基于最小二乘支持向量机的风电功率短期预测研究

发布时间：2018-05-25 02:08

本文选题：风电功率预测 + 最小二乘支持向量机　；参考：《华中科技大学》2014年硕士论文

【摘要】：由于风能的波动性和间歇性，导致风力发电的输出功率的波动性和不稳定性，对风电并网和电力系统调度造成了很大的冲击，严重阻碍了大规模风电接入电力系统。为了在保障电网安全稳定运行的前提下尽可能多的接纳风电，对风电场输出功率做出准确的预测就变得非常重要。本文采用最小二乘支持向量机来建立风电功率的短期预测模型。由于最小二乘支持向量机的核函数和超参数对预测模型的性能影响很大，本文分别基于不同核函数对建立最小二乘支持向量机预测模型，选出了一种最优核函数；采用引力搜索算法对超参数进行选择，建立GSA-LSSVM风电功率预测模型，将结果与支持向量机模型和BP神经网络预测结果比较，结果表明采用ERBF核函数和GSA参数优化的最小二乘支持向量机模型的预测精度更高，表明这种模型是一种更可靠的风电功率短期预测模型。常规的预测问题只能做出确定性的点预测，但是风电功率的预测存在一定水平的误差。为了确定某预测值出现的概率，考虑对风电功率进行不确定性的概率预测，构建每个预测点的置信预测区间。本文采用非参数估计方法进行风电功率的短期区间预测，基于LSSVM回归模型的线性光滑属性，计算预测点的方差和偏差估计值，，来构建不同置信度下的风电场短期预测功率的置信区间。基于上述风电功率的短期预测的单点预测和区间预测理论，本文基于Matlab平台进行风电功率预测算法进行仿真，为方便预测算法的可视化操作，采用Matlab用户图形界面设计实现了风电功率短期预测系统。本系统基于Matlab界面设计平台GUIDE，实现了风电功率的单点预测和区间预测，设计实现了简便美观的用户界面以及菜单栏，实现了数据导入、模型选择等模块，并图形展示了预测结果和预测误差值。
[Abstract]:Because of the volatility and intermittency of wind energy, the output power of wind power is fluctuating and unstable, which has a great impact on wind power grid connection and power system scheduling, and seriously hinders large-scale wind power access to power system. In order to accept wind power as much as possible under the premise of ensuring the safe and stable operation of power grid, it is very important to make accurate prediction of the output power of wind farm. In this paper, the least square support vector machine (LS-SVM) is used to establish the short-term prediction model of wind power. Because the kernel function and superparameter of least square support vector machine (LS-SVM) have great influence on the performance of prediction model, this paper establishes a prediction model of LS-SVM based on different kernel function pairs, and selects an optimal kernel function. The super parameters are selected by gravity search algorithm, and the GSA-LSSVM wind power prediction model is established. The results are compared with those of support vector machine model and BP neural network. The results show that the least squares support vector machine model with ERBF kernel function and GSA parameter optimization is more accurate, which indicates that this model is a more reliable short-term wind power prediction model. The conventional prediction problem can only make deterministic point prediction, but there is a certain level of error in the prediction of wind power. In order to determine the probability of the occurrence of a certain prediction value, the probabilistic prediction of wind power uncertainty is considered, and the confidence prediction interval of each prediction point is constructed. In this paper, the method of nonparametric estimation is used to predict the short-term interval of wind power. Based on the linear smooth property of LSSVM regression model, the variance and deviation estimate of the prediction point are calculated. To construct the confidence interval of short-term predictive power of wind farm under different confidence levels. Based on the theory of single-point prediction and interval prediction for short-term wind power prediction, this paper simulates the wind power prediction algorithm based on Matlab platform, in order to facilitate the visualization operation of the prediction algorithm. The short-term wind power prediction system is designed and implemented with Matlab user interface. Based on the design platform of Matlab interface, the system realizes the single point prediction and interval prediction of wind power, designs and implements a simple and beautiful user interface and menu bar, and realizes the modules of data import, model selection and so on. The result and error of prediction are shown in figure.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM614

【相似文献】