阅读说明：本技术 一种局部遮挡下光伏阵列全局最大功率点追踪方法 (Photovoltaic array global maximum power point tracking method under local shielding ) 是由 纪峰 汪友明 程琳 李永超 王清艺 雷一帆 马一飞 于 2020-11-23 设计创作，主要内容包括：本发明公开了一种局部遮挡下光伏阵列全局最大功率点追踪方法,该方法包括：通过功率-电压变化率将光伏阵列模型对应的P-U曲线划分为若干个区间的P-U曲线,根据区间的P-U曲线确定局部最大功率点的输出电压,将其输入李雅普诺夫稳定控制系统；基于李雅普诺夫第二类方法确定系统状态是否稳定；基于蛙跳-粒子群算法优化的径向基函数(RBF)神经网络确定与所述区间上输出电压对应的占空比函数；将占空比函数确定的局部最大功率点进行比较进而确定全局最大功率点；根据全局最大功率点所对应的占空比数值驱动所述光伏阵列中的MOSFET,以对所述全局最大功率点进行追踪,从而实现能快速稳定地追踪到全局最大功率点,提高光伏发电效率。(The invention discloses a photovoltaic array global maximum power point tracking method under local shielding, which comprises the following steps: dividing a P-U curve corresponding to the photovoltaic array model into a plurality of interval P-U curves through the power-voltage change rate, determining the output voltage of a local maximum power point according to the interval P-U curves, and inputting the output voltage into a Lyapunov stable control system; determining whether the system state is stable based on a Lyapunov second method; determining a duty ratio function corresponding to the output voltage in the interval based on a Radial Basis Function (RBF) neural network optimized by a frog-particle swarm optimization algorithm; comparing the local maximum power points determined by the duty ratio function to determine a global maximum power point; and driving the MOSFET in the photovoltaic array according to the duty ratio value corresponding to the global maximum power point so as to track the global maximum power point, thereby realizing fast and stable tracking of the global maximum power point and improving the photovoltaic power generation efficiency.)

1. A method for tracking a global maximum power point of a photovoltaic array under local shading is characterized by comprising the following steps:

the method comprises the following steps that firstly, a photovoltaic array model is established according to states of a bypass diode and a blocking diode in a photovoltaic array equivalent circuit under local shielding, the bypass diode is connected with each photovoltaic module in the equivalent circuit in parallel, the blocking diode is connected with the positive electrode side of each group of series photovoltaic modules in series, and the states are conduction states or blocking states;

determining the output voltage of a local maximum power point according to a power-voltage P-U curve corresponding to the photovoltaic array model;

constructing a Lyapunov function, determining a stable function of the expected value of the current of the photovoltaic system according to the time derivative of the Lyapunov function, and taking the parameters determined by the state equation as input variables of the RBF neural network;

determining a duty ratio function corresponding to the output voltage based on a Radial Basis Function (RBF) neural network optimized by a frog-particle swarm optimization algorithm;

comparing the local maximum power points determined based on the duty ratio function to determine a global maximum power point;

and step five, driving a metal-oxide semiconductor field effect transistor (MOSFET) in the photovoltaic array according to the duty ratio value corresponding to the global maximum power point so as to track the global maximum power point.

2. The method according to claim 1, wherein the determining the output voltage of the local maximum power point according to the power-voltage P-U curve corresponding to the photovoltaic array model of step one is specifically:

averagely dividing a power-voltage P-U curve into a plurality of sections;

judging and combining the sections according to the power-voltage P-U change rate to obtain the interval where each local maximum value point is located;

and determining the output voltage according to the P-U curve interval where the local maximum value point is located.

3. The method according to claim 1, wherein the step three of determining the duty cycle function corresponding to the output voltage based on the leapfrog-particle swarm optimization based RBF neural network is specifically:

determining an input variable of the RBF neural network based on a Lyapunov second-class method as a judgment condition;

determining the topological structure of the RBF neural network according to the input variable;

determining the optimal radial basis center and the optimal width of the RBF neural network based on a frog-particle swarm algorithm;

and training the RBF neural network according to the optimal radial basis center and the optimal width.

4. The method of claim 3, wherein the input variables of the RBF neural network are determined based on the Lyapunov second class method, in particular:

establishing an array equation of the photovoltaic array according to kirchhoff's law and the output voltage, wherein the array equation comprises the output voltage, the photovoltaic system current, the inductive current, the boost voltage of the photovoltaic array and the duty ratio;

constructing a Lyapunov function according to a preset first tracking error and the array equation;

determining a stabilization function of the expected value of the photovoltaic system current according to the time derivative of the Lyapunov function;

constructing a second tracking error based on the stabilization function and the actual value of the photovoltaic system current;

determining the output voltage, the photovoltaic system current, the first tracking error, and the second tracking error as the input variables.

5. The method of claim 3, wherein the optimal radial basis center and optimal width of the RBF neural network are determined based on a frog-particle swarm optimization algorithm, specifically:

determining frog individuals in a frog-leaping algorithm according to the initial radial base center and the initial width of the RBF neural network;

randomly generating a preset number of frog individuals and determining the maximum iteration times and dimensionality;

determining the fitness value of each frog individual according to a fitness function;

grouping the frog individuals according to the fitness value;

determining particles of a particle swarm algorithm according to the optimal frog individuals in each group of frog individuals;

updating the optimal particles based on the particle swarm algorithm;

outputting the optimal radial basis center and the optimal width when the maximum iteration number is reached.

Technical Field

The application relates to the technical field of photovoltaic power generation, in particular to a photovoltaic array global maximum power point tracking method under local shielding.

