local and global neighborhood searches respectively, thereby improving local accuracy and global search capabilities, thereby achieving simultaneous local fine search and global large-scale search, improving optimization accuracy and increasing the probability of jumping out of the local optimal solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 is a flow chart of the method described in an embodiment of the present invention;

[0087] Figure 2 It is a model schematic diagram of a grid-connected inverter system;

[0088] Figure 3is a principle block diagram of the system according to an embodiment of the present invention;

[0089] Figure 4 It is a principle block diagram of the device described in the embodiment of the present invention. DETAILED DESCRIPTION

[0090] The specific implementation manner and working principle of the present invention are further described in detail below with reference to the accompanying drawings.

[0091] Embodiment 1:

[0092] like Figure 1 As shown, the embodiment of the present invention provides a method for optimizing parameters of a grid-connected inverter based on an improved genetic particle swarm algorithm, and the specific steps are as follows:

[0093] Step 1: Construct the objective function: Construct the objective function of the grid-connected inverter system to be optimized. The specific steps are as follows:

[0094] Step 1.1: Model the grid-connected inverter system to be optimized, such as Figure 2 As shown, a linearized impedance model is obtained, which can be expressed as:

[0095]

[0096] Y v,dd (s)=Y cc (s)-G cc (s)G dc (s)

[0097]

[0098] Among them, ω f is the reference angular frequency of the AC system, k p_cc and k i_cc The proportional and integral gains represented by L constitute the proportional-integral controller of the inner loop. f and R f represents the inductance and resistance of the inverter filter, and F cc (s) = k p_cc +k i_cc / s is the transfer function of the inner loop PI controller, which can describe the dynamic characteristics of the inner loop control, ω cc =k p_cc / L f is the artificially defined inner loop control bandwidth, ω dvc =k p_dvc v sd0 / Cv dc0 is the artificially defined outer loop control bandwidth, ω pll =v sd0 k p_pllIt is the artificially defined phase-locked loop control bandwidth.

[0099] Step 1.2: Establish a power grid model in the dq coordinate system;

[0100] The analytical expression of the power grid can be expressed as:

[0101] v g =Z g (s)i c +v s

[0102] The above expression describes the relationship between the input voltage and current of the VSC. As we all know, since the AC power grid is a symmetrical system, the power grid model can satisfy the following relationship:

[0103] Z g,dd (s) = Z g,qq (s)

[0104] Z g,dq (s)=-Z g,qd (s)

[0105] Its impedance matrix model can be expressed as:

[0106] Among them, ω b is the reference angular frequency of the power grid, expressed as 2πf b (f b =50Hz).

[0107] Step 1.3: Based on the established impedance model and grid model, a grid-connected model of the grid-connected inverter system is constructed, and an objective function of the grid-connected inverter system to be optimized is obtained;

[0108] The characteristics of the grid-connected inverter module can be expressed by the impedance Z vsc (s) represents, and the AC system can be simplified as an infinite voltage source v g , and the power grid matrix Z g (s) connection. Therefore, an impedance model can be formed, which is convenient for studying the small signal stability of the grid-connected inverter system, expressed as the following formula:

[0109]

[0110] After obtaining the small signal impedance model of the grid-connected system, the instability characteristics of the system can be further analyzed theoretically through the zero-pole criterion (i.e. the characteristic root of the closed-loop system).

[0111] In this embodiment, the expression of the objective function is: F(ω cc ,ω dvc ,ω pll ),

[0112] Among them, ω cc is the inner loop control bandwidth; ω dvc is the outer loop control bandwidth; ω pll is the phase-locked loop control bandwidth.

