CN112596575B - A Maximum Power Point Tracking Method Based on NPSO Algorithm and Hierarchical Automatic Restart - Google Patents

Abstract

The invention discloses a maximum power point tracking method based on an NPSO algorithm and a hierarchical automatic restart, which is characterized in that a hierarchical automatic restart method is added before a maximum power point tracking method based on the NPSO algorithm, the maximum power point tracking method based on the NPSO algorithm is characterized in that on the basis of a PSO algorithm, the duty ratio D of a PWM signal is used as the position of a population particle, the output power is used as the adaptive value of the population particle, the population particle is divided into a convergence particle and a free particle, the convergence particle has the same property with the particle in the PSO algorithm, the free particle has no memory, the convergence particle and the free particle are updated at the same time, and the individual optimal position of the convergence particle and the global optimal position of the population are updated through the positions of the convergence particle and the free particle after each update; when the convergence particle falls into the local optimum solution, i.e. the local optimum position, the convergence particle is pulled out of the local optimum solution by using the free particle. The power generation efficiency and the stability are improved.

Description

A Maximum Power Point Tracking Method Based on NPSO Algorithm and Hierarchical Automatic Restart

technical field

The invention belongs to the technical field of new energy photovoltaic power generation, and relates to a maximum power point tracking method based on NPSO (Novel Particle Swarm Optimization) algorithm and hierarchical automatic restart.

Background technique

With the massive investment in various production equipment and household appliances, the global demand for electric energy is increasing. In the face of limited fossil fuels and infinitely growing demand for electricity, the world has turned its attention to renewable energy. Among them, solar energy has gained global popularity due to its advantages of low pollution, low noise and inexhaustible supply. Approved. According to data from the International Renewable Energy Agency (IRENA), the cumulative installed capacity of photovoltaics in the world from 2010 to 2019 maintained a steady upward trend, reaching 578,533MW in 2019, an increase of 20.3% over 2018. In complex environments such as partial shadows, the PV array's P-V output curve is a multi-peak nonlinear curve with multiple local maxima and one maximum value. At present, since the photovoltaic power generation system has the ability to be automated and unattended, once it falls into the local power maximum value, the photovoltaic power generation system will run at the local power maximum value for a long time, resulting in a reduction in efficiency and a waste of energy. . In order to ensure the high power generation efficiency of the photovoltaic power station, the photovoltaic power generation system must run at the maximum power point in real time.

The P-V output characteristic curve of the photovoltaic array is a nonlinear curve with one or more peaks. The output power can be adjusted by adjusting the output voltage of the photovoltaic array. When the output voltage of the photovoltaic array corresponds to the maximum output voltage of the photovoltaic array When the power is high, the tracking of the maximum power point is realized. The traditional MPPT (Maximum PowerPoint Tracking, maximum power point tracking) technology has the defect that it is easy to fall into the local power maximum value. MPPT technology has always been a hot issue that people pay attention to. Domestic and foreign scholars have proposed many effective methods. The methods include disturbance observation method, conductance increment method, etc. These methods are widely used in photovoltaic power generation systems due to their mature technology and easy implementation. However, the above-mentioned traditional MPPT method has the disadvantages of slow convergence speed, large oscillation amplitude and easy to fall into local power maximum value, etc., and is not suitable for application in photovoltaic power generation systems in complex environments. In recent years, people have applied the intelligent control algorithm to the MPPT technology, and some breakthroughs have been made. In Liu Yanli, Zhou Hang, Cheng Ze. MPPT control method of photovoltaic system based on particle swarm optimization [J]. Computer Engineering, 2010, 36(15): 265-267 A MPPT technology based on PSO (Particle SwarmOptimization) algorithm was proposed , which effectively improves the speed and accuracy of the maximum power point tracking, and reduces the possibility of the photovoltaic power generation system falling into the local power maximum value. However, due to the defect of the PSO algorithm itself, it is easy to fall into the local optimal solution, so the MPPT technology does not It completely solves the problem that the photovoltaic power generation system is easy to fall into the local power maximum value.

On the other hand, the external environment is constantly changing, such as the movement of clouds, the shielding area of the photovoltaic array dust, and the shielding area of shadows such as trees, etc. When the external environment changes, the maximum power point of the photovoltaic power generation system will also occur accordingly. Therefore, it is necessary to restart the MPPT algorithm and re-find the maximum power point. However, how the photovoltaic power generation system senses changes in the external environment and how to restart after sensing the changes in the external environment is another problem faced by solving this problem. In this regard, some scholars have proposed a periodic scanning method, which proposes to restart the MPPT algorithm at regular intervals to track the maximum power point in real time. However, the periodic scanning method will cause untimely and useless scanning, which cannot be guaranteed The power generation efficiency of photovoltaic power generation systems. And the fixed-cycle scan method restarts with the maximum search area [0,1] of the PWM duty cycle during the restart process. In this way, even if the external environment changes very little, the restart method is based on The maximum search area [0,1] is restarted, which will prolong the tracking time and increase the disturbance of the power generation system. The larger disturbance generated during the restart process will not only have an impact on the power generation system and the power grid, but also reduce the quality of power, and then It will cause certain damage to electrical equipment, and will also reduce the power generation efficiency of photovoltaic power generation systems, resulting in wasted energy, and these wasted energy will inevitably be dissipated into the atmosphere in the form of heat, accelerating global warming. Therefore, the MPPT technology needs further exploration to solve its shortcomings.

SUMMARY OF THE INVENTION

The purpose of the embodiments of the present invention is to provide a maximum power point tracking method based on the NPSO algorithm, so as to solve the problem that the existing maximum power point tracking algorithm is easy to fall into the local power maximum value and cause low power generation efficiency when it is blocked by local shadows.

Another object of the embodiments of the present invention is to provide a hierarchical automatic restart method of the maximum power point tracking algorithm, so as to solve the problem that the restart algorithm of the existing maximum power point tracking algorithm cannot track the maximum power point in time after the external environment changes , or restart the MPPT algorithm without changing the external environment, resulting in low power generation efficiency and increased power fluctuations.

Another object of the embodiments of the present invention is to provide a maximum power point tracking method based on the NPSO algorithm and hierarchical automatic restart.

In order to solve the above-mentioned technical problem, the embodiment of the present invention provides a maximum power point tracking method based on the NPSO algorithm. On the basis of the PSO algorithm, the duty cycle D of the PWM control signal of the photovoltaic power generation system is used as the position of the population particle, Taking the output power P _pv of the photovoltaic array of the photovoltaic power generation system as the fitness value of the population particles, the population particles are divided into two types: convergent particles and free particles. The properties of convergent particles are consistent with those in the PSO algorithm. Particles and free particles are updated simultaneously, the velocity and position of convergent particles are updated according to the update method of particle velocity and position in the PSO algorithm, the position of each free particle is randomly updated within a given search interval, and the search interval of free particles It divides the global search interval of the convergent particles into equal parts according to the number of free particles, and uses the position of the convergent particle to update the individual optimal position of the convergent particle through the fitness value of the convergent particle and the free particle after each update. update the global optimal position of the population with the position of the free particle; when the convergent particle falls into the local optimal solution, that is, the local optimal position, all free particles continue to search in the global search interval of the convergent particle, and at the position of a certain free particle When it is better than the current global optimal solution, the position of the free particle is used to update the global optimal position, that is, the global optimal solution, and the convergent particle is pulled out of the local optimal solution, so that the convergent particle continues to search for the global optimal solution.

