Maximum power point tracking method and system based on genetic algorithm

Abstract

The invention discloses a maximum power point tracking method and system based on a genetic algorithm, and relates to the technical field of photovoltaic power generation. Firstly, data are collected through a sensor and screened to generate a data set, then a network model comprising an input layer, a membership function layer and the like is established, and training is completed; and finally, optimizing the voltage value in a feasible solution space of the model by using a genetic algorithm. According to the method, data set quality is guaranteed through data screening, and a good foundation is laid for subsequent modeling and optimization; the established power tracking model can better fit the working characteristics of the photovoltaic module, and the accuracy of the model is improved; the application of the genetic algorithm can achieve efficient optimization in a large range, is expected to solve the contradiction between the tracking speed and precision in a traditional method, reduces the calculation complexity and the dependence on a sensor, and improves the power generation efficiency of a photovoltaic power generation system.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power generation, and more particularly to a maximum power point tracking method and system based on a genetic algorithm. Background Art

[0002] With the growing global demand for clean energy, photovoltaic power generation (PV) has gained widespread adoption as a key renewable energy source. In PV systems, the output power of PV modules is significantly affected by environmental factors such as light intensity and temperature, and their output power-voltage (PV) curve exhibits nonlinear characteristics. To improve the efficiency of PV systems, maximum power point tracking (MPPT) technology is required to ensure that PV modules consistently operate near their maximum power point, achieving maximum power output.

[0003] Currently, the commonly used maximum power point tracking methods include the perturbation observation method and the conductance increment method. The perturbation observation method determines the direction of the maximum power point by periodically changing the operating voltage of the photovoltaic module and observing the power change. This method is simple in principle and easy to implement, but there is a contradiction between tracking speed and tracking accuracy. It is easy to produce large power losses when the light intensity changes rapidly, and it will oscillate near the maximum power point in steady state, resulting in energy loss. The conductance increment method adjusts the operating voltage according to the relationship between the slope of the power-voltage curve of the photovoltaic module and zero. Its tracking accuracy is relatively high, but the computational complexity is large, and the sensor accuracy requirements are high. In addition, it is prone to misjudgment under certain working conditions, resulting in tracking failure. Therefore, how to provide a method that can take into account both tracking speed and accuracy while reducing computational complexity and dependence on sensors is a problem that needs to be solved urgently by those skilled in the art. Summary of the Invention

[0004] In view of this, the present invention provides a maximum power point tracking method and system based on a genetic algorithm to solve the problems existing in the background technology.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A maximum power point tracking method based on genetic algorithm, comprising:

[0007] S1. Use sensors to collect the output voltage, output current, ambient temperature, and light intensity of the photovoltaic module. The output voltage, output current, ambient temperature, and light intensity collected at each time point are used as a set of data. The data is filtered as a sample, and groups with obvious data anomalies are excluded to generate a photovoltaic module dataset.

[0008] S2. Establish a power tracking model based on the PV module dataset and the working characteristics of the PV modules;

[0009] S3. In the feasible solution space of the power tracking model, the voltage value is optimized by using a genetic algorithm to obtain the optimal voltage value and its corresponding maximum power point.

[0010] Optionally, the establishing of the power tracking model is specifically:

[0011] S21: Establish a network model, including the input layer, membership function layer, norm layer, normalization layer and output layer;

[0012] S22: Initialize parameters and input the first set of data to generate a fuzzy rule;

[0013] S23: Input the second set of data, start training the network, generate new fuzzy rules and adjust the width and weight of neurons;

[0014] S24: Repeat the steps of S23 until all samples enter the network training;

[0015] S25: Dynamically adjust the model to complete training.

[0016] Optionally, the neuron nodes in the membership function layer are divided into multiple groups, each node represents a membership function, each neuron node is connected to only one input variable, the input connection weight is the center and width of the membership function, and the membership function is represented by a Gaussian function:

[0017]

[0018] Where, is x i The width of the j-th Gaussian membership function, μ ij is x i The j-th membership function, c ij is x i is the center of the j-th Gaussian membership function, r is the number of input variables, and u is the total number of rules in the system.

[0019] Optionally, the IF part of the fuzzy rules of each node in the norm layer is used to match the conditions of the fuzzy rules, reflecting the total number of fuzzy rules of the entire system. The output is the activation degree of each rule. The jth rule R j The output is:

[0020]

[0021] Where X=(x1,x2,…x r)∈Rr is the center of the jth RBF unit, and each node of this layer represents an RBF unit, that is, the number of fuzzy rules.

