摘要
针对径向基函数 (RBF)神经网络用于非线性系统辨识时存在的问题 ,对径向基函数网络的拓扑结构作了改进 ,并给出了改进的径向基函数 (MRBF)神经网络的中心选取方法和权值在线调整算法 ,最后用改进的径向基函数网络对一个典型工业对象 (CSTR)进行了应用研究 ,结果表明方法有效。
For solving some problems in nonlinear system modeling based on radial basis function neural networks, a modified construction of RBF networks is proposed. And the selection of the modified RBF networks centers and the weight tuning algorithms is proposed. The application results show that the MRBF networks successfully model a classical chemical plants.
出处
《计算机仿真》
CSCD
2003年第11期61-63,共3页
Computer Simulation