In order to solve the above technical problem, the embodiment of the present invention also provides a hierarchical automatic restart method of the maximum power point tracking algorithm. The maximum power point tracking algorithm is the above-mentioned maximum power point tracking method based on the NPSO algorithm. Proceed as follows:

The first step is to collect the output voltage U _s of the photovoltaic array in real time through the voltage sensor, and then calculate the rate of change ΔU of the output voltage of the photovoltaic array according to formula (9):

ΔU=|U _s −U _max |/U _max ; (9)

Among them, U _max is the output voltage at the maximum power point collected after the photovoltaic array performed the maximum power point tracking algorithm last time;

The second step is to intelligently perceive the degree of change of the external environment through the hierarchical method according to the rate of change ΔU of the output voltage of the photovoltaic array, and reset the global search interval according to the different degrees of change of the external environment, and finally use the reset global search interval to restart the maximum The power point tracking algorithm re-tracks the maximum power point of the photovoltaic power generation system.

In order to solve the above technical problem, the embodiment of the present invention also provides a maximum power point tracking method based on the NPSO algorithm and hierarchical automatic restart, which adds the above-mentioned method before the above-mentioned maximum power point tracking method based on the NPSO algorithm. Described a kind of automatic restart method of a kind of maximum power point tracking algorithm, its concrete steps are as follows:

Step S1, take the duty ratio D of the PWM control signal of the photovoltaic power generation system as the position of the population particle, and use the photovoltaic array output power P _pv of the photovoltaic power generation system as the fitness value of the population particle, and divide the population particle into a plurality of convergent particles and For two free particles, the global search interval for the position of the convergent particle is [0, 1]. Among the two free particles, the search interval for free particle 1 is [0, 0.5], and the search interval for free particle 2 is [0.5, 1]. , and initialize the particle population;

其中，0＜i＜N_p，i′＝1,2，N_p为收敛粒子总数；Step S2, obtain the real-time output voltage V _pv and real-time output current I _pv of the photovoltaic array, and calculate the adaptive value of the i-th convergent particle after the k-th generation update according to the real-time output voltage V _pv and real-time output current I _pv of the photovoltaic array The fitness value of P _i ^k and the i′-th free particle after the update of the k-th generation

Among them, 0<i<N _p , i′=1,2, N _p is the total number of convergent particles;

更新第i个收敛粒子的个体最优位置Dpbest_i和种群的全体最优位置Dgbest，若P_i ^k＞Ppbest_i，则令Ppbest_i＝P_i ^k、

否则Ppbest_i和Dpbest_i不变；若P_i ^k＞Pgbest，则令Pgbest＝P_i ^k、

否则Pgbest和Dgbest不变；若

则令

否则Pgbest和Dgbest不变，其中，

为第i个收敛粒子在第k代更新后的位置，

为第i′个自由粒子在第k代更新后的位置，Ppbest_i为第i个收敛粒子的个体最优适应值，Pgbest为种群的全体最优适应值；Step S3, according to the updated fitness value P _i ^k of the i-th convergent particle in the k-th generation and the fitness value of the i′-th free particle after the update of the k-th generation

Update the individual optimal position Dpbest _i of the i-th convergent particle and the overall optimal position Dgbest of the population. If P _i ^k >Ppbest _i , then let Ppbest _i =P _i ^k ,

Otherwise, Ppbest _i and Dpbest _i remain unchanged; if P _i ^k >Pgbest, then let Pgbest=P _i ^k ,

Otherwise, Pgbest and Dgbest remain unchanged; if

order

Otherwise Pgbest and Dgbest remain unchanged, where,

is the updated position of the i-th convergent particle in the k-th generation,

is the updated position of the ith free particle in the kth generation, Ppbest _i is the individual optimal fitness value of the ith convergent particle, and Pgbest is the overall optimal fitness value of the population;

Step S4, update the velocity and position of the convergent particle, and the position of the two free particles;

Step S5, determine whether the number of iterations, that is, the update algebra k satisfies k>gen _max , that is, determine whether the number of iterations reaches the maximum number of iterations gen _max , if so, end the iteration, and execute step S6, otherwise add 1 to the number of iterations k and return to step S2 continue to iterate;

Step S6, calculating the rate of change ΔU of the output voltage of the photovoltaic array;

Step S7, according to the rate of change ΔU of the output voltage of the photovoltaic array, intelligently perceive the degree of change of the external environment through a hierarchical method, and judge whether it is necessary to return to step S1 to restart according to the different degrees of change of the external environment, and when it is judged that it is necessary to return to step S1 to restart, for The change degree of the external environment of different levels resets the global search interval, then returns to step S1 to restart, and uses the reset global search interval to update the global search interval of convergent particles and free particles in the original step S1.

The beneficial effects of the embodiments of the present invention are: through the secondary development of the PSO algorithm, an NPSO algorithm is proposed. The algorithm divides the population particles into two types: convergent particles and free particles. Since free particles do not have convergence, they will always be Search in the search interval. When the convergent particle reaches convergence, that is, when the convergent particle falls into the local power maximum, the free particle still searches randomly in the global space. When the free particle searches for a larger power, the free particle will It will share its position with the convergent particles, that is, update the global optimal solution to the solution searched by the free particles at this time, thereby pulling the convergent particles out of the local optimal solution, enhancing the global search ability of the PSO algorithm and making up for the PSO algorithm. After the population particle converges, it will lose the ability of global search, and the NPSO algorithm is applied to the MPPT technology to form a maximum power point tracking method based on the NPSO algorithm, which solves the problem of the existing maximum power point tracking algorithm in photovoltaic arrays. It is easy to fall into the problem of low power generation efficiency due to local power maximum in complex environments such as partial shadow occlusion, which improves the power generation efficiency of photovoltaic power generation systems.

A hierarchical automatic restart method of the maximum power point tracking algorithm is proposed. The change of the external environment will first cause the change of the output voltage of the photovoltaic array. The output voltage of the photovoltaic array is collected by the voltage sensor, and the rate of change of the voltage is used to judge whether it is Restart the maximum power point tracking algorithm, so as to achieve the purpose of real-time tracking, can automatically detect whether the external environment has changed, and realize the function of real-time tracking of the maximum power point; The degree of change of the environment is classified, which can intelligently perceive the degree of change of the external environment, and reasonably formulate the global search interval, so that the particles can search near the original maximum power point, avoid useless searches, and reduce the tracking time and the power fluctuation during the search process. It reduces the power fluctuation when restarting the MPPT algorithm, and effectively solves the problem that the restart algorithm of the existing MPPT algorithm cannot track the MPPT in time after the external environment changes, or restarts without changing the external environment. The maximum power point tracking algorithm is adopted, which causes the problem of low power generation efficiency and increased power fluctuation. The photovoltaic power generation system is more stable and the power generation efficiency is higher.

Description of drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative efforts.

Figure 1 is a graph of the P-V output characteristic of a photovoltaic array.

FIG. 2 is a flowchart of a maximum power point tracking method based on an NPSO algorithm and hierarchical automatic restart according to an embodiment of the present invention.

Figure 3 is a schematic diagram of a photovoltaic system Boost boost circuit.

Figure 4 is a simulation model diagram of a photovoltaic power generation system based on a boost circuit.

Figure 5(a) is a simulation result of the maximum power point tracking method based on the PSO algorithm under the illumination mode 3 with uneven partial shadow occlusion.

Fig. 5(b) is a simulation result diagram of the maximum power point tracking method based on the NPSO algorithm according to the embodiment of the present invention under the illumination pattern 3 with uneven partial shadow occlusion, and the illumination pattern with uneven partial shadow occlusion is the same as that of Fig. 5 ( a) Consistent.

Fig. 6(a) is a simulation result diagram of the maximum power point tracking method based on the periodic scanning method when the external illumination environment changes.