[0022] Optionally, the method further includes the step of detecting the power tracking model, inputting multiple sets of known parameters into the power tracking model for detection, and determining whether the error between the simulation results output by the power tracking model and the collected known data is within a predetermined error range. If so, the training is completed; if not, the parameters are adjusted and retrained.

[0023] Optionally, the S3 is specifically:

[0024] Determining the output power corresponding to individuals in a subpopulation of the initial population using a power tracking model; wherein each individual is composed of a set of parameter combinations of operating point parameters to be optimized for the photovoltaic module;

[0025] Converting the initial power variables corresponding to the output powers into initial gene variables;

[0026] The subpopulation is evolved using the initial gene variables to obtain excellent working point individuals corresponding to the subpopulation and excellent gene variables corresponding to the excellent working point individuals;

[0027] According to the preset fusion and reproduction rounds, the excellent gene variables are used to fuse and reproduce the excellent working point individuals to obtain the optimal working point parameter combination to track the maximum power point.

[0028] Optionally, converting the initial power variables corresponding to the output powers into initial gene variables is specifically as follows:

[0029] Inputting the initial power variable into a first conversion queue;

[0030] Using a first conversion thread, converting the initial power variables in the first conversion queue into initial gene variables according to a first mapping relationship between the initial power variables and the initial gene variables;

[0031] The initial gene variables are continuous or discrete codes, corresponding to the feasible range of the photovoltaic module operating point parameters.

[0032] Optionally, the method of performing fusion and reproduction on the excellent working point individuals using the excellent gene variables according to a preset fusion and reproduction round to obtain an optimal working point parameter combination includes:

[0033] According to the fusion and reproduction rounds, the excellent genetic variables of each subpopulation are subjected to cross-population simulation binary crossover and Gaussian mutation to generate multiple new individuals and their corresponding new genetic variables;

[0034] Inputting the new gene variable into a second conversion queue, and converting it into a new power variable using a second conversion thread;

[0035] The power tracking model is used to determine the output power of each newborn individual, and the newborn individual with the highest power is selected as the current optimal operating point parameter combination to track the maximum power point.

[0036] A maximum power point tracking system based on genetic algorithm, comprising:

[0037] The data acquisition module uses sensors to collect the output voltage, output current, ambient temperature, and light intensity of the photovoltaic modules. The output voltage, output current, ambient temperature, and light intensity collected at each time point are taken as a set of data. The data is filtered as samples, and groups with obvious data anomalies are excluded to generate a photovoltaic module data set.

[0038] The model building module builds a power tracking model based on the PV module dataset and the working characteristics of the PV modules;

[0039] The optimization module uses a genetic algorithm to optimize the voltage value within the feasible solution space of the power tracking model to obtain the optimal voltage value and its corresponding maximum power point.

[0040] The above technical solution demonstrates that, compared to existing technologies, the present invention provides a maximum power point tracking method and system based on a genetic algorithm. This method first collects data through sensors and filters it to generate a data set. A network model, including an input layer and a membership function layer, is then established and trained. Finally, a genetic algorithm is used to optimize the voltage value within the model's feasible solution space. Data screening in the present invention ensures data set quality, laying a solid foundation for subsequent modeling and optimization. The established power tracking model can better fit the operating characteristics of photovoltaic modules, improving model accuracy. The application of a genetic algorithm allows for efficient optimization over a wide range, potentially resolving the conflict between tracking speed and accuracy in traditional methods, reducing computational complexity and sensor dependence, and improving the power generation efficiency of photovoltaic power generation systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0042] Figure 1 The present invention provides a flow chart of the method. DETAILED DESCRIPTION

[0043] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0044] The embodiment of the present invention discloses a maximum power point tracking method based on genetic algorithm, such as Figure 1 Shown, including:

[0045] S1. Use sensors to collect the output voltage, output current, ambient temperature, and light intensity of the photovoltaic module. The output voltage, output current, ambient temperature, and light intensity collected at each time point are used as a set of data. The data is filtered as a sample, and groups with obvious data anomalies are excluded to generate a photovoltaic module dataset.

[0046] S2. Establish a power tracking model based on the PV module dataset and the working characteristics of the PV modules;

[0047] S3. In the feasible solution space of the power tracking model, the voltage value is optimized by using a genetic algorithm to obtain the optimal voltage value and its corresponding maximum power point.

