光伏发电系统改进型最大功率跟踪算法的研究与应用

发布时间：2019-01-24 21:53

【摘要】：能源不仅是当今社会发展的基础,也是世界各国经济建设的动力。太阳能特有的零污染和取之不尽的特点使之成为世界公认的最理想的清洁能源,而太阳能被用来发电是目前最主要的用途之一。开展对光伏发电系统的研究不仅对能源的可持续利用、环境保护提供了有效解决方案,而且对社会的稳定发展具有重要的战略意义。利用最大功率跟踪控制技术是提高光伏发电系统效率的有效方式,对光伏发电行业有重要影响,同时也是这一领域的研究热点。本文以光伏发电系统为研究对象,针对最大功率跟踪问题展开研究,主要内容有:1.根据光伏电池的等效电路推导了其数学模型,利用Matlab/Simulink搭建光伏电池模型以及整个光伏发电系统的仿真平台。利用搭建的仿真平台对光伏电池的V-I输出特性进行仿真研究,并对传统的占空比扰动观察法,进行了分析与仿真研究。2.本文对利用BP神经网络进行最大功率点跟踪时,训练数据存在测量误差的情况进行了分析,指出了基于最小二乘的BP神经网络(LS-NN) MPPT算法预测结果的精准度严重依赖训练数据的准确性,提出了基于准最小二乘的BP神经网络(QLS-NN) MPPT算法。通过仿真分析与实验测试对比了两种不同算法的预测结果。3.本文对利用阻抗匹配原理进行最大功率跟踪的方法也做了重点研究。通过对传统阻抗匹配MPPT算法的理论分析,发现传统阻抗匹配MPPT算法对系统参数设置比较敏感,提出了改进型动态阻抗匹配MPPT算法。并通过仿真和实验对比,本文提出的改进型动态阻抗匹配MPPT算法能够有效的改善光伏发电系统的性能。本研究采用基于准最小二乘的BP神经网络MPPT算法减小了含有测量误差的训练样本对利用神经网进行预测最大功率的影响,提高了系统的鲁棒性;基于改进型动态阻抗匹配的MPPT算法针对传统阻抗匹配MPPT算法能够有效的解决系统参数设定较多以及输出功率受负载影响较大的问题,为提高光伏发电系统的整体效率和稳定性提供了参考依据。
[Abstract]:Energy is not only the foundation of social development, but also the driving force of economic construction. The unique zero pollution and inexhaustible characteristics of solar energy make it the most ideal clean energy in the world, and solar energy is used to generate electricity is one of the most important uses. The research on photovoltaic power generation system not only provides an effective solution for the sustainable use of energy and environmental protection, but also has an important strategic significance for the stable development of society. The use of maximum power tracking control technology is an effective way to improve the efficiency of photovoltaic power generation system, which has an important impact on photovoltaic power generation industry, and is also a research hotspot in this field. In this paper, photovoltaic power generation system as the research object, aiming at the maximum power tracking problem, the main contents are: 1. According to the equivalent circuit of photovoltaic cell, the mathematical model of photovoltaic cell is deduced, and the model of photovoltaic cell and the simulation platform of the whole photovoltaic system are built by using Matlab/Simulink. The V-I output characteristics of photovoltaic cells are simulated by using the built simulation platform, and the traditional duty cycle disturbance observation method is analyzed and simulated. 2. In this paper, the measurement error of the training data in maximum power point tracking using BP neural network is analyzed. It is pointed out that the accuracy of BP neural network (LS-NN) MPPT algorithm based on least squares depends heavily on the accuracy of training data, and a BP neural network (QLS-NN) MPPT algorithm based on quasi least squares neural network is proposed. The prediction results of two different algorithms are compared by simulation analysis and experimental test. 3. 3. The method of maximum power tracking based on impedance matching principle is also studied in this paper. Based on the theoretical analysis of the traditional impedance matching MPPT algorithm, it is found that the traditional impedance matching MPPT algorithm is sensitive to the system parameters, and an improved dynamic impedance matching MPPT algorithm is proposed. The improved dynamic impedance matching MPPT algorithm proposed in this paper can effectively improve the performance of photovoltaic power generation system through the comparison of simulation and experiment. In this study, the BP neural network MPPT algorithm based on quasi least squares is used to reduce the influence of training samples with measurement error on the prediction of maximum power using neural network, and the robustness of the system is improved. The MPPT algorithm based on improved dynamic impedance matching can effectively solve the problem that the system parameters are set more and the output power is greatly affected by the load for the traditional impedance matching MPPT algorithm. It provides a reference for improving the overall efficiency and stability of photovoltaic power generation system.
【学位授予单位】：温州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM615;TP183

【参考文献】