FIG. 6(b) is a simulation result diagram of a hierarchical automatic restart method of a maximum power point tracking algorithm according to an embodiment of the present invention when the external lighting environment changes, and the external lighting environment changes are consistent with FIG. 6(a).

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

Taking the SIEMENS SP75 solar panel as an example to illustrate the situation that the P-V output curve of the photovoltaic array is multi-peak, a photovoltaic array simulation model consisting of three photovoltaic panels in series is built under Simulink, and the parameters of each photovoltaic panel are listed in Table 1. The simulated illumination conditions and simulation results are shown in Table 2, and the P-V output curves of the photovoltaic panel under three illuminations are shown in Figure 1.

Table 1 SIEMENS SP75 photovoltaic panel parameters

parameter value Maximum power (W) 75 Maximum power point voltage (V) 17.0 Maximum power point current (A) 4.4 Short circuit current (A) 4.8 Open circuit voltage (V) 21.7

Table 2 PV array output characteristic data

In the P-V output characteristic curves of the photovoltaic array under the three illumination modes shown in Figure 1, mode 1 is uniform illumination, and modes 2 and 3 are respectively subjected to partial shadows to block uneven illumination. It can be seen from Figure 1 that mode 1 has one peak, mode 2 has two peaks, and mode 3 has three peaks. In mode 2 and mode 3, due to the existence of multiple peaks, there is a risk of tracking a local maximum instead of a maximum when performing MPPT tracking. Most of the existing technical solutions are researched under the condition of ideal uniform illumination and no shadow occlusion, that is, the P-V output characteristic curve of the photovoltaic array is a single peak, which is inconsistent with reality and does not consider uneven illumination or partial shadow occlusion. , all have the problem of easily falling into the local power maximum when they are occluded by uneven illumination or partial shadows. For example, in the MPPT technology based on the PSO algorithm, the population of particles gradually converges in the iterative process, and the searchable interval becomes smaller and smaller. When all the particles converge to a point, the PSO algorithm completely loses the global search ability. The power found at this point is not the maximum power value, then it falls into the local power maximum value and cannot jump out.

和位置

两个参数。然后在全局搜索空间和预先定义的速度范围内，每个粒子随机获取自身的初始位置

和初始速度

即得到第一代粒子的位置和速度。根据目标函数即可计算出每个粒子位置

对应的适应值，其中目标函数是根据要求解的问题而定的。接着在每次迭代中粒子群中的每个粒子通过式(1)和式(2)进行更新自身的速度和位置，并将每个粒子搜索到的目前的最优位置作为粒子的个体最优解xpbest_i，并在粒子群中所有粒子的个体最优解xpbest_i中选取出最优的解作为整个粒子群的全局最优解xgbest。The PSO algorithm is inspired by the behavior of birds to find food. The algorithm has shown high performance in various optimization problems and has a wide range of applications. At the beginning of the PSO algorithm, first define a particle swarm with N particles, as the main body of the algorithm execution, each particle has a speed

and location

two parameters. Then within the global search space and a predefined velocity range, each particle randomly acquires its own initial position

and initial velocity

That is, the position and velocity of the first generation of particles are obtained. According to the objective function, the position of each particle can be calculated

The corresponding fitness value, where the objective function is determined according to the problem to be solved. Then, in each iteration, each particle in the particle swarm updates its own speed and position through Equation (1) and Equation (2), and takes the current optimal position searched by each particle as the individual optimal position of the particle. Solve xpbest _i , and select the optimal solution from the individual optimal solutions xpbest _i of all particles in the particle swarm as the global optimal solution xgbest of the entire particle swarm.

表示种群中第i个粒子在第k代更新后的位置，

表示种群中第i个粒子在第k-1代更新后的位置，

表示种群中第i个粒子在第k代更新后的速度，

表示种群中第i个粒子在第k-1代更新后的速度，i＝1,2,3...N，N是种群中粒子的总数；ω为惯性权重因子，rand()为介于0和1之间的随机数，c₁和c₂是学习因子。where k is the algebra of particle update,

represents the position of the i-th particle in the population after the update of the k-th generation,

represents the position of the ith particle in the population after the update of the k-1th generation,

represents the velocity of the ith particle in the population after the update of the kth generation,

Represents the velocity of the ith particle in the population after the update of the k-1th generation, i=1,2,3...N, N is the total number of particles in the population; ω is the inertia weight factor, rand() is between A random number between 0 and 1, c ₁ and c ₂ are learning factors.

表示粒子的位置，速度仍用

表示，

表示第i个粒子在第k代更新后的位置，

表示第i个粒子在第k更新后的速度，0<i<N，N为种群中的粒子总数。此时粒子的全局搜索空间也即粒子的位置范围为[0,1]，粒子的速度上限为v_max、下限为v_min，当采用公式(4)计算的

小于v_min时，

当采用公式(4)计算的

大于v_max时，

v_max和v_min具体的取值是根据实际的应用场景调试的。限制粒子的移动速度可以避免粒子的位置每次移动的过大，粒子位置移动的过大有可能会使粒子一直错过最优解，不利于粒子的收敛；另外粒子位置移动的过大，会使功率变化的过大，对电网不利。Further applying the PSO algorithm to the photovoltaic MPPT method, the position of the particles in the particle swarm represents the duty cycle D of the PWM control signal of the photovoltaic power generation system, so it can be used

Indicates the position of the particle, the velocity is still used

express,

represents the position of the i-th particle after the update of the k-th generation,

Indicates the velocity of the i-th particle after the k-th update, 0<i<N, where N is the total number of particles in the population. At this time, the global search space of the particle, that is, the position range of the particle is [0,1], the upper limit of the particle speed is v _max , and the lower limit is v _min , when the formula (4) is used to calculate

When less than v _min ,

When calculated using formula (4)

When greater than v _max ,

The specific values of v _max and v _min are debugged according to actual application scenarios. Limiting the moving speed of the particles can prevent the position of the particles from moving too large each time. If the position of the particles moves too much, the particles may miss the optimal solution all the time, which is not conducive to the convergence of the particles. Excessive power changes are detrimental to the power grid.

The objective function of the particle is shown in formula (3).

P _pv = Vpv × Ipv (3)

Among them, P _pv represents the output power of the photovoltaic array of the photovoltaic power generation system, that is, the fitness value of the population particles, Vpv represents the voltage value output by the photovoltaic array in real time, and Ipv represents the current value output by the photovoltaic array in real time.

In each generation, a comparison of the fitness value P _i ^k of the particle is carried out. P _i ^k represents the fitness value of the i-th particle in the k-th generation, and the fitness value is the standard for evaluating the position of the particle. The maximum fitness value in the algebra is recorded as the individual optimal fitness value Ppbest _i , and the position of the particle corresponding to Ppbest _i is defined as the individual optimal solution Dpbest _i of the particle. And select the optimal fitness value from the individual optimal fitness value Ppbest _i of all the particles as the global optimal fitness value Pgbest of the particle swarm, and define the position of the particle corresponding to Pgbest as the global optimal solution Dgbest of the particle swarm. In the photovoltaic MPPT method, the above equations (1) and (2) can be rewritten as equations (4) and (5).

