摘要
基于自适应神经网络控制(DANC)算法提出一种改进的最大功率追踪(MPPT)算法。针对DANC在权值更新的过程中采用的梯度下降算法因初始权值而导致局部收敛的问题,该文根据光伏(PV)阵列部分遮挡情况下的功率-电压(P-V)特征曲线,提出一种反馈负载电压遍历和DANC相结合的算法。首先通过反馈负载电压遍历方法快速达到参考电压附近,然后利用DANC在线学习算法稳定到峰值,通过比较峰值大小,最终找到全局最优解。通过Matlab仿真和硬件实验验证改进算法的可行性。
Based on Adaptive Neural Network Control algorithm(DANC),an improved maximum power point tracking algorithm(MPPT)is proposed in this paper. For the problem of local convergence caused by the initial weights in DANC algorithm,an method combined the feedback load voltage traverse with DANC is proposed,according to the characteristic curve of Power-Voltage(P-V)in case of partial occlusions of Photovoltaic(PV)arrays. First,a point near the reference voltage is reached by the feedback load voltage traverse. Then by using DANC algorithm,the result is focus on peak value. By comparing the peak values,the global optimal solution is obtained. Based on MATLAB simulation and hardware experiment,the feasibility of the improved algorithm is verified.
作者
张晓强
刘宜罡
邹应全
张翠芳
Zhang Xiaoqiang;LiuYigang;Zou Yingquan;Zhang Cuifang(School of Information Science and Technology,Southwest Jiaotong University,Chengdu 610031,China)
出处
《太阳能学报》
EI
CAS
CSCD
北大核心
2019年第11期3095-3102,共8页
Acta Energiae Solaris Sinica
关键词
自适应神经网络控制
电压遍历
光伏阵列
电导增量
最大功率追踪
adaptive neural network control(DANC)
voltage traverse
photovoltaic(PV)
incremental conductance(INC)
maximum power point tracking(MPPT)
作者简介
通信作者:邹应全(1976—),男,博士、副教授,主要从事智能信息采集与处理方面的研究。zouyingq@home.swjtu.edu.cn。