Background

Solar photovoltaic power generation is considered as a new energy technology with the most development prospect in the world at present, and each developed country invests huge capital competition research and development, actively promotes the industrialization process, and vigorously develops market application. The photovoltaic power generation industry has encountered a number of problems in its development: the photovoltaic module has high cost, low photoelectric conversion efficiency and partial shielding damage. MPPT (Maximum Power Point Tracking) is the most direct and effective method for reducing the Power generation cost and improving the Power generation efficiency.

In the actual use process of a photovoltaic system, the influence of the surrounding environment (sky, dark clouds, high-rise buildings and dust) often causes the uneven illumination intensity of a photovoltaic array, and the problem of local shielding is caused. Under the condition that a photovoltaic array is partially shielded, the output Power-voltage P-U characteristic curve of the photovoltaic array presents a plurality of peak characteristics, and the traditional MPPT method is difficult to find out the global MPP from a plurality of local MPPs (Maximum Power points).

Therefore, how to realize fast and stable tracking to the global maximum power point under local shielding and further improve the photovoltaic power generation efficiency is a technical problem to be solved at present.

Disclosure of Invention

The invention provides a photovoltaic array global maximum power point tracking method under local shielding, which is used for solving the technical problem that the global MPP is difficult to quickly and stably find from a plurality of local MPPs under the local shielding in the prior art, and comprises the following steps:

the method comprises the following steps that firstly, a photovoltaic array model is established according to states of a bypass diode and a blocking diode in a photovoltaic array equivalent circuit under local shielding, the bypass diode is connected with each photovoltaic module in the equivalent circuit in parallel, the blocking diode is connected with the positive electrode side of each group of series photovoltaic modules in series, and the states are conduction states or blocking states;

determining the output voltage of a local maximum power point according to a power-voltage P-U curve corresponding to the photovoltaic array model;

constructing a Lyapunov function, determining a stable function of the expected value of the current of the photovoltaic system according to the time derivative of the Lyapunov function, and taking the parameters determined by the state equation as input variables of the RBF neural network;

determining a duty ratio function corresponding to the output voltage based on a Radial Basis Function (RBF) neural network optimized by a frog-particle swarm optimization algorithm;

comparing the local maximum power points determined based on the duty ratio function to determine a global maximum power point;

and step five, driving a metal-oxide semiconductor field effect transistor (MOSFET) in the photovoltaic array according to the duty ratio value corresponding to the global maximum power point so as to track the global maximum power point.

Preferably, the determining the output voltage of the local maximum power point according to the power-voltage P-U curve corresponding to the photovoltaic array model in the step one includes:

averagely dividing a power-voltage P-U curve into a plurality of sections;

judging and combining the sections according to the power-voltage P-U change rate to obtain the interval where each local maximum value point is located;

and determining the output voltage according to the P-U curve interval where the local maximum value point is located.

Preferably, in the third step, the RBF neural network optimized based on the frog-particle swarm optimization determines a duty ratio value corresponding to the output voltage, specifically:

determining an input variable of the RBF neural network based on a Lyapunov second-class method;

determining the topological structure of the RBF neural network according to the input variable;

determining the optimal radial basis center and the optimal width of the RBF neural network based on a frog-particle swarm algorithm;

training the RBF neural network according to the optimal radial basis center and the optimal width;

and outputting the duty ratio function according to the training result.

Preferably, the input variables of the RBF neural network are determined based on a lyapunov second-class method, specifically:

establishing an array equation of the photovoltaic array according to kirchhoff's law and the output voltage, wherein the array equation comprises the output voltage, the photovoltaic system current, the inductive current, the boost voltage of the photovoltaic array and the duty ratio;

constructing a Lyapunov function according to a preset first tracking error and the array equation;

determining a stabilization function of the expected value of the photovoltaic system current according to the time derivative of the Lyapunov function;

constructing a second tracking error based on the stabilization function and the actual value of the photovoltaic system current;

determining the output voltage, the photovoltaic system current, the first tracking error, and the second tracking error as the input variables.

Preferably, the optimal radial basis center and the optimal width of the RBF neural network are determined based on a frog-particle swarm optimization, specifically:

determining frog individuals in a frog-leaping algorithm according to the initial radial base center and the initial width of the RBF neural network;

randomly generating a preset number of frog individuals and determining the maximum iteration times and dimensionality;

determining the fitness value of each frog individual according to a fitness function;

grouping the frog individuals according to the fitness value;

determining particles of a particle swarm algorithm according to the optimal frog individuals in each group of frog individuals;

updating the optimal particles based on the particle swarm algorithm;

outputting the optimal radial basis center and the optimal width when the maximum iteration number is reached.

Correspondingly, this application has still provided photovoltaic array global maximum power point tracking equipment under local sheltering from, equipment includes:

the building module is used for building a photovoltaic array model according to states of a bypass diode and a blocking diode in a photovoltaic array equivalent circuit under local shielding, wherein the bypass diode is connected with each photovoltaic module in the equivalent circuit in parallel, the blocking diode is connected with the positive electrode side of each group of series photovoltaic modules in series, and the states are a conduction state or a blocking state;

the first determining module is used for determining the output voltage of the local maximum power point according to a power-voltage P-U curve corresponding to the photovoltaic array model;

the second determining module is used for determining a duty ratio function corresponding to the output voltage based on a RBF neural network optimized by a frog-particle swarm optimization algorithm;

and the driving module is used for driving the metal-oxide semiconductor field effect transistor (MOSFET) in the photovoltaic array according to the duty ratio value so as to track the global maximum power point.

Preferably, the first determining module is specifically configured to:

averagely dividing a power-voltage P-U curve into a plurality of sections;

judging and combining the sections according to the power-voltage P-U change rate to obtain the interval where each local maximum value point is located;

and determining the output voltage according to the P-U curve interval where the local maximum value point is located.

