基于改进SSA优化BP神经网络的光伏发电量预测

Photovoltaic Power Generation Prediction Based on Improved SSA Optimized BP Neural Network

在分析影响光伏发电因素基础上,针对传统基于BP神经网络光伏发电预测存在问题,提出了一种改进麻雀搜索算法优化BP神经网络的光伏发电预测模型。首先引入T分布变异略优化麻雀搜索算法。然后采用TSSA对BP神经网络的权值和阈值进行优化,克服了BP神经网络中存在的问题传统学习算法,容易陷入局部最优解,收敛速度慢的缺点。利用TSSABP和SSA-BP预测模型与传统BP预测模型进行了对比实验,比较功率实际输出数据。结果表明TSSA-BP预测模型能更好准确地预测功率输出,具有潜在的实用价值。

Based on the analysis of the factors affecting photovoltaic power generation,aiming at the problems existing in traditional photovoltaic power generation prediction based on BP neural network,aphotovoltaic power generation prediction model optimized by improved Sparrow search algorithm and BP neural network was proposed.Firstly,T-distribution variation is introduced to slightly optimize Sparrow search algorithm.Then TSSA is used to optimize the weight and threshold value of BP neural network,which overcomes the shortcomings of traditional learning algorithm in BP neural network,such as easy to fall into local optimal solution and slow convergence speed.In this paper,TSSABP and SSA-BP prediction models are compared with traditional BP prediction models,and the actual power output data are compared.The results show that TSSA-BP prediction model can predict power output more accurately and has potential practical value.