[0048] In a specific embodiment, establishing a power tracking model is specifically as follows:

[0049] S21: Establish a network model, including the input layer, membership function layer, norm layer, normalization layer and output layer;

[0050] The first layer is the input layer, where X1, X2, ..., X r It is the input variable of the power tracking model, specifically the output voltage and output current of the photovoltaic module, as well as the ambient temperature and light intensity.

[0051] The second layer is the membership function layer. Neuron nodes are divided into multiple groups. Each node represents a membership function. Each neuron node is connected to only one input variable. The input connection weight is the center and width of the membership function. The membership function is represented by a Gaussian function:

[0052]

[0053] Where, is x i The width of the j-th Gaussian membership function, μ ij is x i The j-th membership function, c ij is x iis the center of the j-th Gaussian membership function, r is the number of input variables, and u is the total number of rules in the system.

[0054] Among them, the Gaussian function has good smooth local characteristics, which can gather the input data in a small range and make the output data monotonically decrease on both sides with the symmetry axis as the center. Therefore, the Gaussian function is selected as the membership function to divide the entire input space.

[0055] The third layer is the norm layer. The IF part of each node fuzzy rule is used to match the conditions of the fuzzy rule, reflecting the total number of fuzzy rules in the entire system. The output is the activation degree of each rule. The jth rule R j The output is:

[0056]

[0057] Where X=(x1,x2,…x r )∈R r It is the center of the j-th RBF unit. Each node of this layer represents an RBF unit, which represents the number of fuzzy rules.

[0058] The fourth layer is the normalization layer, which normalizes the output of the previous layer. The number of nodes in this layer is equal to the number of fuzzy rule nodes in the third layer. The number of nodes N in the jth layer is j The output is:

[0059]

[0060] The fifth layer is the output layer. Each neuron node represents an output variable, which outputs the latitude and longitude of the coordinate point. The output value is the superposition of all input signals:

[0061]

[0062] Where y is the output of the variable, ω k is the connection weight of the k-th rule.

[0063] S22: Initializing parameters, including: maximum length of input space, minimum length of input space, overlap factor, threshold, attenuation constant, convergence constant, initial width of Gaussian function, predefined maximum error, predefined minimum error, and predefined constant for adjusting the center of Gaussian function; and inputting the first set of data to generate a fuzzy rule;

[0064] S23: Input the second set of data, start training the network, generate new fuzzy rules and adjust the width and weight of neurons;

[0065] S24: Repeat the steps of S23 until all samples enter the network training;

[0066] S25: Dynamically adjust the model to complete training.

[0067] It also includes the step of detecting the power tracking model, using multiple sets of known parameters to input the power tracking model for detection, and judging whether the error between the simulation results output by the power tracking model and the collected known data is within a predetermined error range. If so, the training is completed; if not, the parameters are adjusted and retrained.

[0068] In a specific embodiment, S3 is specifically:

[0069] S301, using a power tracking model to determine the output power corresponding to individuals in a subpopulation of an initial population; wherein each individual is composed of a set of parameter combinations of operating point parameters of a photovoltaic module to be optimized;

[0070] S302, converting the initial power variables corresponding to the output powers into initial gene variables;

[0071] It can be understood that, through the calculation in S301, each initial power variable corresponds to a power, and using the mapping relationship between the initial power variable and the initial gene variable, the initial power variable corresponding to each power can be converted into an initial gene variable.

[0072] S303. Using the initial gene variables, the subpopulation is evolved to obtain excellent working point individuals corresponding to the subpopulation and excellent gene variables corresponding to the excellent working point individuals;

[0073] It is understood that a multi-population genetic algorithm can be used to evolve each subpopulation within the initial population based on the initial genetic variables. Specific evolutionary methods include crossover and mutation. Typical crossover methods include, but are not limited to, simulated binary crossover; typical mutation methods include, but are not limited to, Gaussian mutation and Cauchy mutation. After completing crossover and mutation, excellent working point individuals corresponding to each subpopulation and the excellent genetic variables corresponding to these excellent working point individuals can be obtained.

[0074] S304: According to the preset fusion and reproduction rounds, the excellent gene variables are used to fuse and reproduce the excellent working point individuals to obtain the optimal working point parameter combination to track the maximum power point.

[0075] It is understandable that in the embodiment of the present application, the number of fusion and reproduction rounds will be preset according to actual needs and the computing power of the high-performance computing system. It should be noted that the more fusion and reproduction rounds there are, the better the individuals formed after evolution will be, that is, closer to the maximum power point.