[0113] Step 2, algorithm initialization: Use the chaotic mapping Logistic mapping algorithm to initialize the population position and set the algorithm's hyperparameters;

[0114] In the initialization stage, the chaotic map is used to generate the initial position and velocity of the particles to ensure the uniform distribution of the population in the search space. The specific method is to use the Logistic mapping algorithm of the chaotic map for initialization, which can be expressed as:

[0115] x n+1 =μ x ·x n ·(1-x n )

[0116] Among them, x n is the position of the particle in the nth chaotic sequence; is the chaos control parameter; φ0 is the initial phase; int(t) is the current iteration number; T max is the maximum number of iterations. Preferably,

[0117] Furthermore, let the parameter to be optimized be ω∈{ω ilc ,ω dvc ,ω pll}, for each parameter, its initialization method is expressed as:

[0118]

[0119] In the formula, x0 is the initial value generated by the Logistic mapping; is the initial velocity of parameter ω; is the maximum speed of parameter ω; is the initial value of the i-th group of parameters to be optimized; ω min is the minimum value of the parameter to be optimized; ω max is the maximum value of the parameter to be optimized. Preferably,

[0120] In this case, the algorithm’s hyperparameters include:

[0121] Population size: N = 50;

[0122] Maximum number of iterations: T max =1000;

[0123] Chaos control parameter: μ x =3.99;

[0124] Radioactive decay coefficient: λ r =0.95;

[0125] Nonlinear acceleration factor exponent: β=2.5.

[0126] Step 3, initial fitness value calculation: Calculate the fitness value of each particle at the initial moment through the objective function, record the optimal solution, and save the initial global optimal particle and individual optimal particle information;

[0127] For each particle, calculate its performance in the objective function F(ω cc ,ω dvc ,ω pll ), specifically, for the i-th particle, the way to calculate its fitness value is expressed as:

[0128]

[0129] Among them, f i is the fitness value of the i-th particle; is the inner loop control bandwidth parameter value corresponding to the i-th particle; is the outer loop control bandwidth parameter value corresponding to the i-th particle; is the phase-locked loop control bandwidth corresponding to the i-th particle; φ() is the activation function; F() is the objective function; K is the number of parameters to be optimized; ψ k is the weight coefficient, satisfying represents the value of the kth parameter to be optimized corresponding to the i-th particle; is the target value of the kth parameter to be optimized.

[0130] In one embodiment, the activation function is a Sigmoid activation function based on Gaussian error. Assuming its input is z, the calculation method is expressed as:

[0131]

[0132] Among them, γ f is the slope of the activation function. Preferably,

[0133] Moreover, in this embodiment, the dynamic adjustment method of the weight coefficient is expressed as:

[0134]

[0135] Among them, γ ψ is the weight coefficient adjustment factor; is the variance of the kth parameter in the population; is the variance of the jth parameter in the population. Preferably, γψ =10.

[0136] Furthermore, the variance of the kth parameter in the population is calculated as:

[0137]

[0138] in, is the mean of the kth parameter.

[0139] Further, record the optimal solution and update the individual optimal position and the global optimal position G best , and save the current global optimal particle and individual optimal particle information, expressed as:

[0140]

[0141] in, represents the individual optimal position, that is, the optimal particle position of the current iteration; G best represents the global optimal position; is the particle swarm position of the tth iteration; f i is the fitness value of the i-th particle; is the historical optimal fitness value of the ith particle; f global _ best is the fitness value of the particle corresponding to the global optimal position.

[0142] Step 4: Randomly perturb particle speed and position: According to the principle of radioactive decay, add random perturbations to the speed and position of the particles;

[0143] According to the principle of radioactive decay, random perturbations are added to the velocity and position of the particle, and the amplitude of the perturbation decreases with the increase of the number of iterations. Specifically, the way to perturb the velocity of the particle is expressed as:

[0144]

[0145] in, is the velocity of the i-th particle at the t-th iteration; ← is the parameter update symbol; δ is the perturbation factor of the particle; is the disturbance term;

[0146] In this embodiment, the velocity disturbance term of the particle is The calculation method is expressed as:

[0147]

[0148] in, is the first random variable that follows a standard normal distribution, is the velocity disturbance intensity.

[0149] In the embodiment of the present invention, the velocity disturbance intensity It can be calculated using the fractional-order attenuation strategy, and the calculation is expressed as:

[0150]

[0151] in, is the initial velocity disturbance intensity, λ r is the radioactive decay coefficient, λ r t is the radioactive decay coefficient at the current iteration; λ cre is the velocity disturbance attenuation exponent. In this example, preferably, λ cre =1.2.