表示种群中第i个收敛粒子在第k代更新后的速度，

表示种群中第i个收敛粒子在第k-1代更新后的速度，v_max为收敛粒子的速度上限，v_min为收敛粒子的速度下限，当采用公式(4)计算的

小于v_min时，

当采用公式(4)计算的

大于v_max时，

表示种群中第i个收敛粒子在第k代更新后的位置，

表示种群中第i个收敛粒子在第k-1代更新后的位置，i＝1,2,3...N，N是种群中粒子的总数，1<k<gen_max，gen_max为最大迭代次数；ω为惯性权重因子，rand()为[0,1]区间内的随机数，c₁和c₂是学习因子。in,

represents the velocity of the ith convergent particle in the population after the update of the kth generation,

Represents the velocity of the ith convergent particle in the population after the update of the k-1th generation, v _max is the upper limit of the velocity of the convergent particle, v _min is the lower limit of the velocity of the convergent particle, when calculated by formula (4)

When less than v _min ,

When calculated using formula (4)

When greater than v _max ,

represents the position of the i-th convergent particle in the population after the update of the k-th generation,

Represents the position of the ith convergent particle in the population after the update of the k-1th generation, i=1,2,3...N, N is the total number of particles in the population, 1<k<gen _max , gen _max is the maximum The number of iterations; ω is the inertia weight factor, rand() is a random number in the [0,1] interval, and c ₁ and c ₂ are learning factors.

The steps of the MPPT (Maximum Power Point Tracking) method based on the PSO algorithm are as follows:

种群粒子的位置

的全局区间为[0,1]，以光伏发电系统的光伏阵列输出功率P_pv作为粒子的适应值P_i ^k，并定义种群粒子的速度的上限v_max、下限v_min及最大迭代次数gen_max的值；初始化粒子群，即初始化种群中所有粒子在第一代的位置

在第一代的速度

粒子总数N、学习因子c₁和c₂以及惯性权重因子ω。Step 1. Take the duty cycle D of the PWM control signal of the photovoltaic power generation system as the position of the population particle

the position of the population particle

The global interval is [0,1], the output power P _pv of the photovoltaic array of the photovoltaic power generation system is used as the fitness value P _i ^k of the particles, and the upper limit v _max , the lower limit v _min and the maximum number of iterations gen _max of the speed of the population particles are defined The value of ; initialize the particle swarm, that is, initialize the positions of all particles in the swarm in the first generation

Speed in the first generation

The total number of particles N, the learning factors c ₁ and c ₂ and the inertia weight factor ω.

Step 2: Calculate the fitness value P _i ^k of the particle according to formula (3).

否则Ppbest_i和Dpbest_i不变；若P_i ^k＞Pgbest，则令Pgbest＝P_i ^k、

否则Pgbest和Dgbest不变；Step 3. If P _i ^k >Ppbest _i , then let Ppbest _i =P _i ^k ,

Otherwise, Ppbest _i and Dpbest _i remain unchanged; if P _i ^k >Pgbest, then let Pgbest=P _i ^k ,

Otherwise Pgbest and Dgbest remain unchanged;

Step 4: Update the velocity and position of the population particles according to the iterative formulas (4) and (5);

Step 5. If the termination condition k>gen _max of the particle swarm optimization algorithm is satisfied, that is, the maximum number of iterations is reached, then the iteration is ended, and step 6 is executed, otherwise, return to step 2;

Step 6: If the current environment does not change, output the global optimal solution Dgbest, and use Dgbest as the duty cycle of the PWM control signal of the photovoltaic power generation system, otherwise return to Step 1.

Example 1

The embodiment of the present invention proposes an NPSO algorithm through the secondary development of the PSO algorithm. The algorithm divides the population particles into two types: convergent particles and free particles, which enhances the global search ability of the PSO algorithm and makes up for the easy trapping of the PSO algorithm. Defects of local optimal solutions and apply NPSO algorithm to MPPT technology.

速度为

收敛粒子和自由粒子进行同代更新，在每次迭代时按照式(4)和式(5)对收敛粒子的速度以及位置进行更新，且目标函数如式(3)所示，即按照式(3)获取收敛粒子和自由粒子的适应值P_i ^k；自由粒子不具有记忆性，每次更新时在给定的搜索区间内随机取值，自由粒子的适应值用

表示，若自由粒子搜索到的解，也即自由粒子的位置

优于当前的全局最优解Dgbest时，把全局最优解更新为此时自由粒子搜索到的解，即令

当NPSO算法中的收敛粒子陷入局部最优解后，收敛粒子就失去了全局搜索能力，由于自由粒子仍然在全局搜索空间内进行搜索，因此当自由粒子更新全局最优解后，根据公式(3)和式(4)可知收敛粒子就会向着更新后的全局最优解方向移动，那么自由粒子就将收敛粒子拉出了局部最优，进而找到全局最优解。The NPSO algorithm divides the population particles into two categories: convergent particles and free particles. The number of free particles is two, which are divided into free particles 1 and free particles 2. The convergent particle is consistent with the particle properties in the PSO algorithm, and its position is

speed is

The convergent particle and the free particle are updated in the same generation, and the velocity and position of the convergent particle are updated according to formula (4) and formula (5) in each iteration, and the objective function is shown in formula (3), that is, according to formula ( 3) Obtain the fitness value P _i ^k of the convergent particle and the free particle; the free particle has no memory, and randomly selects a value in a given search interval during each update, and the fitness value of the free particle is

represents, if the solution found by the free particle is the position of the free particle

When it is better than the current global optimal solution Dgbest, update the global optimal solution to the solution searched by free particles at this time, that is, let

When the convergent particle in the NPSO algorithm falls into the local optimal solution, the convergent particle loses the global search ability. Since the free particle is still searching in the global search space, when the free particle updates the global optimal solution, according to formula (3 ) and equation (4), it can be known that the convergent particle will move towards the updated global optimal solution direction, then the free particle will pull the convergent particle out of the local optimal solution, and then find the global optimal solution.

The search process of free particles is divided into two stages by the threshold gen of the free particle search stage. The first stage is that the number of iterations of free particles and convergent particles is within the generation of gen, and free particles are searched in the global search space, and the global search space is [ 0,1], and defining the search interval of free particles is to divide the global search interval of convergent particles into equal parts according to the number of free particles. The role of free particles at this stage is to search in the global search interval to help convergent particles jump out of the local optimal solution. The second stage is that the number of population particle iterations is after the gen generation, free particles no longer perform global search, but take the global optimal solution Dgbest as the center, and perform random search in a tiny area with a radius of r ₁ . The role of free particles at this stage is to enable the current population particles to search for the global optimal solution more carefully, to improve the search accuracy, and to speed up the convergence speed of convergent particles.

The maximum power point tracking method based on NPSO algorithm, the specific steps are as follows:

收敛粒子在第一代的速度

收敛粒子的速度上限v_max、速度下限v_min，自由粒子在第一代位置

收敛粒子总数N_p，学习因子c₁和c₂，惯性权重因子ω，最大迭代次数gen_max以及自由粒子的搜索代数阈值gen，其中，0＜i＜N_p，i′＝1,2；Step 1. Take the duty cycle D of the PWM control signal of the photovoltaic power generation system as the position of the population particle, use the photovoltaic array output power P _pv of the photovoltaic power generation system as the fitness value of the population particle, and divide the population particle into multiple convergent particles. and two free particles, the global search interval of the position of the convergent particle is [0, 1], the search interval of free particle 1 of the two free particles is [0, 0.5], and the search interval of free particle 2 is [0.5, 1] ], the search interval of free particles is to divide the global search interval of convergent particles into equal parts according to the number of free particles; then initialize the particle population, that is, initialize the position of convergent particles in the first generation

Velocity of convergent particles in the first generation

The upper limit v _max and the lower limit v _min of the convergent particle, the free particle is at the position of the first generation

The total number of convergent particles N _p , the learning factors c ₁ and c ₂ , the inertia weight factor ω, the maximum number of iterations gen _max and the search algebraic threshold gen of free particles, where 0<i<N _p , i′=1,2;

