An Improved IPL Algorithm for Constructing RBF Network and its Application in Fault Classifier

摘要

This paper adjusts the sampling operator of Incremental Projection Learning (IPL) algorithm combined with three learning phases for Radial Basis Function (RBF) networks, because of the simplicity of the new method, the improved IPL algorithm’s calculation speed can be faster than before. The simulation results show that the new algorithm can induce a simpler network structure than the former algorithm, and the output of the new IPL inducing RBF network is more accurate than before. A circuit’s fault diagnosis is studied as a test case, which is based on the inducing RBF network, and the experimental results show that this approach may serve as the needs of fault classification process.