[0152] In this embodiment, the expression for adding random perturbations to the positions of particles is:

[0153]

[0154] in, is the position of the i-th particle at the t-th iteration; is the perturbation term of the particle’s position;

[0155] In this embodiment, the calculation method of the particle position disturbance term is expressed as:

[0156]

[0157] in, is the second random variable that follows the standard normal distribution; is the position disturbance intensity.

[0158] In this embodiment, the position disturbance intensity of the particle The calculation method is:

[0159]

[0160] in, is the initial position disturbance intensity. Preferably,

[0161] In the specific implementation process, this embodiment also performs speed disturbance intensity Adaptive adjustment of , that is, dynamically adjusting the disturbance intensity according to the diversity and convergence speed of the current population to prevent premature convergence. The way to adjust the disturbance intensity is expressed as:

[0162]

[0163] Among them, D t For the diversity of populations.

[0164] In the embodiment, the diversity threshold for adjusting the disturbance intensity is set to D threshold , when D t <D threshold When , the disturbance intensity is increased, which can be expressed as:

[0165]

[0166] in, is the disturbance gain coefficient. Preferably,

[0167] Step 5: Update the particle's velocity and position as follows:

[0168] When updating the particle speed in this embodiment, the particle speed is updated by using a nonlinear acceleration factor in combination with the particle's own experience, individual optimal solution and global optimal solution. The nonlinear acceleration factor is expressed as:

[0169]

[0170] in, is the nonlinear acceleration factor; α max is the maximum nonlinear acceleration factor value, α min is the minimum nonlinear acceleration factor value; int(t) represents the current iteration number. Preferably, α max =0.9,α min =0.4.

[0171] Next, the nonlinear acceleration factor is used to update the particle velocity, which can be expressed as:

[0172]

[0173] in, is the particle velocity at the t+1th iteration; is the particle speed of the tth iteration; c1 is the first speed learning factor, c2 is the second speed learning factor; and All are random numbers in the interval [0, 1]; is the optimal position of the individual; G best is the global optimal position; is the particle swarm position of the tth iteration; It is the velocity disturbance term of the particle added in step 4. In this example, preferably, c1=c2=2.05.

[0174] Next, the particle position is updated. In this embodiment, the new position of the particle is calculated according to the updated speed, and a boundary check is performed. The particle position update method is expressed as follows:

[0175]

[0176] in, is the particle position at the t+1th iteration; is the particle position at the tth iteration; is the particle velocity at the t+1th iteration; The position disturbance term of the particle added in step 4;

[0177] In this example, in order to prevent particles from crossing the boundary, the mirror reflection method is used for boundary processing, which is expressed as:

[0178]

[0179] Among them, ω min is the minimum value of the parameter to be optimized; max is the maximum value of the parameter ω to be optimized.

[0180] Step 6: Perform multi-point crossover and mutation operations based on chaotic sequences;

[0181] In this embodiment, in the multi-point crossover operation, the selected particle pair is subjected to a multi-point crossover operation to exchange some parameters and improve the diversity of the population. Specifically, the selected particle pair is (X i ,X j ), perform multi-point crossover based on chaotic sequence:

[0182] Step A1: update the chaotic sequence. The update method is expressed as:

[0183] θ n+1 =μ θ ·θ n ·(1-θ n )

[0184] Among them, θ n+1 is the n+1th chaotic sequence; θ n is the nth chaotic sequence; μ θ is a multi-point cross-weight term; preferably, μ θ =3.99.

[0185] Step A2: perform a multi-point crossover operation. The expression of the multi-point crossover operation is:

[0186]

[0187] in, is the position of the ith particle after the multi-point crossover operation, for The kth parameter index of X i is the position of the ith particle, X jis the position of the jth particle, (X i ) k For X i The kth parameter index, (X j ) k For X j The kth parameter index of ; β is the mixing factor of the crossover operation.