Step 2. Obtain the real-time output voltage V _pv and real-time output current I _pv of the photovoltaic array, and calculate and calculate the i-th convergent particle after the k-th generation update according to the real-time output voltage V _pv and real-time output current I _pv by formula (3). The fitness value P _i ^k and the fitness value of the i′-th free particle after the update of the k-th generation

更新第i个收敛粒子的个体最优位置Dpbest_i和种群的全体最优位置Dgbest，若P_i ^k＞Ppbest_i，则令Ppbest_i＝P_i ^k、

否则Ppbest_i和Dpbest_i不变；若P_i ^k＞Pgbest，则令Pgbest＝P_i ^k、

否则Pgbest和Dgbest不变；若

则令

否则Pgbest和Dgbest不变；Step 3. According to the updated fitness value P _i ^k of the i-th convergent particle in the k-th generation and the fitness value of the i′-th free particle after the update of the k-th generation

Update the individual optimal position Dpbest _i of the i-th convergent particle and the overall optimal position Dgbest of the population. If P _i ^k >Ppbest _i , then let Ppbest _i =P _i ^k ,

Otherwise, Ppbest _i and Dpbest _i remain unchanged; if P _i ^k >Pgbest, then let Pgbest=P _i ^k ,

Otherwise, Pgbest and Dgbest remain unchanged; if

order

Otherwise Pgbest and Dgbest remain unchanged;

Step 4. Update the velocity and position of the convergent particles according to formulas (4) and (5), and update the positions of the two free particles according to formulas (6) to (8):

为自由粒子1在第k代更新后的位置，

为自由粒子2在第k代更新后的位置；d_min为收敛粒子的全局搜索区域的下限，d_max为收敛粒子的全局搜索区域的上限，即收敛粒子的全局搜索区域为[d_min,d_max]；rand()为区间[0,1]内的随机数，k为粒子迭代代数，自由粒子随机搜索的搜索半径r₁由式(8)确定：in,

is the updated position of free particle 1 in the kth generation,

is the updated position of free particle 2 in the kth generation; d _min is the lower limit of the global search area of the convergent particle, d _max is the upper limit of the global search area of the convergent particle, that is, the global search area of the convergent particle is [d _min , d _max ]; rand() is a random number in the interval [0,1], k is the particle iterative algebra, and the search radius r ₁ of the random search of free particles is determined by formula (8):

表示第1个收敛粒子在第k次更新即迭代后的位置，d₁ ^k-1表示第1个收敛粒子在第k-1次迭代后的位置；

表示第2个收敛粒子在第k次迭代后的位置，

表示第2个收敛粒子在第k-1次迭代后的位置；

表示第i个收敛粒子在第k次迭代后的位置，

表示第i个收敛粒子在第k-1次迭代后的位置；

表示第N_p个收敛粒子在第k次迭代后的位置，

表示第N_p个收敛粒子在第k-1次迭代后的位置。in,

Represents the position of the first convergent particle after the k-th update, that is, the iteration, d ₁ ^k-1 represents the position of the first convergent particle after the k-1th iteration;

represents the position of the second convergent particle after the kth iteration,

Indicates the position of the second convergent particle after the k-1th iteration;

represents the position of the i-th convergent particle after the k-th iteration,

represents the position of the ith convergent particle after the k-1th iteration;

represents the position of the N _pth convergent particle after the kth iteration,

represents the position of the N _pth convergent particle after the k-1th iteration.

Step 5. Determine whether the number of iterations, that is, the update algebra k satisfies k>gen _max , that is, determine whether the maximum number of iterations is reached, if so, end the iteration, and execute step 6, otherwise add 1 to the number of iterations k and return to step 2 to continue the iteration;

Step 6. Determine whether the current external environment has changed. If the current external environment has not changed, output the global optimal solution Dgbest, and use Dgbest as the duty cycle of the PWM control signal of the photovoltaic power generation system, otherwise return to step 1.

Example 2

Aiming at the problems of poor real-time performance and large disturbance during the restarting process of the existing restarting system, a hierarchical automatic restarting method of maximum power point tracking algorithm is proposed. The specific implementation steps are as follows:

The first step is to collect the output voltage U _s of the photovoltaic array in real time through the voltage sensor, and then calculate the rate of change ΔU of the output voltage of the photovoltaic array according to formula (9):

ΔU=|U _s −U _max |/U _max ; (9)

Wherein, U _max is the output voltage at the maximum power point collected after the photovoltaic array performed the maximum power point tracking algorithm last time. When the external environment changes, the output voltage of the photovoltaic array will inevitably change. Therefore, if the output voltage of the collected photovoltaic array changes, it means that the external environment has changed, otherwise the external environment has not changed.

The second step is to intelligently sense the degree of change of the external environment through the hierarchical method according to the change rate ΔU of the output voltage of the photovoltaic array, and reset the global search interval for the restart of the MPPT algorithm, that is, the maximum power point tracking algorithm, according to the different degrees of change of the external environment. Re-track the maximum power point of the PV system.

When a change in the external environment is detected, the MPPT technology needs to be restarted. When restarting, the change rate ΔU of the output voltage of the photovoltaic array is analyzed and processed, and the external environment can be extracted from the data of the change rate ΔU of the output voltage of the photovoltaic array. The degree of change, according to the degree of change of the external environment, reasonably formulate the global search interval, and finally re-track the maximum power point of the photovoltaic power generation system according to the MPPT technology based on the NPSO algorithm.

In the process of intelligently sensing the degree of change of the external environment by the hierarchical method, three thresholds of voltage change rate η ₁ , η ₂ and η ₃ are firstly set, and the voltage change rate thresholds η ₁ , η ₂ and η ₃ and the output voltage of the photovoltaic array are used. The magnitude of the change rate ΔU divides the degree of change of the external environment into three levels: when 0≤ΔU<η ₁ , the corresponding degree of change in the external environment is the first level; when η ₁ ≤ΔU<η ₂ , the corresponding external environment The change degree of the environment is the second level; when η ₂ ≤ΔU<η ₃ , the corresponding change degree of the external environment is the third level. The first level means that the external environment only changes slightly, and the output power of the photovoltaic power generation system also changes slightly. The maximum power point of the photovoltaic power generation system at this level does not move or moves to a very small extent. The first-level MPPT algorithm does not need to be restarted; the third-level indicates that the external environment has undergone great changes, and the maximum power point of the photovoltaic power generation system at this level has moved greatly. Therefore, when the MPPT algorithm restarts, it is necessary to define the global search as the maximum search The interval is [0,1]; the second level indicates that the external environment has undergone minor changes, and the maximum power point of the photovoltaic power generation system at this level has also moved accordingly. Therefore, when the MPPT algorithm is restarted, it needs to be based on The size of the change rate ΔU of the output voltage of the photovoltaic array can reasonably redefine the search interval. The specific method is to set the upper and lower limits of the search interval to be D _max and D _min respectively, that is, the search interval is [D _min , D _max ], where D _min ≥ 0 and D _max ≤ 1; the values of D _max and D _min Determined by formula (10) and formula (11):

D _max =Dgbest+r ₂ ; (10)

D _min =Dgbest-r ₂ ; (11)

Among them, Dgbest is the global optimal solution searched by the last execution of the MPPT algorithm, r ₂ is the search radius of the search with Dgbest as the center, and the value of r ₂ is determined according to the change rate ΔU of the output voltage of the photovoltaic array. See formula (12):

Wherein, both η ₂₁ and η ₂₂ are two voltage change rate sub-thresholds located between the voltage change rate thresholds η ₁ and η ₂ , and η ₁ <η ₂₁ <η ₂₂ <η ₂ . The specific values of η ₂₁ and η ₂₂ are determined according to the specific photovoltaic power generation system and can be obtained through on-site debugging. Here, under the premise of η ₁ ≤ΔU ≤ η ₂ , when ΔU is small, r ₂ takes 0.1, when ΔU is large, r ₂ takes 0.2, and when ΔU is large, r ₂ takes 0.3.