[0076] Specifically, after obtaining the excellent working point individuals corresponding to each sub-population and the excellent genetic variables corresponding to the excellent working point individuals, it is necessary to further evolve, fuse and reproduce the excellent working point individuals in each sub-population to obtain the best new individuals and use them as the optimal working point parameter combination.

[0077] In a specific embodiment, the initial power variables corresponding to the output powers are converted into initial gene variables as follows:

[0078] Inputting the initial power variable into the first conversion queue;

[0079] It is understandable that in a multi-population genetic algorithm, the initial power variables need to be converted into initial gene variables before evolutionary fusion and reproduction can proceed. To this end, the initial power variables need to be input into the first conversion queue. These initial power variables correspond to the power of individuals in the subpopulations within the initial population.

[0080] Using the first conversion thread, converting the initial power variables in the first conversion queue into initial gene variables according to a first mapping relationship between the initial power variables and the initial gene variables;

[0081] Among them, the initial gene variables are continuous or discrete codes, corresponding to the feasible range of the operating point parameters of the photovoltaic module.

[0082] It is understandable that during the conversion, the high-performance computing nodes in the high-performance computing system will set up multiple first conversion threads. These first conversion threads use parallel execution to cyclically process the initial power variables in the first conversion queue in the computer background, and convert the initial power variables into initial gene variables.

[0083] It's important to note that before the conversion, the initial population must be divided into multiple subpopulations. The initial power variables are then converted to initial genetic variables for each individual in each subpopulation. Ideally, each subpopulation should contain approximately equal individuals. Furthermore, the subpopulations are divided only to facilitate thread scheduling, enabling parallel computing. Each subpopulation executes the conversion task within its assigned conversion thread. Besides serving as the basic unit for thread scheduling, the subpopulation also serves as the smallest unit for executing steps such as crossover and mutation in evolution.

[0084] It should also be noted that there is a certain mapping relationship between the initial power variable and the initial gene variable, and this application does not limit the mapping relationship itself.

[0085] In the embodiment of the present application, the initial gene variables are used to complete the evolution of the subpopulation, and the excellent working point individuals corresponding to the subpopulation and the excellent gene variables corresponding to the excellent working point individuals are obtained, including:

[0086] The initial gene variables are simulated by binary crossover and Gaussian mutation to obtain the excellent working point individuals corresponding to the subpopulation and the excellent gene variables corresponding to the excellent working point individuals.

[0087] It should be noted that population evolution methods include crossover and mutation. The specific algorithms used for crossover and mutation vary, and this application does not limit them. To illustrate the feasibility of the method of this application, in one embodiment, binary crossover and Gaussian mutation can be simulated on the initial genetic variables to obtain the excellent working point individuals corresponding to the subpopulation and the excellent genetic variables corresponding to the excellent working point individuals.

[0088] In a specific embodiment, according to a preset fusion and reproduction round, excellent genetic variables are used to fuse and reproduce excellent working point individuals to obtain an optimal working point parameter combination, including:

[0089] According to the fusion and reproduction rounds, the excellent genetic variables of each sub-population are subjected to cross-population simulation binary crossover and Gaussian mutation to generate multiple new individuals and their corresponding new genetic variables;

[0090] It should be noted that the initial population in the embodiments of this application is divided into multiple subpopulations. After completing the evolution of each subpopulation, excellent working point individuals and excellent genetic variables corresponding to each subpopulation can be obtained. At this time, it is necessary to fuse and multiply the excellent working point individuals and excellent genetic variables corresponding to each subpopulation to achieve a better evolutionary effect. During the fusion and multiplication, the evolutionary method of simulating binary crossover and Gaussian mutation can still be used to ultimately obtain multiple new individuals and their corresponding new genetic variables.

[0091] It should be noted that after each sub-population evolves, there may be multiple excellent working point individuals generated, and this application is not limited to this.

[0092] Inputting the new gene variable into the second conversion queue and converting it into a new power variable using the second conversion thread;

[0093] It is understood that after finding multiple new individuals and their corresponding new genetic variables, the new genetic variables can be input into a second conversion queue. The so-called second conversion queue refers to the queue that converts the new genetic variables into new power variables. It should also be noted that there is a certain mapping relationship between the new genetic variables and the new power variables, and this application does not limit the mapping relationship itself.

[0094] The power tracking model is used to determine the output power of each newborn individual, and the newborn individual with the highest power is selected as the current optimal operating point parameter combination to track the maximum power point.

