摘要
针对入侵杂草算法易陷入局部最优、后期寻优精度较低等不足,提出一种差分进化入侵杂草(DEIWO)算法用于训练前向神经网络,结合入侵杂草算法的种群多样性和差分进化算法的启发式搜索等特质以增强算法的全局搜索能力和局部挖掘能力,建立基于DEIWO算法的神经网络预测模型。通过实例验证了本文改进的算法具有较好的寻优精度和收敛速度,预测模型可行和有效。
In view of the defects of being easily trapped in the local optimal and low accuracy of later optimiza-tion in invasive weed optimization （IWO） algorithm, a differential evolution invasive weed optimization （DE-IWO） algorithm is proposed for training feed-forward neural network. It takes the advantage of the features ofthe population diversity of IWO and heuristic search of DE, and enhances the global search ability and localmining ability of algorithm. Then, a neural network prediction model based on DEIWO algorithm is estab-lished. The results show that better optimization accuracy and convergence speed of the modified algorithm,and the feasibility and effectiveness of the prediction model are verified by the example verification.
出处
《辽宁科技大学学报》
CAS
2014年第6期561-568,共8页
Journal of University of Science and Technology Liaoning
基金
辽宁省首批"十百千高端人才引进工程"项目资助
关键词
入侵杂草优化算法
差分进化入侵杂草算法
神经网络学习算法
神经网络预测模型
invasive weed optimization algorithm
differential evolution invasive weed optimization algo-rithm
learning algorithm of neural network
prediction model of neural network