The maximum power tracking algorithm in this embodiment, that is, the MPPT algorithm, can either use the existing maximum power point tracking method, such as the maximum power point tracking method based on the PSO algorithm, or the maximum power point tracking method based on the NPSO algorithm proposed in Embodiment 2. Point tracking method.

Example 3

In photovoltaic power generation systems, a two-stage system structure is often used. The so-called two-stage system means that a first-level DC/DC (direct current/direct current) conversion link is added before the DC/AC (direct current/alternating current) conversion link. The addition of the DC/DC conversion link can control the grid-connected technology and the maximum power point tracking algorithm separately, making the control more convenient and flexible. In the embodiment of the present invention, the DC/DC link adopts a boost circuit, and the implementation of the maximum power point tracking algorithm is also performed in the boost circuit. The boost circuit of the photovoltaic system is shown in FIG. 3 . This embodiment proposes a maximum power point tracking method based on NPSO algorithm and hierarchical automatic restart, which is an algorithm for adding one level of automatic reset global search area to the previous stage of the original maximum power point tracking method based on NPSO algorithm. As shown in Figure 2, follow these steps:

收敛粒子在第一代的速度

收敛粒子的速度上限v_max、速度下限v_min，自由粒子在第一代位置

收敛粒子总数N_p，学习因子c₁和c₂，惯性权重因子ω，最大迭代次数gen_max，自由粒子在的搜索代数阈值gen，分等级重启的电压变化率阈值η₁、η₂和η₃以及电压变化率子阈值η₂₁和η₂₂，其中，0＜i＜N_p，i′＝1,2；Step S1, take the duty ratio D of the PWM control signal of the photovoltaic power generation system as the position of the population particle, and use the photovoltaic array output power P _pv of the photovoltaic power generation system as the fitness value of the population particle, and divide the population particle into a plurality of convergent particles and For two free particles, the global search interval for the position of the convergent particle is [0, 1], that is, the lower limit d _min of the global search interval for the position of the convergent particle is 0, and the upper limit d _max of the global search interval for the position of the convergent particle is 1, and the two Among the free particles, the search interval of free particle 1 is [0, 0.5], and the search interval of free particle 2 is [0.5, 1], and the particle population is initialized, that is, the position of the initial convergent particle in the first generation

Velocity of convergent particles in the first generation

The upper limit v _max and the lower limit v _min of the convergent particle, the free particle is at the position of the first generation

total number of convergent particles N _p , learning factors c ₁ and c ₂ , inertia weight factor ω, maximum number of iterations gen _max , search algebra threshold gen for free particles, voltage rate thresholds η ₁ , η ₂ and η ₃ for hierarchical restart and the voltage change rate sub-thresholds η ₂₁ and η ₂₂ , where 0<i<Np, i' ₌ 1,2;

Step S2, obtain the real-time output voltage V _pv and real-time output current I _pv of the photovoltaic array, and use formula (3) to calculate the i-th convergent particle in the k-th generation according to the real-time output voltage V _pv and real-time output current I _pv of the photovoltaic array. The updated fitness value P _i ^k and the updated fitness value of the i′-th free particle in the k-th generation

更新第i个收敛粒子的个体最优位置Dpbest_i和种群的全体最优位置Dgbest，若P_i ^k＞Ppbest_i，则令Ppbest_i＝P_i ^k、

否则Ppbest_i和Dpbest_i不变；若P_i ^k＞Pgbest，则令Pgbest＝P_i ^k、

否则Pgbest和Dgbest不变；若

则令

否则Pgbest和Dgbest不变；Step S3, according to the updated fitness value P _i ^k of the i-th convergent particle in the k-th generation and the fitness value of the i′-th free particle after the update of the k-th generation

Update the individual optimal position Dpbest _i of the i-th convergent particle and the overall optimal position Dgbest of the population. If P _i ^k >Ppbest _i , then let Ppbest _i =P _i ^k ,

Otherwise, Ppbest _i and Dpbest _i remain unchanged; if P _i ^k >Pgbest, then let Pgbest=P _i ^k ,

Otherwise, Pgbest and Dgbest remain unchanged; if

order

Otherwise Pgbest and Dgbest remain unchanged;

Step S4, update the velocity and position of the convergent particle according to formulas (4) and (5), and update the positions of the two free particles according to formulas (6) to (8);

Step S5, determine whether the number of iterations, that is, the update algebra k satisfies k>gen _max , that is, determine whether the maximum number of iterations is reached, if so, end the iteration, and execute step S6, otherwise add 1 to the number of iterations k and return to step S2 to continue the iteration;

Step S6, calculating the rate of change ΔU of the output voltage of the photovoltaic array by formula (9);

自由粒子2的搜索区间[0.5,1]替换为

若η₂≤ΔU＜η₃，直接返回步骤S1进行重启。Step S7, according to the rate of change ΔU of the output voltage of the photovoltaic array, intelligently perceive the degree of change of the external environment through a hierarchical method, and judge whether it is necessary to return to step S1 to restart according to the different degrees of change of the external environment, and when it is judged that it is necessary to return to step S1 to restart, for The degree of change of the external environment at different levels resets the global search interval, then returns to step S1 to restart, and replaces the global search interval of the convergent particles in the original step S1 with the reset global search interval. Specifically, if 0≤ΔU<η ₁ , there is no need to return to step S1 to restart, the global optimal position Dgbest is directly output, Dgbest is used as the duty cycle of the PWM control signal of the photovoltaic power generation system, and the process returns to step S6 to continue to calculate the output of the photovoltaic array The rate of change of voltage; if η ₁ ≤ΔU<η ₂ , return to step S1 to restart, and reset the global search area to [D _min , D _max ], that is, the global search interval [0,1 of the converged particle in step S1 ] is updated to [D _min , D _max ], that is, the lower limit d _min of the global search interval of the convergent particles is updated to D _min , and the upper limit d _max of the global search interval of the convergent particles is replaced by D _max . The search interval [0,0.5] of free particle 1 is replaced by

The search interval [0.5,1] of free particle 2 is replaced by

If η ₂ ≤ΔU<η ₃ , directly return to step S1 to restart.

Example 4

In order to verify the maximum power point tracking method based on the NPSO algorithm, the simulation model of the photovoltaic power generation system based on the boost circuit shown in Figure 3 is built in the Simulink environment, as shown in Figure 4. The PV array is a photovoltaic array with three photovoltaic cells connected in series, and the parameters of one photovoltaic cell are shown in Table 1, C1=10μF, L=1.5mH, C2=50μF, R1=53Ω.

Under the three illumination modes in Table 2, the simulation tests of the maximum power point tracking method based on the PSO algorithm and the NPSO algorithm are carried out respectively. The number of particles in both the PSO algorithm and the NPSO algorithm is 6, the maximum number of iterations is 20, and the parameter gen value of the NPSO algorithm is 12.

Fig. 5(a) and Fig. 5(b) are a simulation result diagram of the maximum power point tracking method based on the PSO algorithm and the NPSO algorithm under the illumination mode 3. From Figure 5(a) and Figure 5(b), it can be seen that the MPPT method based on the PSO algorithm before the improvement is trapped in the local power maximum, and under the same initial conditions of the particle swarm, the improved NPSO-based method The maximum power point tracking method of the algorithm jumped out of the local power maximum value at 0.33s due to the global search effect of free particles, and all particles converged to the power maximum value of 96.8W at 1.43s. The global search capability of the particle swarm algorithm is enhanced. When the photovoltaic array is blocked by local shadows, the success rate of the maximum power point tracking of the photovoltaic power generation system is effectively improved, the problem of falling into a local power maximum value is solved, and the power generation is improved. Efficiency and economy of photovoltaic power plants, reducing energy waste.