[0095] According to the fusion and reproduction rounds, the excellent working point individuals are fused and reproduced using excellent gene variables to determine multiple optimal newborn individuals and their corresponding optimal newborn gene variables, including:

[0096] Whenever the number of sub-populations that have completed evolution reaches the preset number of fusion and reproduction, the excellent genetic variables corresponding to the excellent working point individuals in each sub-population are simulated by binary crossover and Gaussian mutation to obtain excellent new individuals and excellent new genetic variables;

[0097] It is understood that after each subpopulation completes its evolution, it generates excellent working point individuals and corresponding excellent genetic variables. Assuming there are R subpopulations, these subpopulations may complete their evolution at different times as they evolve within the high-performance computing nodes of a parallel computing system. In other words, the evolution of the R subpopulations is completed sequentially, rather than simultaneously. Thus, when the number of subpopulations that have completed evolution reaches the preset fusion and reproduction number K, binary crossover and Gaussian mutation simulations can be performed on the excellent genetic variables corresponding to the excellent working point individuals in these subpopulations to obtain new individuals and new genetic variables corresponding to these subpopulations. Then, when the number of other subpopulations that have completed their evolution reaches the preset fusion and reproduction number K, binary crossover and Gaussian mutation simulations can be performed on the excellent genetic variables corresponding to the excellent working point individuals in these other subpopulations to obtain new individuals and new genetic variables corresponding to these other subpopulations. Similarly, binary crossover and Gaussian mutation simulations can be performed on the excellent genetic variables corresponding to the excellent working point individuals in all subpopulations to obtain new individuals and new genetic variables corresponding to the entire R subpopulation. The advantage of this is that the efficiency of fusion and reproduction can be improved. It should be noted that, in the embodiment of the present application, a counter can be used to count when the number of sub-populations that have completed fusion and reproduction reaches the fusion and reproduction number K.

[0098] The excellent de novo genetic variants were entered into the third transition cohort;

[0099] using a third conversion thread to convert the newborn gene variables in the third conversion queue into newborn power variables according to a third mapping relationship between the newborn power variables and the newborn gene variables;

[0100] It is understandable that during the conversion, the high-performance computing nodes in the high-performance computing system will set up multiple third conversion threads. These third conversion threads use parallel execution to cyclically process the new gene variables in the third conversion queue in the computer background, and convert the new gene variables into new power variables.

[0101] It should be noted that, during the conversion, the newly generated gene variables corresponding to each subpopulation can utilize the third conversion thread to implement parallel computing operations and improve conversion efficiency.

[0102] It should also be noted that there is a certain mapping relationship between the new gene variables and the new power variables, and this application does not limit the mapping relationship itself.

[0103] According to the newborn power variable, the power corresponding to each newborn individual is determined using the power tracking model, and the first L newborn individuals with higher power are selected as multiple optimal newborn individuals, and the newborn gene variables corresponding to the multiple newborn individuals with higher power are used as the corresponding optimal newborn gene variables; where L is a positive integer and is less than the total number of newborn individuals.

[0104] From the above description, it can be seen that the method provided in this application can use new gene variables to fuse and reproduce new individuals according to the fusion and reproduction rounds, and determine multiple optimal new individuals and their corresponding optimal new gene variables.

[0105] A maximum power point tracking system based on genetic algorithm, comprising:

[0106] The data acquisition module uses sensors to collect the output voltage, output current, ambient temperature, and light intensity of the photovoltaic modules. The output voltage, output current, ambient temperature, and light intensity collected at each time point are taken as a set of data. The data is filtered as samples, and groups with obvious data anomalies are excluded to generate a photovoltaic module data set.

[0107] The model building module builds a power tracking model based on the PV module dataset and the working characteristics of the PV modules;

[0108] The optimization module uses a genetic algorithm to optimize the voltage value within the feasible solution space of the power tracking model to obtain the optimal voltage value and its corresponding maximum power point.

[0109] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0110] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A maximum power point tracking method based on genetic algorithm, characterized in that: include: S1. Use sensors to collect the output voltage, output current, ambient temperature, and light intensity of the photovoltaic module. The output voltage, output current, ambient temperature, and light intensity collected at each time point are used as a set of data. The data is filtered as a sample, and groups with obvious data anomalies are excluded to generate a photovoltaic module dataset. S2. Establish a power tracking model based on the PV module dataset and the working characteristics of the PV modules; S3. In the feasible solution space of the power tracking model, the voltage value is optimized by using a genetic algorithm to obtain the optimal voltage value and its corresponding maximum power point.

