智能水滴算法优化SVM的光伏最大功率跟踪研究

发布时间：2019-07-09 08:19

【摘要】：对化石能源的不断开采和使用,对生态环境造成了严重的影响,为了更好地保护地球的生态环境,利用绿色能源和可再生能源以实现可持续发展迫在眉睫。太阳能是一种备受人们亲睐的清洁能源,有着广阔的发展前景。光伏发电作为目前开发太阳能的主要方式,具有安全可靠、应用形式灵活、安装维护简单等优势。但是光伏发电存在电能转化率低、成本高等问题,目前广泛使用的最大功率跟踪控制算法(MPPT),是提高其转换效率的重要手段。首先介绍了光伏发电输出预测的研究背景、意义以及预测中所采用的方法和技术。然后分析了光伏电池输出特性,引出MPPT的必要性和重要性。接着介绍了一种新兴群智能算法-智能水滴算法(Intelligent Water Drops,IWD),该算法最初是针对离散优化问题提出来的,而对SVM参数的优化是连续优化问题,同时传统智能水滴算法在结点的选择上易出现停滞和阻塞现象。针对以上问题对传统智能水滴算法提出改进,然后利用传统标准算法优化测试函数对改进的智能水滴进行测试,并将结果同传统蚁群算法、标准智能水滴算法进行对比,实验结果证明,本文提出的改进智能算法用于函数优化问题,具有良好的收敛性和准确性。其次介绍了支持向量机算法(SVM)基本原理,提出利用智能水滴算法优化支持向量机参数。然后利用UCI网站提供的标准测试集进行相应性能测试,并将实验结果与传统优化算法寻优结果进行比较,实验结果证明利用改进的智能水滴算法优化支持向量机能获得更高的回归预测精度。最后以某光伏电站为研究对象,利用光伏电站历史数据进行实验,对遗传算法、粒子群算法和智能水滴算法的参数寻优结果进行了对比分析,给出了在matlab/simulink环境下搭建的基于IWD-SVM的光伏MPPT的仿真模型。结果表明了本文给出的利用智能水滴优化SVM模型在光伏系统最大功率跟踪中的可行性。
文内图片：

图片说明：光伏电池仿真模型
[Abstract]:The continuous exploitation and use of fossil energy has a serious impact on the ecological environment. In order to better protect the ecological environment of the earth, it is urgent to use green energy and renewable energy to achieve sustainable development. Solar energy is a kind of clean energy which is favored by people and has a broad development prospect. Photovoltaic power generation, as the main way to develop solar energy, has the advantages of safety and reliability, flexible application form, simple installation and maintenance, and so on. However, there are some problems in photovoltaic power generation, such as low power conversion rate and high cost. The maximum power tracking control algorithm (MPPT), which is widely used at present, is an important means to improve its conversion efficiency. Firstly, the research background, significance and methods and techniques of photovoltaic power generation output prediction are introduced. Then the output characteristics of photovoltaic cells are analyzed, and the necessity and importance of MPPT are introduced. Then, a new swarm intelligent algorithm, intelligent water drop algorithm (Intelligent Water Drops,IWD), is introduced, which is originally proposed for discrete optimization problem, and the optimization of SVM parameters is a continuous optimization problem. At the same time, the traditional intelligent water drop algorithm is prone to stagnation and blocking in node selection. In view of the above problems, the traditional intelligent water drop algorithm is improved, and then the improved intelligent water droplet algorithm is tested by using the traditional standard algorithm to optimize the test function, and the results are compared with the traditional ant colony algorithm and the standard intelligent water drop algorithm. The experimental results show that the improved intelligent algorithm proposed in this paper has good convergence and accuracy in the function optimization problem. Secondly, the basic principle of support vector machine (SVM) algorithm is introduced, and the intelligent water drop algorithm is proposed to optimize the parameters of support vector machine (SVM). Then the corresponding performance test is carried out by using the standard test set provided by UCI website, and the experimental results are compared with those of the traditional optimization algorithm. The experimental results show that the improved intelligent water drop algorithm is used to optimize the support vector function to obtain higher regression prediction accuracy. Finally, taking a photovoltaic power station as the research object, the parameter optimization results of genetic algorithm, particle swarm optimization algorithm and intelligent water drop algorithm are compared and analyzed by using the historical data of photovoltaic power station, and the simulation model of photovoltaic MPPT based on IWD-SVM in matlab/simulink environment is given. The results show the feasibility of using intelligent water droplets to optimize SVM model in maximum power tracking of photovoltaic system.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18;TM615

【参考文献】