摘要
为解决传统算法收敛速度慢、精度低的问题,提出一种改进粒子群算法(improved particle swarm optimization, IPSO),通过在寻优过程中动态调整惯性因子ω和加速因子c1和c2,提高算法的寻优效率;利用改进算法优化BP(back propagation)网络的权值和阈值,建立尾矿坝地下水位预测模型,结合实例数据对预测模型进行验证。研究结果表明,改进算法的收敛速度得到改善,预测模型对坝体地下水位的预测精度得到提高。
To solve the low convergence speed and poor precision problem of the traditional prediction algorithm, an improved particle swarm optimization(IPSO) algorithm was proposed. The inertia factor ω and the accelerating factor c1 and c2 of the algorithm were dynamically adjusted during the searching process to improve the optimization effectiveness. The weights and thresholds of back propagation(BP) network were optimized by the improved algorithm. And the prediction model of groundwater levels in tailing dam was built and verified according to its instance data. The test results showed that the convergence speed of algorithm and the accuracy of prediction model was improved.
作者
郑店坤
许同乐
尹召杰
孟庆民
ZHENG Diankun;XU Tongle;YIN Zhaojie;MENG Qingmin(Mechanical Engineering School,Shandong University of Technology,Zibo 255049,Shandong,China;Shanbo Anjifu Gear Motor Co.,Ltd.,Zibo 255200,Shandong,China)
出处
《山东大学学报(工学版)》
CAS
CSCD
北大核心
2019年第3期108-113,共6页
Journal of Shandong University(Engineering Science)
基金
山东省自然科学基金资助项目(ZR2013FM005)
淄博市科学技术发展计划资助项目(No.JY20151587)
关键词
动态优化
地下水位
粒子群
BP神经网络
预测模型
dynamic optimization
groundwater levels
particle swarm
BP neural network
prediction model