Preferably, the second determining module is specifically configured to:

determining an input variable of the RBF neural network based on a Lyapunov second-class method;

determining the topological structure of the RBF neural network according to the input variable;

determining the optimal radial basis center and the optimal width of the RBF neural network based on a frog-particle swarm algorithm;

training the RBF neural network according to the optimal radial basis center and the optimal width;

and outputting the duty ratio function according to the training result.

Preferably, the second determining module is further specifically configured to:

establishing an array equation of the photovoltaic array according to kirchhoff's law and the output voltage, wherein the array equation comprises the output voltage, the photovoltaic system current, the boosting voltage of the photovoltaic array and the duty ratio;

constructing a Lyapunov function according to a preset first tracking error and the array equation;

determining a stabilization function of the expected value of the photovoltaic system current according to the time derivative of the Lyapunov function;

constructing a second tracking error based on the stabilization function and the actual value of the photovoltaic system current;

determining the output voltage, the photovoltaic system current, the first tracking error, and the second tracking error as the input variables.

Preferably, the second determining module is further specifically configured to:

determining frog individuals in a frog-leaping algorithm according to the initial radial base center and the initial width of the RBF neural network;

randomly generating a preset number of frog individuals and determining the maximum iteration times and dimensionality;

determining the fitness value of each frog individual according to a fitness function;

grouping the frog individuals according to the fitness value;

determining particles of a particle swarm algorithm according to the optimal frog individuals in each group of frog individuals;

updating the optimal particles based on the particle swarm algorithm;

outputting the optimal radial basis center and the optimal width when the maximum iteration number is reached.

Compared with the prior art, the invention has the following beneficial effects:

the invention discloses a photovoltaic array global maximum power point tracking method under local shielding, which comprises the following steps: establishing a photovoltaic array model according to states of a bypass diode and a blocking diode in the photovoltaic array equivalent circuit under local shielding; determining the output voltage of a local maximum power point according to a power-voltage P-U curve corresponding to the photovoltaic array model; constructing a Lyapunov function, determining a stable function of the expected value of the current of the photovoltaic system according to the time derivative of the Lyapunov function, and taking the parameters determined by the state equation as input variables of the RBF neural network; determining a duty ratio function corresponding to the output voltage based on a Radial Basis Function (RBF) neural network optimized by a frog-particle swarm optimization algorithm; comparing the local maximum power points determined by the duty ratio function to determine a global maximum power point; and driving a metal-oxide semiconductor field effect transistor (MOSFET) in the photovoltaic array according to the duty ratio value corresponding to the global maximum power point so as to track the global maximum power point. Therefore, the global maximum power point can be quickly and stably tracked, the unstable state of photovoltaic voltage and current near the maximum power point is improved, and the photovoltaic power generation efficiency is improved.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

Fig. 1 is a schematic flowchart of a method for tracking a global maximum power point of a photovoltaic array under local shielding according to an embodiment of the present invention;

FIG. 2 is a schematic diagram illustrating a method for tracking a global maximum power point of a photovoltaic array under a local occlusion according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a photovoltaic array according to an embodiment of the present invention;

FIG. 4 is a schematic illustration of a photovoltaic array in accordance with an embodiment of the present invention;

FIG. 5 is a schematic diagram of a photovoltaic array simulation model under partial occlusion in an embodiment of the invention;

FIG. 6 is a schematic sectional view of a P-U curve according to an embodiment of the present invention;

FIG. 7 is a schematic diagram of a photovoltaic power generation system in accordance with an embodiment of the present invention;

FIG. 8 is a schematic diagram of a hardware design structure of a photovoltaic power generation system according to an embodiment of the present invention;

FIG. 9 is a flow chart of an RBF neural network optimized by a frog-particle swarm optimization algorithm;

FIG. 10 is a graph illustrating the output voltage curve of a photovoltaic array combining a segmented search with a Lyapunov stabilization function according to an embodiment of the present invention;

FIG. 11 is a schematic diagram of the output voltage curve of a photovoltaic array combining the sectional search and the conventional perturbation and observation method in the embodiment of the present invention;

FIG. 12 is a graph illustrating the output power curve of a photovoltaic array combining a segmented search with a Lyapunov stabilization function according to an embodiment of the present invention;

FIG. 13 is a graph illustrating the output power curve of a photovoltaic array combining a segmented search with a conventional perturbed view method according to an embodiment of the present invention;

fig. 14 is a schematic structural diagram of a photovoltaic array global maximum power point tracking device under local shading according to an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

As described in the background, it is difficult to quickly and stably find a global MPP from a plurality of local MPPs under local occlusion in the prior art.

In order to solve the above problem, an embodiment of the present application provides a method for tracking a global maximum power point of a photovoltaic array under local shielding, in which interval segmentation search is combined with a lyapunov stabilization function, so that the global maximum power point can be quickly and stably tracked, and photovoltaic power generation efficiency is improved.

Fig. 1 shows a schematic flow chart of a method for tracking a global maximum power point of a photovoltaic array under local shading according to an embodiment of the present invention, where the method includes the following steps:

s101, a photovoltaic array model is established according to states of a bypass diode and a blocking diode in the photovoltaic array equivalent circuit under local shielding. And determining the output voltage of the local maximum power point according to the P-U curve corresponding to the photovoltaic array model.