FIG. 6( a ) and FIG. 6( b ) are simulation diagrams of adopting the periodic scanning method and the hierarchical automatic restart method according to the embodiment of the present invention. The experimental conditions were set to change from illumination mode 3 to illumination mode 2 at 1.3 s, and the periodic scanning method performed a scan every 1.5 s. It can be seen from the simulation results that the periodic scanning method cannot immediately respond to restart the MPPT algorithm when the external environment changes, but must wait for the scanning time to restart the MPPT algorithm. On the other hand, the external environment did not change at 3s, but the scanning time of the periodic scanning method came, so he had to restart the MPPT algorithm, which was unnecessary scanning and reduced the power generation efficiency. In the hierarchical automatic restart method, when the external environment changes within 1.3s, the system can automatically detect the changes in the external environment and restart the MPPT algorithm immediately. In addition, in Figure 6(a), the grading method is not added during the restart of the MPPT algorithm, and in Figure 6(b), the grading method is added during the restart of the MPPT algorithm. By comparison, it can be seen that in Figure 6(b) The power fluctuation range of the MPPT algorithm in the restart process is roughly 110 to 150W, while the power fluctuation range of the MPPT algorithm restart process in Figure 6(a) is roughly 0 to 150W. The restart of the algorithm reduces power fluctuations, makes the photovoltaic power generation system more stable, increases the efficiency of photovoltaic power generation, and reduces energy waste.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims (9)

1. a maximum power point tracking method based on NPSO algorithm, is characterized in that, on the basis of PSO algorithm, with the duty ratio D of the PWM control signal of the photovoltaic power generation system as the position of the population particle, with the position of the population particle of the photovoltaic power generation system. The output power P _pv of the photovoltaic array is used as the fitness value of the population particles, and the population particles are divided into two types: convergent particles and free particles. The properties of convergent particles are consistent with those in the PSO algorithm, and free particles have no memory. The same generation update, the speed and position of the convergent particle are updated according to the speed and position update method of the particle in the PSO algorithm, the position of each free particle is randomly updated in the given search interval, and the search interval of the free particle is the convergence particle. The global search interval is divided into equal parts according to the number of free particles. After each update, the position of the convergent particle is used to update the individual optimal position of the convergent particle through the fitness value of the convergent particle and the free particle, and the position of the convergent particle and the free particle is used. Update the global optimal position of the population; when the convergent particle falls into the local optimal solution, that is, the local optimal position, all free particles continue to search in the global search interval of the convergent particle, and the position of a free particle is better than the current global position When the optimal solution is obtained, the position of the free particle is used to update the global optimal position, that is, the global optimal solution, and the convergent particle is pulled out of the local optimal solution, so that the convergent particle continues to search for the global optimal solution; Specific steps are as follows: Step 1. Take the duty ratio D of the PWM control signal of the photovoltaic power generation system as the position of the population particle, and use the photovoltaic array output power P _pv of the photovoltaic power generation system as the fitness value of the population particle, and divide the population particle into multiple convergent particles and For two free particles, the global search interval for the position of the convergent particle is [0, 1]. Among the two free particles, the search interval for free particle 1 is [0, 0.5], and the search interval for free particle 2 is [0.5, 1]. , and initialize the particle population; Step 2. Obtain the real-time output voltage V _pv and real-time output current I _pv of the photovoltaic array, _{and calculate} the fitness value P _i ^k and the The fitness value of i′ free particles in the kth generation

Among them, 0<i<N _p , i′=1,2, N _p is the total number of convergent particles; Step 3. According to the updated fitness value P _i ^k of the i-th convergent particle in the k-th generation and the fitness value of the i′-th free particle after the update of the k-th generation

Update the individual optimal position Dpbest _i of the i-th convergent particle and the overall optimal position Dgbest of the population: If P _i ^k >Ppbest _i , then let Ppbest _i =P _i ^k ,

Otherwise Ppbest _i and Dpbest _i remain unchanged; If P _i ^k >Pgbest, then let Pgbest=P _i ^k ,

Otherwise Pgbest and Dgbest remain unchanged; like

order

Otherwise Pgbest and Dgbest remain unchanged; in,

is the updated position of the i-th convergent particle in the k-th generation,

is the updated position of the ith free particle in the kth generation, Ppbest _i is the individual optimal fitness value of the ith convergent particle, and Pgbest is the overall optimal fitness value of the population; Step 4. Update the velocity and position of the convergent particle and the position of the two free particles; Step 5. Determine whether the number of iterations, that is, the update algebra k satisfies k>gen _max , that is, determine whether the number of iterations reaches the maximum number of iterations gen _max , if so, end the iteration, and execute step 6, otherwise add 1 to the number of iterations k and return to step 2 continue to iterate; Step 6: Determine whether the current external environment has changed, if the current external environment has not changed, output the global optimal solution Dgbest, otherwise return to step 1. 2. a kind of maximum power point tracking method based on NPSO algorithm according to claim 1, is characterized in that, described step 4 updates the speed and the position of convergent particle according to following formula (4)～(5):

in,

represents the velocity of the ith convergent particle in the population after the update of the kth generation,

Represents the velocity of the ith convergent particle in the population after the update of the k-1th generation, v _max is the upper limit of the velocity of the convergent particle, v _min is the lower limit of the velocity of the convergent particle, when calculated by formula (4)

When less than v _min ,

When calculated using formula (4)

When greater than v _max ,

Represents the position of the ith convergent particle in the population after the update of the k-1th generation, 1<k<gen _max ; ω is the inertia weight factor, rand() is a random number in the interval [0,1], c ₁ and _c2 is the learning factor. 3. a kind of maximum power point tracking method based on NPSO algorithm according to claim 1, is characterized in that, described step 4 updates the position of two free particles according to following formula (6)～(7):

in,

is the updated position of free particle 1 in the kth generation,

is the updated position of free particle 2 in the kth generation; d _min is the lower limit of the global search interval of the convergent particle, d _max is the upper limit of the global search interval of the convergent particle; rand() is the random number in the interval [0,1], gen is the search algebraic threshold of free particles. After the number of iterations is gen, the free particles no longer perform a global search, but take the global optimal solution Dgbest as the center, and conduct a random search of a tiny area with a radius of r _1. The free particles randomly search The search radius r ₁ of the search is determined by equation (8):

in,

Represents the position of the first convergent particle after the k-th update, that is, the iteration, d ₁ ^k-1 represents the position of the first convergent particle after the k-1th iteration;

represents the position of the second convergent particle after the kth iteration,

Indicates the position of the second convergent particle after the k-1th iteration;

represents the position of the i-th convergent particle after the k-th iteration,

represents the position of the ith convergent particle after the k-1th iteration;

represents the position of the N _pth convergent particle after the kth iteration,

represents the position of the N _pth convergent particle after the k-1th iteration. 4. a kind of maximum power point tracking method based on NPSO algorithm according to claim 1, is characterized in that, initializing particle population is the position of initializing convergent particle in the first generation

Velocity of convergent particles in the first generation

The upper limit v _max and the lower limit v _min of the convergent particle, the free particle is at the position of the first generation

