Convergence Analysis of Multi-innovation Learning Algorithm Based on PID Neural Network

Abstract

In order to improve the identification accuracy of dynamic system, multi-innovation learning algorithm based on PID neural networks is presented, which can improve the online identification performance of the networks. The multi-innovation gradient type algorithms use the current data and the past data that make it more effective than the BP algorithm in view of accuracy and convergence rate. Simulation results showed that the proposed algorithm is effect.

Keywords

• Multi-innovation
• PID neural networks
• System identification
• Nonlinear system