基于L-SVM的太阳能短期上网发电量预测方法研究

对光伏上网发电量进行短期预测,可以为电力部门的调度以及用电计划的调整提供参考.提出了一种基于最小二乘支持向量机(least square support vector machines,LS-SVM)对短期光伏上网发电量的预测方法,LSSVM方法具有好的泛化能力.以甘肃某地区电厂的并网发电全年实测数据为实例,同时考虑到短期太阳辐射和光伏电池温度对光伏发电量的影响,建立了基于LS-SVM的短期预测模型.与现有的前向神经网络预测方法进行比较,实验结果表明,该方法能获得更好的预测效果,具有一定的应用潜力.

Prediction Method of Short-Term Electricity of Solar Power Generation Based on LS-SVM

Prediction of short-term photovoltaic electricity can provide reference for electric power dispatching department to dispatch and adj ust the power plan.The prediction method of short-term photovoltaic electricity is presented based on Least Square Support Vector Machines （LS-SVM）with good generalization ability.Taking the annual measured data of grid-connected photo-voltaic power plant as example and considering the influence of short-term solar radiation and temperature of photovoltaic cells on the photovoltaic generation,we establish the short-term pre-diction model based on LS-SVM.This method is compared with the existing forward neural net-work prediction method.The experimental results show that this method can obtain the good pre-diction performance and has great potential application.