2. The maximum power point tracking method based on genetic algorithm according to claim 1, characterized in that: The specific steps of establishing the power tracking model are as follows: S21: Establish a network model, including the input layer, membership function layer, norm layer, normalization layer and output layer; S22: Initialize parameters and input the first set of data to generate a fuzzy rule; S23: Input the second set of data, start training the network, generate new fuzzy rules and adjust the width and weight of neurons; S24: Repeat the steps of S23 until all samples enter the network training; S25: Dynamically adjust the model to complete training.

3. The maximum power point tracking method based on genetic algorithm according to claim 2, characterized in that: The neuron nodes in the membership function layer are divided into multiple groups, each node represents a membership function, each neuron node is connected to only one input variable, and the input connection weight is the center and width of the membership function. The membership function is represented by a Gaussian function: Where, is x i The width of the j-th Gaussian membership function, μ ij is x i The j-th membership function, c ij is x i is the center of the j-th Gaussian membership function, r is the number of input variables, and u is the total number of rules in the system.

4. The maximum power point tracking method based on genetic algorithm according to claim 3, characterized in that: The IF part of the fuzzy rules of each node in the norm layer is used to match the conditions of the fuzzy rules, reflecting the total number of fuzzy rules in the entire system. The output is the activation degree of each rule. The jth rule R j The output is: Where X=(x1,x2,…x r )∈Rr is the center of the jth RBF unit, and each node of this layer represents an RBF unit, that is, the number of fuzzy rules.

5. The maximum power point tracking method based on genetic algorithm according to claim 2, characterized in that: It also includes the step of detecting the power tracking model, using multiple sets of known parameters to input the power tracking model for detection, and judging whether the error between the simulation results output by the power tracking model and the collected known data is within a predetermined error range. If so, the training is completed; if not, the parameters are adjusted and retrained.

6. The maximum power point tracking method based on genetic algorithm according to claim 1, characterized in that: The S3 is specifically: Determining the output power corresponding to individuals in a subpopulation of the initial population using a power tracking model; wherein each individual is composed of a set of parameter combinations of operating point parameters to be optimized for the photovoltaic module; Converting the initial power variables corresponding to the output powers into initial gene variables; The subpopulation is evolved using the initial gene variables to obtain excellent working point individuals corresponding to the subpopulation and excellent gene variables corresponding to the excellent working point individuals; According to the preset fusion and reproduction rounds, the excellent gene variables are used to fuse and reproduce the excellent working point individuals to obtain the optimal working point parameter combination to track the maximum power point.

7. The maximum power point tracking method based on genetic algorithm according to claim 6, characterized in that: The conversion of the initial power variables corresponding to the output powers into initial gene variables is specifically as follows: Inputting the initial power variable into a first conversion queue; Using a first conversion thread, converting the initial power variables in the first conversion queue into initial gene variables according to a first mapping relationship between the initial power variables and the initial gene variables; The initial gene variables are continuous or discrete codes, corresponding to the feasible range of the photovoltaic module operating point parameters.

8. The maximum power point tracking method based on genetic algorithm according to claim 6, characterized in that: According to the preset fusion and reproduction rounds, the excellent gene variables are used to fuse and reproduce the excellent working point individuals to obtain the optimal working point parameter combination, including: According to the fusion and reproduction rounds, the excellent genetic variables of each subpopulation are subjected to cross-population simulation binary crossover and Gaussian mutation to generate multiple new individuals and their corresponding new genetic variables; Inputting the new gene variable into a second conversion queue, and converting it into a new power variable using a second conversion thread; The power tracking model is used to determine the output power of each newborn individual, and the newborn individual with the highest power is selected as the current optimal operating point parameter combination to track the maximum power point.

9. A maximum power point tracking system based on genetic algorithm, characterized in that: A maximum power point tracking method based on a genetic algorithm according to any one of claims 1 to 8 is applied, comprising: The data acquisition module uses sensors to collect the output voltage, output current, ambient temperature, and light intensity of the photovoltaic modules. The output voltage, output current, ambient temperature, and light intensity collected at each time point are taken as a set of data. The data is filtered as samples, and groups with obvious data anomalies are excluded to generate a photovoltaic module data set. The model building module builds a power tracking model based on the PV module dataset and the working characteristics of the PV modules; The optimization module uses a genetic algorithm to optimize the voltage value within the feasible solution space of the power tracking model to obtain the optimal voltage value and its corresponding maximum power point.

Images