[0188] In this example, the calculation method of the mixing factor β of the crossover operation can be expressed as:

[0189]

[0190] in, is a random number in the interval [0, 1].

[0191] In this embodiment, the specific steps of performing the mutation operation based on the chaotic sequence are as follows:

[0192] Step B1: According to the number of iterations, the variation function is used to perform non-uniform variation on the parameters of the particles. The variation amplitude decreases with time. The expression of the variation function is:

[0193]

[0194] Among them, Δ(int(t), y) is the variation function; r is a random number in the interval [0, 1], γ Δ is the variation index; y represents the input of the variation function; κ t is the adaptive scaling factor, int(t) is the current iteration number; T max is the maximum number of iterations. In this example, preferably, γ Δ =4.

[0195] In this example, the adaptive scaling factor κ t The calculation method is expressed as:

[0196]

[0197] Among them, σ κ is the scaling amplitude; T κ is the scaling period; φ κ is the scaling phase. Preferably, σ κ =0.1, T κ =50,φ κ =0.5.

[0198] Step B2: Perform mutation operation based on the mutation function, the expression is:

[0199]

[0200] in, is the position of the ith particle after the mutation operation, for The kth parameter index of The kth parameter index of r k is a random number in the interval [0, 1]; Δ() is the variation function, ω min is the minimum value of the parameter to be optimized; max is the maximum value of the parameter ω to be optimized.

[0201] Step 7: Update the particle fitness value, and use a Laplace distribution-based fine search method to perform a local area search and a Cauchy distribution-based search method to perform a global neighborhood search according to the current optimal solution, and update the individual optimal position and the global optimal position;

[0202] This embodiment first performs a local neighborhood search, that is, a fine search is performed in a small range for the current optimal solution to improve the accuracy of the solution. In one embodiment, a fine search method based on Laplace distribution is used to perform a local neighborhood search, which is expressed as:

[0203] X local =G best +σ local ·Laplace(0,1)+sgn(u)·ln(1-2|u|)

[0204] Among them, X local is the particle position after local neighborhood search; G best is the global optimal position; σ local is the local search range; Laplace(0,1) represents the Laplace distribution in the interval from 0 to 1; u is a uniform random number in the interval [-0.5, 0.5]; sgn() is the sign function. Preferably, σ local =0.02.

[0205] Next, a global neighborhood search is performed, that is, random sampling is performed in the entire search space to prevent falling into a local optimum. In one embodiment, the search is performed in the entire search space based on Cauchy distribution to enhance the ability to jump out of the local optimum, which is expressed as:

[0206]

[0207] Among them, X global is the particle position after global neighborhood search; X min Indicates the minimum value of the global particle position, X min represents the maximum value of the global particle position; Cauchy(0,1) represents the Cauchy distribution in the range of 0 to 1; v rew is a uniform random number in the interval [0, 1].

[0208] Step 8: Check the convergence conditions. If the maximum number of iterations is met or the convergence index is less than the preset threshold, terminate the iteration, output the optimal solution, and obtain the optimal parameters of the grid-connected inverter system, that is, return the optimal parameter combination. Otherwise, return to step 4 to iterate.

[0209] Optimized parameter combination The dynamic performance of the control system is improved, meeting specific industrial application requirements and improving production efficiency and product quality.

[0210] Embodiment 2:

[0211] like Figure 2 As shown, an embodiment of the present invention provides a grid-connected inverter parameter optimization system based on an improved genetic particle swarm algorithm, which is used to implement the method described in Example 1, including:

[0212] An objective function building module, used to build an objective function of the grid-connected inverter system to be optimized;

[0213] Algorithm initialization module, used to initialize the population position using the chaotic mapping Logistic mapping algorithm and set the algorithm's hyperparameters;

[0214] The first calculation module is used to calculate the fitness value of each particle at the initial moment through the objective function, record the optimal solution, and save the initial global optimal particle and individual optimal particle information;

[0215] The random perturbation adding module is used to add random perturbations to the velocity and position of particles according to the principle of radioactive decay;

[0216] The second calculation module is used to update the speed and position of the particles;

[0217] Multi-point crossover and mutation operation module, used to perform multi-point crossover and mutation operations based on chaotic sequences;

[0218] The third calculation module is used to update the fitness value of the particle, and according to the current optimal solution, a fine search method based on Laplace distribution is used to perform a local area search, and a search method based on Cauchy distribution is used to perform a global neighborhood search, and update the individual optimal position and the global optimal position;

[0219] The judgment module is used to judge whether the convergence condition meets the maximum number of iterations or the convergence index is less than a preset threshold. If so, the optimal solution is output to obtain the optimal parameters of the grid-connected inverter system.