As shown in fig. 3, the photovoltaic modules are mostly connected in series-parallel to form a photovoltaic array (i.e. M photovoltaic modules are connected in series to form a row, and N photovoltaic modules are connected in parallel), and a bypass diode D is added on the basis_bAnd a blocking diode D_pThe bypass diode mainly acts to eliminate the "hot spot effect" that occurs in the case of partial shadows. Blocking diode D_pThe function is to prevent reverse current from flowing through the photovoltaic module. Under the condition of local shielding, the current-voltage (I-U) and power-voltage (P-U) curves of the photovoltaic array are influenced by the bypass diode and the blocking diode, and the multimodal phenomenon occurs. And establishing a photovoltaic array model according to the states of a bypass diode and a blocking diode in the photovoltaic array equivalent circuit under the local shielding. Specifically, a corresponding I-U curve and a corresponding P-U curve may be drawn according to the photovoltaic array model, and as shown in fig. 6, the output voltage of the global maximum power point may be determined according to the P-U curve.

In order to accurately determine the output voltage of the global maximum power point, in a preferred embodiment of the present application, the output voltage of the local maximum power point is determined according to a power-voltage P-U curve corresponding to the photovoltaic array model, specifically:

averagely dividing a power-voltage P-U curve into a plurality of sections;

judging and combining the sections according to the power-voltage P-U change rate to obtain the interval where each local maximum value point is located;

and determining the output voltage according to the P-U curve interval where the local maximum value point is located.

Specifically, a power-voltage P-U curve is averagely divided into a plurality of sections; judging and combining the sections according to the power-voltage P-U change rate to obtain the interval where each local maximum value point is located; and determining the output voltage according to the P-U curve interval where the local maximum value point is located. In a specific application scenario of the present application, as shown in fig. 6, a plurality of P-U curve segments are judged and combined to obtain three intervals a-B-C, C-D, and D-E. Each interval has a local peak corresponding to P_m＝I_m*U_m. Therefore, one coordinate point (U) exists for each section_m,I_m) And determining the output voltage according to the P-U curve interval where the local maximum power point is located. It should be noted that the above solution of the preferred embodiment is only one specific implementation solution proposed in the present application, and other ways of determining the output voltage of the global maximum power point according to the power-voltage P-U curve all belong to the protection scope of the present application.

S102, a Lyapunov function is constructed, a stability function of the expected value of the current of the photovoltaic system is determined according to the time derivative of the Lyapunov function, and the parameters determined by the Lyapunov function are used as input variables of the RBF neural network.

Specifically, when a linear constancy system is researched, a plurality of criteria such as algebraic stability criterion and nyquist stability criterion can be used for judging the stability of the system. The Lyapunov stability theory can be simultaneously applied to the analysis of the stability of a linear system, a nonlinear system, a steady system and a time-varying system, and is a more general stability analysis method. The Lyapunov stability theory mainly refers to the second class of Lyapunov methods, also known as Lyapunov direct methods. The second class Lyapunov method can be used for systems of any order, and the stability can be directly judged by using the method without solving a system state equation. For non-linear systems and time-varying systems, the solution of the state equations is often difficult, and thus lyapunov second class of methods shows great advantages. And determining a stable function of the expected value of the current of the photovoltaic system according to the time derivative of the Lyapunov function, and taking the parameter determined by the state equation as an input variable of the RBF neural network.

S103, determining a duty ratio function corresponding to the output voltage in the interval based on a Radial Basis Function (RBF) neural network optimized by a frog-particle swarm optimization algorithm.

Specifically, the frog-leaping algorithm is a brand-new heuristic group evolution algorithm, has high-efficiency calculation performance and excellent global search capability, and the particle swarm algorithm is a random search algorithm based on group cooperation and developed by simulating foraging behavior of a bird group. A prediction model of a duty ratio D is established by adopting a RBF neural network optimized based on a frog-particle swarm optimization algorithm, the radial base center and the width of the RBF neural network are substantially used as frogs in the frog-leaping algorithm, the frogs in the frog-leaping algorithm are optimized by adopting the particle swarm optimization algorithm, parameters are updated by continuous optimization, and the learning and training of the network are realized.

In order to determine an accurate duty ratio value, in a preferred embodiment of the present application, the RBF neural network optimized based on the frog-particle swarm optimization determines a duty ratio function corresponding to the output voltage, specifically:

determining an input variable of the RBF neural network based on a Lyapunov second-class method;

determining the topological structure of the RBF neural network according to the input variable;

determining the optimal radial basis center and the optimal width of the RBF neural network based on a frog-particle swarm algorithm;

training the RBF neural network according to the optimal radial basis center and the optimal width;

and outputting the duty ratio function according to the training result.

Specifically, an input variable of the RBF neural network is determined by a Lyapunov second-class method; determining the topological structure of the RBF neural network according to the input variable; then determining the optimal radial basis center and the optimal width of the RBF neural network based on a frog-particle swarm algorithm; and finally, training the RBF neural network according to the optimal radial basis center and the optimal width and then outputting a duty ratio function.

It should be noted that the scheme of the above preferred embodiment is only a specific implementation scheme proposed by the present application, and other manners of determining the duty ratio value corresponding to the output voltage by using the RBF neural network optimized based on the frog-particle swarm optimization all belong to the protection scope of the present application.

S104, comparing the local maximum power points determined by the duty ratio function to determine a global maximum power point.

As described above, the global maximum power point is determined by comparing the local maximum power points determined by the duty cycle function.

In order to accurately determine the input variables of the RBF neural network, in a preferred embodiment of the present application, the input variables of the RBF neural network are determined based on a lyapunov second method, specifically:

establishing an array equation of the photovoltaic array according to kirchhoff's law and the output voltage, wherein the array equation comprises the output voltage, the photovoltaic system current, the inductive current, the boost voltage of the photovoltaic array and the duty ratio;

constructing a Lyapunov function according to a preset first tracking error and the array equation;

determining a stability function of the expected value of the photovoltaic system current according to the time derivative of the Lyapunov function;

constructing a second tracking error based on the stabilization function and the actual value of the photovoltaic system current;

determining the output voltage, the photovoltaic system current, the first tracking error, and the second tracking error as the input variables.

