摘要
介绍了一种基于多层前向神经网络的电流控制逆变器，给出了神经网络及逆变器的具体实现方法与设计．为了提高神经网络的学习训练速度，笔者采用具有较快收敛速度的递推最小二乘算法（ＲＬＳ）用于网络权值的训练．应用研究表明该逆变器较传统逆变器具有动态响应速度快，静态精度较高等优点．
In this paper is introduced a current-controlled voltage source inverter based on multiayer forward neural network, with the design and implementation of the neural network and the inverter given. Recursive least square (RLS) with comparatively high convergent speed is adopted to update training in network weights. Results obtained show that the inverter has quicker dynamic response and higher steady-state accuracy.
出处
《汕头大学学报（自然科学版）》
2000年第2期15-18,共4页
Journal of Shantou University：Natural Science Edition