[0220] Embodiment 3:

[0221] like Figure 3As shown, an embodiment of the present invention provides a computer device, including a memory, a processor, a multimedia component, an I / O interface, a communication component, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method described in Example 1 when executing the computer program.

[0222] Embodiment 4:

[0223] An embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method described in the above embodiment 1 are implemented.

[0224] In summary, the present invention optimizes the three key parameters of the inner loop control bandwidth, the outer loop control bandwidth and the phase-locked loop control bandwidth in the control system based on the chaotic radioactive principle genetic particle swarm optimization algorithm. On the basis of the traditional genetic particle swarm optimization algorithm, the initial position and speed of the particles are generated by using Logistic chaotic mapping to ensure the uniform distribution of the population in the search space, thereby enhancing the diversity and globality of the search, thereby avoiding the concentration of the population in a certain local area in the initial stage and improving the ability of the algorithm to jump out of the local optimal solution; firstly, the chaos control parameters are dynamically adjusted according to the number of iterations, thereby enhancing the diversity of the initial population and gradually guiding the search to converge, thereby expanding the search space in the early stage of the search, avoiding the algorithm from converging to the local optimum too early, and accelerating the convergence to the global optimum in the later stage; then, based on the principle of radioactive decay, the speed and position of the particles are disturbed, and the disturbance amplitude decreases with the number of iterations, thereby increasing the exploration ability in the early stage of iteration, and fine adjustment in the later stage of iteration is helpful to improve the accuracy of the global optimal solution, and further Avoid falling into the local optimum; then, the perturbation intensity of the particle velocity is adjusted through the fractional-order attenuation strategy, which improves the adaptability of the algorithm, so that the appropriate perturbation intensity is adopted in different stages, which enhances the flexibility of the algorithm in different search stages and further accelerates the convergence speed; then, the nonlinear acceleration factor is used to replace the traditional linear velocity update rule, so that the algorithm can more flexibly combine the individual optimal and global optimal information, thereby improving the search efficiency of the population and enhancing the convergence speed of the algorithm, especially when searching for a solution close to the global optimal solution, it converges faster and more stably; then, multi-point crossover and non-uniform mutation operations based on chaotic sequences are used to improve the diversity of the population during the optimization process, thereby avoiding the population from falling into the local optimal solution too early and ensuring the exploration of the global optimal solution; finally, the Laplace distribution and the Cauchy distribution are combined to perform local and global neighborhood searches respectively, which improves the local accuracy and the ability of global search, thereby realizing local fine search and global large-scale search at the same time, improving the optimization accuracy and increasing the probability of jumping out of the local optimal solution.

[0225] The technical solution provided by the present invention is described in detail above. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. It should be pointed out that for ordinary technicians in this technical field, without departing from the principle of the present invention, the present invention can also be improved and modified in a number of ways, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.

Claims

1. A method for optimizing parameters of grid-connected inverter based on improved genetic particle swarm algorithm, characterized in that: The steps include: Step 1: construct an objective function of the grid-connected inverter system to be optimized; Step 2: Use the chaotic mapping Logistic mapping algorithm to initialize the population position and set the hyperparameters of the algorithm; Step 3: Calculate the fitness value of each particle at the initial moment through the objective function, record the optimal solution, and save the initial global optimal particle and individual optimal particle information; Step 4: Add random perturbations to the particle's velocity and position based on the principle of radioactive decay; Step 5: Update the velocity and position of the particle; Step 6: Perform multi-point crossover and mutation operations based on chaotic sequences; Step 7: Update the particle fitness value, and use a Laplace distribution-based fine search method to perform a local area search and a Cauchy distribution-based search method to perform a global neighborhood search according to the current optimal solution, and update the individual optimal position and the global optimal position; Step 8: Check the convergence conditions. If the maximum number of iterations is met or the convergence index is less than the preset threshold, terminate the iteration, output the optimal solution, and obtain the optimal parameters of the grid-connected inverter system. Otherwise, return to step 4 for an iterative cycle.

