【作者】 张鹏；

【导师】 随慧斌； 郑天茹；

【作者基本信息】 山东大学 ， 电气工程（专业学位）， 2021， 硕士

【摘要】 在21世纪的今天,能源问题仍是一个至关重要的问题,特别是在新冠疫情大流行引发的前所未有的全球健康危机和经济危机的背景下,如果继续大量使用传统的化石能源必将对环境、社会造成不可弥补的伤害。在目前存在的能源发电种类中,光伏发电在选址建设和污染物排放上相比于传统化石能源具有明显的优势,且在沙漠、山区以及极地等环境恶劣的地区能够充分使用太阳能会对当地居民带来极大的方便。凭借其所具有的优势和潜力,各国学者在光伏发电领域投入了越来越多的时间和精力进行了深入研究。目前光伏发电技术在发电方式和原理上还有许多关键技术问题值得深入研究,其中一个就是最大功率点跟踪技术的研究。从光伏电池的输出P-V特性曲线可知,输出功率有一个极大值,进行最大功率点跟踪可以使光伏电池工作在功率峰值点处,提高工作效率。传统的最大功率点跟踪方法可以找到这个极值点从而输出最大功率,本文结合了相关的传统算法并进行了改进,提出了一种基于恒定电压法启动结合K分法的改进变步长扰动观察算法。该算法以一种恒定电压启动,然后判断扰动前后两次的dP/dU符号,如果同号则步长不变,反之步长除以K(K值为大于1的数),距离最大功率点较远处采用大步长,较近时采用小步长。通过不断改变扰动步长使输出越来越接近功率最大值,最后保持恒定。通过仿真验证,证明了该算法K的取值有一个最佳范围,在该范围内比传统的定步长扰动观察法和电导增量法跟踪速度更快、精度更高,且在光照强度和温度变化的情况下,K分法也能快速、准确地找到新的功率最大值。单个光伏电池的输出功率较低,在实际电力构架中往往不能满足电压等级要求,因此为了获得更高的输出电压和功率,采用光伏阵列。但是光伏阵列在局部遮阴的情况下其输出P-V特性曲线会有多个极值点,传统算法在此情形下容易陷入局部极值点从而无法输出最大功率。为了解决传统算法在光伏阵列局部遮阴情况下无法准确找到功率最值点的问题,本文选用了智能算法中的差分进化算法作为局部遮阴情况下光伏阵列的寻优算法,并对标准差分进化算法进行了改进,如粒子初始位置的设定、根据适应度确定的最优变异策略及缩放因子的自适应控制、交叉因子的自适应控制、算法早熟控制及结束和重启的条件等方面。并且在Matlab/Simulink仿真平台上进行了仿真实验,设置了三种光伏阵列的遮阴情况,首先将改进差分进化算法与标准差分进化算法和粒子群算法进行了比较,其次改进差分进化算法又与扰动观察法进行了比较。从光伏阵列的输出情况来看,在三种遮阴情况下,改进差分进化算法比标准差分进化算法和粒子群算法具有更高的收敛速度和精度,功率损失小,跟踪效率高,且扰动观察法在遮阴情况2时陷入了局部极值点,但是改进差分进化算法准确无误的找到了功率最大值点。当外部环境发生变化时,如光照强度和温度发生突变时,改进差分进化算法也能有效、快速地追踪到新的最大功率点。更多还原

【Abstract】 In the 21st century,energy is still a crucial issue.Especially in the background of the hitherto unknown global health and economic crisis triggered by the COVID-19 pandemic,if we continue to use a lot of traditional fossil energy,it will cause irreparable harm to the environment and society.In many new energy power generation,solar energy has obvious environmental advantages and great development potential,and making good use of solar energy in remote areas will certainly bring great convenience.Therefore,in the field of photovoltaic power generation,scholars from all over the world have put more and more energy into in-depth research.Photovoltaic power generation has many key technical problems to be studied,one of which is the research of maximum power point tracking(MPPT)technology.The output characteristic curve of photovoltaic cell has a maximum value.The MPPT can make photovoltaic cells work at the peak power point and improve work efficiency.The traditional MPPT methods can find this extreme point and output the maximum power.In this paper,an improved variable step size perturbation and observation MPPT algorithm is proposed based on the constant voltage method and the K division method.The algorithm starts with a constant voltage,and then judges the dP/dU symbol before and after the perturbation twice.If the sign is same,the step size remains unchanged;otherwise,the step size is divided by K(the value of K is greater than 1).In the distance from the maximum power point,the large step length is used,and in the near,the small step length is used.By constantly changing the perturbation step size,the output is closer and closer to the maximum power value,and finally remains constant.Through simulation,it is proved that the algorithm has an optimal range of K value.In this range,the tracking speed is faster and the accuracy is higher than the traditional perturbation observation method with fixed step size and incremental conductance method.Moreover,under condition of light intensity and temperature change,the new maximum power point can also be found quickly and accurately by the K division method.Because the output power of a single photovoltaic cell often can not meet the requirements,therefore,photovoltaic cells need to be connected in series and parallel to form photovoltaic array.In the case of partial shading,the output characteristic curve of photovoltaic array will have multiple extreme points,and the traditional algorithms are easy to fall into local extreme points and cause power loss.In order to solve this problem,this paper selects differential evolution(DE)algorithm as the optimization algorithm of photovoltaic array in the case of partial shading,and improved the standard DE algorithm,such as the setting of the initial position of particles,the optimal mutation strategy determined according to fitness and the adaptive control of scaling factor,the adaptive control of crossover factor,the premature control of algorithm and conditions for termination and restart and so on.Experiments were carried out on the Matlab/Simulink simulation platform,and three shading conditions of photovoltaic array were set.Firstly,the improved DE algorithm is compared with standard DE algorithm and particle swarm optimization algorithm.Secondly,the improved DE algorithm is compared with perturbation and observation algorithm.From the output of photovoltaic array,under three kinds of shading conditions,the improved DE algorithm has higher convergence speed,less power loss and high tracking efficiency than the standard DE algorithm and particle swarm optimization algorithm.The perturbation and observation method falls into local extreme value in the case of shading 2,but the improved DE algorithm can find the maximum power point accurately.When the external environment changes,such as light intensity and temperature suddenly change,the improved DE algorithm can also track the new maximum power point effectively and quickly.更多还原