弹性自适应人工鱼群-BP神经网络模型及在短期电价预测中的应用

Short-term electricity price prediction based on resilient adaptive artificial fish school algorithm and back propagation neural network hybrid algorithm

为了提高短期电价预测精度,分析了人工鱼群算法及其缺点,提出了一种弹性自适应人工鱼群算法(RAAFSA)。应用RAAFSA算法训练BP神经网络,实现了BP神经网络参数优化,形成弹性自适应人工鱼群-BP神经网络混合算法(RAAFSA-BP),对贵州电网进行短期电价预测。仿真表明,弹性自适应人工鱼群优化的BP神经网络算法收敛速度快于BP神经网络算法和人工鱼群-BP神经网络算法,该混合算法克服了BP神经网络和人工鱼群算法易陷于局部极值、搜索质量差和精度不高的缺点,改善了BP神经网络的泛化能力,输出稳定性好,预报精度显著提高,各日预测电价的平均百分比误差可控制在2%以内,平均绝对误差最大值为1.762$/MWh。该混合算法可有效用于电力市场短期电价预测。

Abstract： To improve the accuracy of short-term price forecast of power market, this paper analyzes the shortcomings of the traditional artificial fish school algorithm （AFSA）, and a resilient adaptive AFSA （RAAFSA） is put forth. The forecast model adopts a hybrid algorithm RAAFSA- BP that is a combination of RAAFSA with back propagation （BP） neural network, or neural network trained by RAAFSA. This model can be used for short-term price forecast and the model parameters are automatically determined from sample data. Test results show a faster convergence and higher accuracy than that of AFSA or BP neural network with a maximum error 1. 762 ￥/MWh and a mean relative error 2%.