2. The method for optimizing parameters of grid-connected inverter based on improved genetic particle swarm algorithm according to claim 1, characterized in that: In step 1, the grid-connected inverter system is modeled, and the specific steps of constructing the objective function are as follows: Step 1.1, modeling the grid-connected inverter system to be optimized to obtain a linearized impedance model; Step 1.2, establish a power grid model in the dq coordinate system; Step 1.3: Based on the established impedance model and grid model, a grid-connected model of the grid-connected inverter system is constructed, and an objective function of the grid-connected inverter system to be optimized is obtained; The objective function is expressed as: F(ω cc ,ω dvc ,ω pll ), Among them, ω cc is the inner loop control bandwidth; ω dvc is the outer loop control bandwidth; ω pll is the phase-locked loop control bandwidth.

3. The method for optimizing parameters of grid-connected inverter based on improved genetic particle swarm algorithm according to claim 1, characterized in that: In step 2, the formula for initializing the population position using the chaotic mapping Logistic mapping algorithm is: x n+1 =μ x ·x n ·(1-x n ) Among them, x n is the position of the particle in the nth chaotic sequence; is the chaos control parameter; φ0 is the initial phase; int(t) is the current iteration number; T max is the maximum number of iterations.

4. The method for optimizing parameters of grid-connected inverters based on improved genetic particle swarm algorithm according to claim 3 is characterized in that: In step 2, when the chaotic mapping Logistic mapping algorithm is used to initialize the population position, the formula for initializing each parameter ω to be optimized is: in, is the initial value of the i-th group of parameters to be optimized; min is the minimum value of the parameter to be optimized; max is the maximum value of the parameter ω to be optimized; x0 is the initial value generated by the Logistic mapping; is the initial speed of the parameter ω to be optimized; is the maximum speed of the parameter ω to be optimized.

5. The method for optimizing parameters of grid-connected inverter based on improved genetic particle swarm algorithm according to claim 1, characterized in that: In step 3, the calculation formula for calculating the fitness value of each particle at the initial moment through the objective function is: Among them, f i is the fitness value of the i-th particle; is the inner loop control bandwidth parameter value corresponding to the i-th particle; is the outer loop control bandwidth parameter value corresponding to the i-th particle; is the phase-locked loop control bandwidth corresponding to the i-th particle; φ() is the activation function; F() is the objective function; K is the number of parameters to be optimized; ψ k is the weight coefficient, satisfying represents the value of the kth parameter to be optimized corresponding to the i-th particle; is the target value of the kth parameter to be optimized.

6. The method for optimizing parameters of grid-connected inverter based on improved genetic particle swarm algorithm according to claim 1, characterized in that: In step 4, the expression for adding random perturbations to the particle's velocity is: in, is the velocity of the i-th particle at the t-th iteration; ← is the parameter update symbol; δ is the perturbation factor of the particle; is the disturbance term, and its calculation formula is is the first random variable that follows a standard normal distribution; is the velocity disturbance intensity, is the initial velocity disturbance intensity; cre is the velocity disturbance attenuation exponent; r t is the radioactive decay coefficient of the current iteration; int(t) is the number of current iterations; T max is the maximum number of iterations; The expression for adding random perturbations to the particle's position is: in, is the position of the i-th particle at the t-th iteration; is the perturbation term of the particle’s position; is the second random variable that follows the standard normal distribution; is the position disturbance intensity; is the initial position disturbance intensity; σ is the position disturbance attenuation exponent.

