为了较精确地表征超级电容的对外特性,提出了一种基于RBF-ELM(Radical Basis Function-Extreme Learning Machine)神经网络的超级电容建模方法。通过分析超级电容工作原理,提出并表征了影响超级电容对外特性的一个重要参数Q;介绍了所选网型RBF-ELM的原理及结构;在Matlab环境下,结合超级电容实际状态下的工作数据,选用RBF-ELM网络进行建模,仿真结果证明了所提参数Q的有效性。比较了其他网型的建模性能,表明该方法具有较好的实时性和精度。
An supercapacitor modeling method based on RBF-ELM(Radical Basis Function-Extreme Learning Machine)neural network was proposed in order to describe the output characteristics of supercapacitor with high accuracy. Firstly, by analyzing the theory of supercapacitor's working, an important parameter Q which had a significant impact on its performances was found and described. Then the theory of RBF-ELM was described. At last, by taking the Q as a part of the input vector of the neural network, the model of supercapacitor was established in Matlab with the data generated from experiments, whose performance was proved the validation of Q as well, and the real-time performance and better accuracy were also demonstrated by comparing with some kinds of other neural networks.