A WNN model trained with orthogonal colliding bodies optimization for accurate identification of hammerstein plant

The identification of Hammerstein plant plays a key role in modeling practical dynamic systems. It's structure comprise of non-linearity associated with a static FIR block whose output is in series with a linear dynamic block and contaminated with noise. To identify such complex plant a new model is developed consist of wavelet neural network associated with FIR block, IIR block and noise. A new meta-heuristic Orthogonal Colliding Bodies Optimization (OCBO) is proposed based on the orthogonal combination of the position of the bodies which enables them to move towards a minimum energy level after collision (thus helpful in achieving a global minimum on a surface). All the weights associated with the proposed model are collectively trained with the OCBO to reduce the mean square error (MSE). For comparative analysis the proposed model is also trained with Colliding Bodies Optimization and Orthogonal Particle Swarm Optimization. Superior performance of OCBO trained model is achieved in terms of close response matching between plant and model output and lower sum of square error obtained during testing.