The total number of converged particles N _p , the learning factors c ₁ and c ₂ , the inertia weight factor ω, the maximum number of iterations gen _max , and the search algebraic threshold gen for free particles. 5. A hierarchical automatic restart method of a maximum power point tracking algorithm, wherein the maximum power point tracking algorithm is the maximum power point tracking method based on the NPSO algorithm according to any one of claims 1 to 4, and the specific The implementation steps are as follows: The first step is to collect the output voltage U _s of the photovoltaic array in real time through the voltage sensor, and then calculate the rate of change ΔU of the output voltage of the photovoltaic array according to formula (9): ΔU=|U _s −U _max |/U _max ; (9) Among them, U _max is the output voltage at the maximum power point collected after the photovoltaic array performed the maximum power point tracking algorithm last time; The second step is to intelligently perceive the degree of change of the external environment through the hierarchical method according to the rate of change ΔU of the output voltage of the photovoltaic array, and reset the global search interval according to the different degrees of change of the external environment, and finally use the reset global search interval to restart the maximum The power point tracking algorithm re-tracks the maximum power point of the photovoltaic power generation system. 6 . The hierarchical automatic restart method of a maximum power point tracking algorithm according to claim 5 , wherein the degree of change of the external environment is intelligently sensed according to the rate of change ΔU of the output voltage of the photovoltaic array through a hierarchical method, 7 . And reset the global search interval according to the different degrees of change of the external environment. First, set different thresholds of voltage change rate, and divide the degree of change of the external environment into multiple thresholds according to different thresholds of voltage change rate and the change rate ΔU of the output voltage of the photovoltaic array. level, and then determine whether the MPPT algorithm needs to be restarted for the degree of change of the external environment at different levels, and when it is determined that the MPPT algorithm needs to be restarted, set different global search intervals for the degree of change of the external environment at different levels. 7. The hierarchical automatic restart method of a maximum power point tracking algorithm according to claim 5, characterized in that, according to the rate of change ΔU of the output voltage of the photovoltaic array, the degree of change of the external environment is intelligently sensed by the hierarchical method, and according to the change rate of the output voltage of the photovoltaic array ΔU The different degree of change of the external environment resets the global search interval. Finally, when restarting the MPPT algorithm using the reset global search interval, first set the three voltage change rate thresholds of η ₁ , η ₂ and η ₃ , and then use the voltage The rate of change thresholds η ₁ , η ₂ and η ₃ and the rate of change ΔU of the output voltage of the photovoltaic array divide the degree of change of the external environment into three levels: when _0≤ΔU <η1, the corresponding change in the external environment The degree is the first level, and the MPPT algorithm of the first level does not need to be restarted; when η ₁ ≤ΔU<η ₂ , the corresponding change degree of the external environment is the second level, and the global search interval is reset to [D _min , D _max ], where D _min ≥ 0 and D _max ≤ 1; when η ₂ ≤ΔU<η ₃ , the corresponding degree of change of the external environment is the third level, and the global search interval is set as the maximum search interval at this time, That is [0,1]; The values of D _max and D _min in the global search interval [D _min , D _max ] are determined by the following formulas (10) and (11): D _max =Dgbest+r ₂ ; (10) D _min =Dgbest-r ₂ ; (11) Among them, Dgbest is the global optimal solution searched by the last execution of the maximum power point tracking algorithm, r ₂ is the search radius centered on Dgbest, and the value of r ₂ is determined according to the change rate ΔU of the output voltage of the photovoltaic array. Determined, see formula (12) for details:

Wherein, both η ₂₁ and η ₂₂ are two voltage change rate sub-thresholds located between the voltage change rate thresholds η ₁ and η ₂ , and η ₁ <η ₂₁ <η ₂₂ <η ₂ . 8. A maximum power point tracking method based on NPSO algorithm and hierarchical automatic restart, it is characterized in that, after a kind of maximum power point tracking method based on NPSO algorithm described in any one of claims 1 to 4, an additional method is added. The hierarchical automatic restart method of a maximum power point tracking algorithm according to claim 5, its concrete steps are as follows: Step S1, take the duty ratio D of the PWM control signal of the photovoltaic power generation system as the position of the population particle, and use the photovoltaic array output power P _pv of the photovoltaic power generation system as the fitness value of the population particle, and divide the population particle into a plurality of convergent particles and For two free particles, the global search interval for the position of the convergent particle is [0, 1]. Among the two free particles, the search interval for free particle 1 is [0, 0.5], and the search interval for free particle 2 is [0.5, 1]. , and initialize the particle population; Step S2, obtain the real-time output voltage V _pv and real-time output current I _pv of the photovoltaic array, and calculate the adaptive value of the i-th convergent particle after the k-th generation update according to the real-time output voltage V _pv and real-time output current I _pv of the photovoltaic array The fitness value of P _i ^k and the i′-th free particle after the update of the k-th generation

Among them, 0<i<N _p , i′=1,2, N _p is the total number of convergent particles; Step S3, according to the updated fitness value P _i ^k of the i-th convergent particle in the k-th generation and the fitness value of the i′-th free particle after the update of the k-th generation

Update the individual optimal position Dpbest _i of the i-th convergent particle and the overall optimal position Dgbest of the population. If P _i ^k >Ppbest _i , then let Ppbest _i =P _i ^k ,

Otherwise, Ppbest _i and Dpbest _i remain unchanged; if P _i ^k >Pgbest, then let Pgbest=P _i ^k ,

Otherwise, Pgbest and Dgbest remain unchanged; if

order

Otherwise Pgbest and Dgbest remain unchanged, where,

is the updated position of the i-th convergent particle in the k-th generation,

is the updated position of the ith free particle in the kth generation, Ppbest _i is the individual optimal fitness value of the ith convergent particle, and Pgbest is the overall optimal fitness value of the population; Step S4, update the velocity and position of the convergent particle, and the position of the two free particles; Step S5, determine whether the number of iterations, that is, the update algebra k satisfies k>gen _max , that is, determine whether the number of iterations reaches the maximum number of iterations gen _max , if so, end the iteration, and execute step S6, otherwise add 1 to the number of iterations k and return to step S2 continue to iterate; Step S6, calculating the rate of change ΔU of the output voltage of the photovoltaic array; Step S7, according to the rate of change ΔU of the output voltage of the photovoltaic array, intelligently perceive the degree of change of the external environment through a hierarchical method, and judge whether it is necessary to return to step S1 to restart according to the different degrees of change of the external environment, and when it is judged that it is necessary to return to step S1 to restart, for The degree of change of the external environment at different levels resets the global search interval, then returns to step S1 to restart, and replaces the global search interval of the convergent particles in the original step S1 with the reset global search interval. 9. a kind of maximum power point tracking method based on NPSO algorithm and grading automatic restart according to claim 8, is characterized in that, in described step S1, initialize particle population, namely initialize the position of convergent particle in the first generation

Velocity of convergent particles in the first generation

The upper limit v _max and the lower limit v _min of the convergent particle, the free particle is at the position of the first generation

total number of convergent particles N _p , learning factors c ₁ and c ₂ , inertia weight factor ω, maximum number of iterations gen _max , search algebra threshold gen for free particles, voltage rate thresholds η ₁ , η ₂ and η ₃ for hierarchical restart and the voltage rate of change sub-thresholds n ₂₁ and n ₂₂ ; In the step S7, if 0≤ΔU<η ₁ , then there is no need to return to step S1 to restart, directly output the global optimal position Dgbest, and return to step S6 to continue to calculate the rate of change of the output voltage of the photovoltaic array; if η ₁ ≤ΔU<η _2. Return to step S1 to restart, and reset the global search area to [D _min , D _max ], that is, replace the global search interval [0,1] of the convergent particle in step S1 with [D _min , D _max ], free The search interval [0,0.5] of particle 1 is replaced by

The search interval [0.5,1] of free particle 2 is replaced by

If η ₂ ≤ΔU<η ₃ , directly return to step S1 to restart.

Images