基于模块化回声状态神经网络光伏发电量预测

针对光伏发电的不确定性、间歇性给电力系统并网运行带来的安全问题,提出了一种基于模块化回声状态网络模型对发电量进行预测.首先利用模块化神经网络按季节建立预测子模型,再将子模型按相同日类型进行数据划分后,与平均气温一同作为样本,利用回声状态网络对子模型进行训练和发电量预测,最后集成输出结果.结果表明:此预测模型在日类型相同时预测误差较小,而在日类型不同时预测误差较大,但与ESN和BP预测模型相比均具有更高的预测精度和更快的预测速度.

Forecasting of Photovoltaic Power Generation Based on Modular-Echo State Network

In order to solve the security problems of the power system interconnection caused by uncertainty and intermittent of photovoltaic power generation,a model based on modular echo state network is proposed to forecast power generation.Firstly,the forecast sub-model was established according to the seasons by using modularized neural networks.Then,the sub-models were divided based on similar days from history data of photovoltaic power generation and together with the average temperature as samples to train the sub-model and forecast the power generation by the echo state network.Finally,the result was integrated output.The results show that this forecasting model has a small forecasting error when the day type is the same,while the prediction error is larger when the day type is different.However,compared with the ESN and BP prediction models,the forecasting model has higher forecasting precision and faster forecasting speed.