Specifically, kirchhoff's law includes kirchhoff's current law and kirchhoff's voltage law, an array equation of the photovoltaic array is established according to kirchhoff's law and the output voltage, the array equation includes the output voltage, the photovoltaic system current, the boost voltage and the duty ratio of the photovoltaic array, then a lyapunov function is constructed according to a preset first tracking error and the array equation, a stable function of a desired value of the photovoltaic system current is determined according to a time derivative of the lyapunov function, a second tracking error is constructed based on the stable function and an actual value of the photovoltaic system current, and finally the output voltage, the photovoltaic system current, the first tracking error and the second tracking error are determined as the input variables, wherein an embodiment of the above-mentioned process-specific calculation formula in a specific application scenario is described in detail, and will not be described in detail herein.

It should be noted that the solution of the above preferred embodiment is only a specific implementation solution proposed in the present application, and other ways of determining the input variable of the RBF neural network based on the lyapunov second method all belong to the protection scope of the present application.

In order to determine an accurate optimal radial basis center and an optimal width of the RBF neural network, in a preferred embodiment of the present application, the optimal radial basis center and the optimal width of the RBF neural network are determined based on a frog-particle swarm algorithm, which specifically includes:

determining frog individuals in a frog-leaping algorithm according to the initial radial base center and the initial width of the RBF neural network;

randomly generating a preset number of frog individuals and determining the maximum iteration times and dimensionality;

determining the fitness value of each frog individual according to a fitness function;

grouping the frog individuals according to the fitness value;

determining particles of a particle swarm algorithm according to the optimal frog individuals in each group of frog individuals;

updating the optimal particles based on the particle swarm algorithm;

outputting the optimal radial basis center and the optimal width when the maximum iteration number is reached.

Specifically, the method comprises the steps of initializing radial basis centers and widths, inputting the radial basis centers and the widths as frog individuals into a frog jump algorithm, randomly generating a preset number of frog individuals by initializing a frog population, determining the maximum iteration times and dimensions, calculating the fitness value of each frog individual according to a fitness function, grouping the frog individuals based on the fitness values, determining particles of a particle swarm algorithm according to the optimal frog individuals in each group of frog individuals, updating the optimal particles based on the particle swarm algorithm, and outputting the optimal radial basis centers and the optimal widths when the maximum iteration times are reached. The maximum iteration times can be divided into local maximum iteration times (maximum iteration times of a particle swarm algorithm) corresponding to groups and global maximum iteration times (maximum iteration times of a frog-leaping algorithm) corresponding to all the groups, and when the local maximum iteration times and the global maximum iteration times are reached at the same time, the maximum iteration times are determined.

It should be noted that the above solution of the preferred embodiment is only a specific implementation solution proposed by the present application, and other ways of determining the optimal radial base center and the optimal width of the RBF neural network based on the frog-particle swarm optimization all belong to the protection scope of the present application.

S105, driving a metal-oxide semiconductor field effect transistor (MOSFET) in the photovoltaic array according to a duty ratio value corresponding to the global maximum power point so as to track the global maximum power point.

Specifically, a MOSFET switch tube in the circuit is driven according to a PWM (Pulse width modulation) signal corresponding to a duty ratio value, so as to track a global maximum power point.

By applying the technical scheme, a photovoltaic array model is established according to the states of a bypass diode and a blocking diode in the photovoltaic array equivalent circuit under local shielding; determining the output voltage of a local maximum power point according to a power-voltage P-U curve corresponding to the photovoltaic array model; constructing a Lyapunov function, determining a stability function of an expected value of the current of the photovoltaic system according to a time derivative of the Lyapunov function, and taking a parameter determined by the Lyapunov function as an input variable of the RBF neural network; determining a duty ratio function corresponding to the output voltage based on a Radial Basis Function (RBF) neural network optimized by a frog-particle swarm optimization algorithm; comparing the local maximum power points determined by the duty ratio function to determine a global maximum power point; and driving a metal-oxide semiconductor field effect transistor (MOSFET) in the photovoltaic array according to the duty ratio value corresponding to the global maximum power point so as to track the global maximum power point. Therefore, the global maximum power point can be quickly and stably tracked, the unstable state of photovoltaic voltage and current near the maximum power point is improved, and the photovoltaic power generation efficiency is improved.

In order to further illustrate the technical idea of the present invention, the technical solution of the present invention will now be described with reference to specific application scenarios.

The embodiment of the invention provides a photovoltaic array global maximum power point tracking method under local shielding. Under the condition of local shielding, the output characteristic curve of the photovoltaic array is in a plurality of local peak values, and Maximum Power Point Tracking (MPPT) of the photovoltaic array is influenced. The traditional MPPT algorithm enables the photovoltaic system to work at a local maximum power point, and a real global maximum power point is missed. As shown in fig. 2, the embodiment of the present invention adopts a segment search, and performs segment combination on the P-U output curve of the photovoltaic array, so as to determine the interval where the local maximum power point is located and the voltage value corresponding to each interval, and use the interval as the input of the lyapunov stable control system. In the Lyapunov stable control, a duty ratio function of a power switch device is determined by setting a tracking error of a photovoltaic power generation system according to a Lyapunov function and a RBF neural network optimized based on a frog-particle swarm optimization, and finally a new DC-DC controller is constructed, so that the global maximum power point of the photovoltaic power generation system is obtained. An integral link is introduced into the constructed Lyapunov function, so that the overall gradual stability of the photovoltaic system in the operation process is ensured, the robustness of the nonlinear control of the photovoltaic system is improved, the maximum power point can be quickly and stably tracked, the unstable state of photovoltaic voltage and current near the maximum power point is improved, and the energy utilization efficiency of the photovoltaic cell is improved.

