基于GA-PSO-BP与灰色关联的光伏短期功率预测

Short-term PV Power Prediction Based on GA-PSO-BP and Grey Correlation

光伏功率预测本身是一个较为复杂的非线性问题,为此首先介绍了在一般情况下影响光伏发电功率预测的客观因素,然后根据客观影响因素进行灰色关联度分析,把预测日作为参考,找到满足条件的日期作为训练样本。针对反向传播(BP)神经网络易于陷入局部极值的问题,利用粒子群优化(PSO)算法与遗传算法(GA)融合寻优,将粒子群中最优粒子进行排序,对高适应度粒子进行重组,反之进行变异;同时引入莱维飞行与随机游动策略增强算法跳出局部最优的能力,加入自适应权重因子平衡算法的探索与寻优能力。仿真实例证明,在不同的天气条件下,GA-PSO-BP模型与BP、PSO-BP、极限学习机(ELM)模型相比,预测效果更好,预测误差更小。

Photovoltaic(PV)power generation forecast is a more complex nonlinear problem,so this paper first introduces the objective factors that affect PV power generation forecast under normal circumstances,and then conducts grey correlation analysis according to the objective factors,takes the forecast date as a reference,and searches for the date meeting the conditions as a training sample.In order to solve the problem that BP neural network is easy to fall into local extremum,particle swarm(PSO)and genetic algorithm(GA)are used to search for optimization.The optimal particles in the particle swarm are sorted,the high-fitness particles are recombined,and vice versa.At the same time,Levy flight and random walk strategies are introduced to enhance the ability of the algorithm to jump out of the local optimal,and the exploration and optimization ability of the adaptive weight factor balance algorithm is added.Simulation examples show that GA-PSO-BP model has better prediction effect and smaller error than BP,PSO-BP and ELM model under different weather conditions,and can effectively avoid prematurity.