7. The method for optimizing parameters of grid-connected inverters based on improved genetic particle swarm algorithm according to claim 6 is characterized in that: When the diversity of the particle population is less than the diversity threshold, the velocity perturbation intensity Adjust according to the following formula: in, is the disturbance gain coefficient, D threshold To adjust the diversity threshold of disturbance intensity, D t For the diversity of populations.

8. The method for optimizing parameters of grid-connected inverters based on improved genetic particle swarm algorithm according to claim 1, characterized in that: In step 5, the formula for updating the particle velocity is: in, is the particle velocity at the t+1th iteration; is the particle velocity at the tth iteration; is the nonlinear acceleration factor; α max is the maximum nonlinear acceleration factor value; α min is the minimum nonlinear acceleration factor value; β is the nonlinear acceleration factor index; int(t) represents the current iteration number; T max is the maximum number of iterations; c1 is the first speed learning factor; c2 is the second speed learning factor; and All are random numbers in the interval [0, 1]; is the optimal position of the individual; G best is the global optimal position; is the particle swarm position of the tth iteration; The velocity disturbance term of the particle added in step 4; The update formula of particle position is: in, is the particle position at the t+1th iteration; is the particle position at the tth iteration; is the particle velocity at the t+1th iteration; The position disturbance term of the particle added in step 4; In the process of updating the particle position, the mirror reflection method is used for boundary processing, as shown below: Among them, ω min is the minimum value of the parameter to be optimized; max is the maximum value of the parameter ω to be optimized.

9. The method for optimizing parameters of grid-connected inverters based on improved genetic particle swarm algorithm according to claim 1, characterized in that: In step 6, the specific steps of performing the multi-point crossover operation based on the chaotic sequence are as follows: Step A1: update the chaotic sequence. The update method is expressed as: i n+1 =μ θ ·i n ·(1-θ n ) Among them, θ n+1 is the n+1th chaotic sequence; θ n is the nth chaotic sequence; μ θ is the multi-point cross-weight term; Step A2: perform a multi-point crossover operation. The expression of the multi-point crossover operation is: in, is the position of the ith particle after the multi-point crossover operation, for The kth parameter index of X i is the position of the ith particle, X j is the position of the jth particle, (X i ) k For X i The kth parameter index, (X j ) k For X j The kth parameter index of ; is the mixing factor of the crossover operation; is a random number in the interval [0, 1]; The specific steps of performing mutation operation based on chaotic sequence are as follows: Step B1: According to the number of iterations, the variation function is used to perform non-uniform variation on the parameters of the particles. The expression of the variation function is: Among them, Δ(int(t), y) is the variation function; r is a random number in the interval [0, 1], γ Δ is the variation index; y represents the input of the variation function; is the adaptive scaling factor, σ κ is the scaling amplitude; T κ is the scaling period; φ κ is the scaling phase; int(t) is the current iteration number; T max is the maximum number of iterations; Step B2: Perform mutation operation based on the mutation function, the expression is: in, is the position of the ith particle after the mutation operation, for The kth parameter index of The kth parameter index of r k is a random number in the interval [0, 1]; Δ() is the variation function, ω min is the minimum value of the parameter to be optimized; max is the maximum value of the parameter ω to be optimized.

10. The method for optimizing parameters of grid-connected inverter based on improved genetic particle swarm algorithm according to claim 1, characterized in that: The formula for local area search using the Laplace distribution-based fine search method in step 7 is: X local (G best +σ local ·Laplace(0.1)+sgn(u)·ln(1-2|u|) Among them, X local is the particle position after local neighborhood search; G best is the global optimal position; σ local is the local search range; Laplace(0,1) represents the Laplace distribution in the interval from 0 to 1; u is a uniform random number in the interval [-0.5, 0.5]; sgn() is the sign function; The formula for global neighborhood search using the search method based on Cauchy distribution is: Among them, X global is the particle position after global neighborhood search; X min Indicates the minimum value of the global particle position, X min represents the maximum value of the global particle position; Cauchy(0,1) represents the Cauchy distribution in the range of 0 to 1; v rew is a uniform random number in the interval [0, 1].

Images