The photovoltaic modules mostly adopt a series-parallel connection mode to form a photovoltaic array, namely M photovoltaic modules are connected in series to form a line, N series of photovoltaic modules are connected in parallel, the photovoltaic array formed by M multiplied by N photovoltaic modules is shown in figure 3, and diodes which are reversely connected in parallel at two ends of the series photovoltaic modules in the figure are bypass diodes D_bIts main role is to eliminate the "hot spot effect" that occurs in the case of local occlusion. In addition, when one photovoltaic module is shielded or breaks down to stop power generation, forward bias is formed at two ends of the diode, so that normal power generation of other photovoltaic modules is not influenced, and the photovoltaic modules are protected from being damaged by higher forward bias or heat. The diodes connected in series on each string of photovoltaic devices are blocking diodes D_pThe function of the photovoltaic module is to prevent reverse current from flowing through the photovoltaic module. Under the condition of local shielding, the current-voltage (I-U) and power-voltage (P-U) curves of the photovoltaic array are influenced by the bypass diode and the blocking diode, and the multimodal phenomenon occurs.

As shown in fig. 4, which is a schematic diagram of the principle of the photovoltaic array in the embodiment of the present invention, in the case of uniform illumination, the current flowing through the cells of the photovoltaic module is the same, and the parallel bypass diodes are in a blocking state under reverse bias; when a certain photovoltaic module on the photovoltaic array is shielded, namely illumination on the photovoltaic module is unevenly distributed, the bypass diode is in a conducting state under forward bias.

FIG. 5 is a schematic diagram of a photovoltaic array simulation model under partial occlusion in an embodiment of the invention. In a simulation experiment, 3 photovoltaic modules are set to be subjected to different illumination intensities, and the output curve of the photovoltaic array is multi-peak phenomenon under local shielding through simulation verification.

The photovoltaic array global maximum power point tracking method under the local shielding mainly comprises the following steps:

step one, establishing a photovoltaic array model according to states of a bypass diode and a blocking diode in a photovoltaic array equivalent circuit under local shielding, wherein the bypass diode is connected with each photovoltaic module in the equivalent circuit in parallel, the blocking diode is connected with the positive electrode side of each group of series photovoltaic modules in series, and the states are a conduction state or a blocking state, as shown in fig. 3.

And step two, determining the interval of the local maximum power point according to the power-voltage P-U curve of the photovoltaic array. The method specifically comprises the following steps:

step a, segmentation

And dividing the P-U curve into 10G small intervals, wherein G is the number of the photovoltaic cells.

Step b, merging

Calculating the power-voltage P-U change rate between each cellAnd determining the interval of the local maximum power point.

1) If the sign of the rate of change of the power-voltage P-U is the same between adjacent zones, i.e.Combining the intervals to form a new interval;

2) if the sign of the change rate of the power-voltage P-U is different between the adjacent intervals, the change rate of the P-U is equal to that of the adjacent intervalsToIf so, merging, otherwise, not merging.

As shown in FIG. 6, the P-U curves are segmented and merged into three intervals A-B-C, C-D, D-E. Each interval has a local peak corresponding to P_m＝I_m*U_m. Therefore, one coordinate point (U) exists for each section_m,I_m) Is the local maximum power point of the segment. When interference factors such as temperature and illumination exist, the global maximum power point is difficult to determineThe neural network can track the maximum power point in the interference environment, and has the advantages of high tracking speed, interference resistance and the like, so that the divided sections need to be deeply searched through the neural network, and the local maximum power point P is determined_mAnd determining the output voltage according to the P-U curve interval where the local maximum value point is located.

And step three, determining a state equation of the photovoltaic power generation system.

As shown in fig. 7, which is a schematic diagram of a principle of a photovoltaic power generation system in an embodiment of the present invention, equations of the photovoltaic power generation system are listed according to kirchhoff current law and kirchhoff voltage law:

in the equation:wherein x is₁Current piecewise function output voltage U as global maximum power point_pv，x₂For the inductive current I in photovoltaic power generation systems_L,x₃For photovoltaic power generation system current I_pvAnd D is the duty ratio of a power switching device in a DC-DC booster circuit in the photovoltaic power generation system. Duty cycle D and x₁,x₂,x₃,There is a non-linear relationship for a total of 5 parameters.

Step four, judging the stability of the system by the Lyapunov function

The second method of lyapunov judging stability is to analyze the judging stability by defining lyapunov scalar function. Closed loop automatic control techniques are based on the concept of feedback to reduce uncertainty. The elements of the feedback theory include three parts, measurement, comparison andand (6) executing. What is essential to the measurement is the actual value of the controlled variable, which is compared with the desired value, and this deviation is used to correct the response of the system and to perform the regulation control. First, a tracking error is set to e₁Definition of e₁＝y-y_refWherein y is_refIs the desired output value of the PV array, then e₁The time derivative of (a) is:

in engineering practice, the most widely used regulator control law is proportional, integral and derivative control, abbreviated as PID control, also known as PID regulation. A PID controller (proportional-integral-derivative controller) is a common feedback loop component in industrial control applications, consisting of a proportional unit P, an integral unit I and a derivative unit D. The basis of PID control is proportional control; integral control may eliminate steady state errors, but may increase overshoot; differential control can accelerate the response speed of the large inertia system and weaken the overshoot tendency. Thus, the lyapunov function is constructed from three parts (proportional-integral-derivative). The Lyapunov function was constructed as:

and setting a virtual control quantity through a state equation of the system, constructing a proper Lyapunov function, calculating a control rate, and judging the stability of the system through the Lyapunov function.

And step five, determining a duty ratio function of the output voltage of the local maximum power point.

