基于改进萤火虫算法的小波神经网络短期负荷预测方法  

Short-term load forecasting method of optimized wavelet neural network based on modified firefly algorithm

作　　者：刘丹[1] 张腾飞[1] 

机构地区：[1]南京邮电大学自动化学院,江苏南京210023

出　　处：《微型机与应用》2016年第23期56-58,共3页Microcomputer & Its Applications

基　　金：国家自然科学基金项目(61105082);江苏省"青蓝工程"基金(QL2016);南京邮电大学"1311人才计划"基金(NY2013);江苏省普通高校研究生科研创新计划项目(SJLX15_0387)

摘　　要：传统的小波神经网络以梯度下降法训练网络,而梯度下降法易导致网络出现收敛早熟、陷入局部极小等问题,影响网络训练的精度。文章将萤火虫算法用于训练小波神经网络,在全局内搜寻网络的最优参数。为了提高萤火虫算法参数寻优的能力,在训练过程中自适应调节γ值。同时利用高斯变异来提高萤火虫个体的活性,在保证收敛速度的同时避免算法陷入局部极小。将优化后的小波神经网络用于短期负荷预测,实验证明改进后的预测模型非线性拟合能力较强、预测精度较高。The traditional wavelet neural network is trained by the gradient descent algorithm, and the algorithm can easily lead to premature convergence and trap in local minimum, which affect the training accuracy of the network. In this paper, the firefly algorithm is used to train the wavelet neural network to search the optimal parameters of the network in the global. In order to improve the firefly algorithm＇ s ability of parameter optimization, the value of ＂y is adjusted adaptively in the training process. At the same time, Gauss variation is used to improve the activity of firefly individuals in order to ensure the convergence speed and avoid falling into local minimum. The optimized wavelet neural net-work is applied to short-term load forecasting, and the simulation results show that the improved prediction model has strong nonlinear fitting a- bility and high precision.

关 键 词：小波神经网络 萤火虫算法 负荷预测 全局寻优 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]