(11-3-4)基于共轭梯度神经网络算法的风电功率预测的优化.docVIP

基于共轭梯度人工神经网络技术：一种迅速、精确的风功率预测方法 李天1，李永倩1，廖明伟1，王炜康1，曾楚杰1 （1 071000） A Fast and Accurate Wind Power Forecasting Approach Based on Conjugated Gradient Neural Network 1 Li Tian1, Li Yong-qian1, Liao Ming-wei1, Wang Wei-kang1, Zeng Chu-jie1 (1.North China Electric Power University, Baoding 071000, Hebei China)ABSTRACT: Prediction of the output power of wind plants has great significance for running a power system connected with large amount of wind generators. Depending on the prediction results, the power plants are able to allocate power generating quotas and distribute resources in scientific and reasonable way. In the past, the Grey Neural Network has been widely applied in predicting wind power. However, this method could hardly satisfy the engineering requirements due to the structure of ANN. The problem of slow convergence speed and large amount of iterations, especially when faced with large scale data, would pose challenges to the power prediction and the sensitivity of automatic control. This paper optimizes ANN model by applying conjugate gradient descent in weights updating process. Experiments performed on different scale datasets proved that the performance of CGNN improves substantially. The average iterations decreases by almost 90% without the sacrifice of prediction accuracy. KEY WORDS: Wind Power Prediction, Conjugated Gradient Decent, Artificial Neural Network, Gray Theory 摘要: 风电场输出功率预测对接入大量风电的电力系统运行有重要意义，通过风功率预测，电厂可合理分配发电配额，调度发电资源。灰色神经网络在预测风功率中得到，但由于神经网络自身的性质，该方法预测结果本文共轭梯度优化，表明，共轭梯度算法灰色神经网络有很大90%，在保证风电功率预测的精确度的情况下，预测速度大幅提高。 关键词: 风功率预测；共轭梯度；神经网络；灰色理论 0 引言 在全球能源危机环境的情况下，发展清洁能源是我国乃至全世界所必须的必经之路着科技的发展的降低，风能全世界发展迅速的清洁能源之一。预测将风电并入电网中着重要的意义。的国家都对风能预测系统进行和研究。是一个风能的，大约占据全球的26.3%[1]。对风预测分为两方面，一方面是因素，湿度等）变化预测功率的变化，另外方法是直接通过历史数据来预测未来的风功率对于风发电的课题已经做了相当多的研究工作，了一些关于风电场风速和功率的预测方法，其中包括持续法2]、卡尔曼波法[3]时间序列法[4-6]、人工神经网络法[7-11]、模糊逻辑法[12]、灰色预测法[13-14] 、专家系统、小波分析理论[19,23]、支持向量机[15,20]、最近临搜索、功率观测器、空间相关性法[15,17]等常用的风速预测方法。 [3]。随机时间序列法利用大量的历史数据来建模，经过模型识别、参数估计、模型检验来确定能够描述所研究时