A prediction model of a duty ratio D is established by adopting a RBF neural network based on a frog-particle swarm algorithm, the radial basis center mu and the width sigma of the RBF neural network are substantially used as frogs in the frog-particle swarm algorithm, parameters are updated by continuous optimization, and learning and training of the network are realized.

The photovoltaic cell model is adopted, so that the model works under different environments and illumination intensities, and every time is recordedOutput voltage U of the scale_pvInductance current I_LBoost voltage U_deThe output current y of the photovoltaic array and the duty cycle D at the maximum power point corresponding thereto. In the patent, 200 groups of data are collected, 180 groups of data are randomly selected to serve as a training sample set, and the rest 20 groups of data serve as the training sample set. The measured variable is subjected to a derivation system equation to obtain an input quantity x₁,x₂,x₃,The algorithm flow chart is shown in fig. 9, and the specific steps are as follows:

1. determining RBF neural network topology

Radial basis function ofWhere x is the input to the network, mu_k,σ_kThe center and width of the kth hidden layer neuron, respectively.

(1) Determining the number of the neural network layers as 3: the RBF neural network is a three-layer neural network, which comprises an input layer, a hidden layer and an output layer.

(2) Determining the number of neurons in an input layer; because the input variable is 5The number of input layer neurons was 5.

(3) Determining the number of neurons in an output layer to be 1; the output variable is the duty ratio D, so the number of neurons in the output layer is 1.

(4) The number of neurons in the hidden layer was determined to be 20.

2. The radial basis center μ and width σ are initialized.

3. The radial basis center μ and width σ are input to the frog-leap algorithm as frog individuals.

4. The frog population and associated parameters are initialized. Randomly generating P frog individuals, and determining related parameters (total iteration times and dimension D) of an algorithm; wherein the ith frog can be coded as X_i＝(X_i1,X_i2X_i3ΛX_iD)。

5. And calculating the fitness. And calculating the fitness value of each frog individual according to the fitness function.

And mapping the cost function of the neural network into a target function (fitness function) to be converged by the mixed frog-leaping particle swarm optimization. The fitness function calculation formula is as follows:wherein E is_avNet (N) is the actual output of the neural network, d (N) is the corresponding ideal output, and N is the number of training samples of the neural network. And substituting the training sample value into a fitness function calculation formula to calculate the fitness value of each frog individual.

6. The frogs are grouped. All frogs are sorted by goodness of fitness value and grouped. Dividing the data into m subgroups according to the sequence of the fitness values, enabling the frog individual arranged at the first position to enter a subgroup 1, enabling the frog individual arranged at the second position to enter a subgroup 2, enabling the frog individual arranged at the third position to enter a subgroup 3, …, enabling the frog individual arranged at the mth position to enter a subgroup m, and enabling the frog individual arranged at the m +1 position to enter the subgroup 1 until the end. Until all the frog individuals are divided, the division of m subgroups is completed.

7. And calculating the optimal frog individuals of each group. Carrying out local search, calculating the optimal frog individual of each group, and recording the optimal frog individual in the current subgroup as X in each subgroup_b。

8. And substituting the optimal frog individuals of each group as particles in the particle swarm into a particle swarm algorithm.

9. And updating the optimal particles. The position and speed updating strategy of the particles in the algorithm is as follows:

w is the inertial weight, c1, c2, c3 represent the weight factors of the corresponding parts, r1, r2, r3 are [0,1]Random number between, mu₁，μ₂Is a speed control factor, p_idIs the best position of the ith particle, p_igdIs the optimal solution in the subgroup, and p_gdIs the global optimal solution of the population. Taking a speed control factor mu₁＝0.5，μ₂＝0.25。

10. And judging whether the particle swarm algorithm reaches the maximum iteration number.

11. If the maximum iteration times of the particle swarm algorithm are reached, judging whether the leapfrog algorithm reaches the maximum iteration times; if not, returning to the step 9;

12. if the maximum iteration times of the frog-leaping algorithm is reached, outputting the optimal radial basis center mu and the optimal width sigma; if not, returning to the step 5;

13. substituting the output optimal radial basis center mu and width sigma into the RBF neural network;

14. calculating the training error (error function) of the network; the difference between the real and expected values is represented using a cross entropy loss function,where y represents the true value, i.e., the output value of the RBF neural network,indicating the expected value.

15. Readjusting the network parameters through the calculated error values;

16. judging whether the maximum iteration number of the RBF neural network is reached;

17. if yes, outputting a duty ratio D; if not, returning to 14;

and step six, comparing the local maximum power points determined by the duty ratio function to determine a global maximum power point.

And step seven, driving a metal-oxide semiconductor field effect transistor (MOSFET) in the photovoltaic array according to the duty ratio value corresponding to the global maximum power point, and tracking the maximum power point of the photovoltaic power generation system.

As shown in fig. 8, the entire photovoltaic system consists of three parts: the photovoltaic module, DC-DC Boost booster circuit, MPPT controller and battery. The hardware design mainly comprises a sampling circuit, a control circuit, a driving circuit, a voltage stabilizing circuit and the like. The purpose of adding the driving circuit is to drive a MOSFET switching tube in a PWM (Pulse width modulation) signal driving circuit output by the singlechip. The specific working scheme is as follows: firstly, a voltage and current sampling circuit is utilized to send the detected output voltage and output current of the photovoltaic module to an STM32 single chip microcomputer. And then the single chip microcomputer changes the duty ratio of the PWM wave output by the single chip microcomputer according to the duty ratio function. The PWM wave is input into a driving circuit to drive a power switch tube in the BOOST circuit, so that the global maximum power point tracking under local shielding is realized.

The DSP samples the voltage and current output by the photovoltaic cell through a sampling circuit of the DSP, then an internal A/D conversion module processes the sampled voltage and current, analog quantity is converted into 0/1 digital quantity, PWM control signals are generated through an MPPT control program, the PWM signals are amplified and strengthened through a driving circuit, switching-on and switching-off of a switching tube Q are achieved, the duty ratio of a Boost circuit is changed, matching of internal impedance and external equivalent impedance of a photovoltaic module is achieved, and MPPT control of a photovoltaic array is completed.

As shown in fig. 8, when the photovoltaic cell is in a working state, firstly, the output of the photovoltaic cell is continuously sampled by a sampling circuit, the signal is sent to an IO port of a processor after being processed by a conditioning circuit, the processor receives the signal and analyzes and calculates the signal, the output power value of the current photovoltaic cell is calculated by using a formula P ═ UI, the output power value is stored in a memory of the processor, the output power value is compared and calculated, then, a maximum power point tracking algorithm in a main program is called, and the duty ratio of a switching tube in a Boost circuit is modified in real time, so that the output voltage is continuously changed, the requirement of the output maximum power value is met, and the purpose of tracking and controlling the maximum power point is achieved. After the voltage output by the photovoltaic cell is boosted by the Boost circuit, the voltage is subjected to DC/DC conversion and DC/AC conversion by the voltage conversion circuit so as to meet the requirements of alternating current and direct current loads.

The perturbation and observation method is a maximum power tracking method which is earlier applied to a photovoltaic power generation system and is called as a traditional maximum power tracking method. The disturbance observation method has simple control thought and convenient realization, can realize the tracking of the maximum power point and improve the utilization efficiency of the system. However, the disturbance observation method only takes the output power of the photovoltaic module twice before and after as the target to study, and does not consider the influence of the external environmental condition change on the output power of the photovoltaic array twice before and after, so that the misjudgment of the method is easy to occur in the using process, the misjudgment increases the tracking time, the output efficiency of the photovoltaic array is reduced, and the tracking failure is caused seriously, so that the method cannot accurately track the maximum output power. In order to verify the superiority of the photovoltaic array MPPT method under the local shielding in the embodiment of the invention, the MPPT method is compared with the MPPT algorithm combining the segmented search and the traditional disturbance observation method in a simulation mode, so that the two MPPT algorithms are set to be consistent in the variation trend of the illumination intensity. The output voltage curve of the photovoltaic array is shown in fig. 10 and fig. 11, the output power curve of the photovoltaic array is shown in fig. 12 and fig. 13, and as can be seen from a comparison graph of two MPPT simulation results, the MPPT algorithm based on the combination of the segmented search and the Lyapunov stable function can stably find the maximum power point of the photovoltaic array under the condition of local shielding, the output curve harmonic wave in the optimization process is less, the waveform is stable, and the oscillation near the maximum power point is low.

In order to achieve the above technical object, an embodiment of the present application further provides a device for tracking a global maximum power point of a photovoltaic array under local shading, as shown in fig. 14, the device including:

the building module 401 is configured to build a photovoltaic array model according to states of a bypass diode and a blocking diode in a photovoltaic array equivalent circuit under local shielding, where the bypass diode is connected in parallel with each photovoltaic module in the equivalent circuit, the blocking diode is connected in series with a positive electrode side of each group of series-connected photovoltaic modules, and the states are a conducting state or a blocking state;

a first determining module 402, configured to determine an output voltage of a global maximum power point according to a power-voltage P-U curve corresponding to the photovoltaic array model;

a second determining module 403, configured to determine, based on a leapfrog-particle swarm optimization RBF neural network, a duty cycle function corresponding to the output voltage;

and the driving module 404 is configured to drive a metal-oxide semiconductor field effect transistor MOSFET in the photovoltaic array according to the duty ratio value corresponding to the global maximum power point, so as to track the global maximum power point.

In a specific application scenario, the first determining module 402 is specifically configured to:

averagely dividing a power-voltage P-U curve into a plurality of sections;

judging and combining the sections according to the power-voltage P-U change rate to obtain the interval where each local maximum value point is located;

and determining the output voltage according to the P-U curve interval where the local maximum value point is located.

In a specific application scenario, the second determining module 403 is specifically configured to:

determining an input variable of the RBF neural network based on a Lyapunov second-class method;

determining the topological structure of the RBF neural network according to the input variable;

determining the optimal radial basis center and the optimal width of the RBF neural network based on a frog-particle swarm algorithm;

training the RBF neural network according to the optimal radial basis center and the optimal width;

and outputting the duty ratio function according to the training result.

In a specific application scenario, the second determining module 403 is further specifically configured to:

establishing an array equation of the photovoltaic array according to kirchhoff's law and the output voltage, wherein the array equation comprises the output voltage, the photovoltaic system current, the inductive current, the boost voltage of the photovoltaic array and the duty ratio;

constructing a Lyapunov function according to a preset first tracking error and the array equation;

determining a stabilization function of the expected value of the photovoltaic system current according to the time derivative of the Lyapunov function;

constructing a second tracking error based on the stabilization function and the actual value of the photovoltaic system current;

determining the output voltage, the photovoltaic system current, the first tracking error, and the second tracking error as the input variables.

In a specific application scenario, the second determining module 403 is further specifically configured to:

determining frog individuals in a frog-leaping algorithm according to the initial radial base center and the initial width of the RBF neural network;

randomly generating a preset number of frog individuals and determining the maximum iteration times and dimensionality;

determining the fitness value of each frog individual according to a fitness function;

grouping the frog individuals according to the fitness value;

determining particles of a particle swarm algorithm according to the optimal frog individuals in each group of frog individuals;

updating the optimal particles based on the particle swarm algorithm;

outputting the optimal radial basis center and the optimal width when the maximum iteration number is reached.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not necessarily depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.