Identification of nonlinear systems using quadratic Volterra series, kernels expansion on wavelet with selection of best basis

Identification of a time-varying nonlinear system using a Volterra model is what our work is concerned about. To parameterize the Volterra kernels for quadratic Volterra series different methods can be used. Using orthonormal basis is one of the popular methods. The system’s time variation is approximated by a weighted sum of appropriate basis sequences. We use wavelet packets as orthonormal basis that can increase the flexibility of the model by offering an appropriate basis to be selected. We also use best basis algorithm with entropy criterion to select the wavelet packets to approximate the true system’s model. In this algorithm the minimum entropy criterion determines the best and the most efficient basis approximation that makes minimum number of sequences causing the best model.

کلیدواژه ها:

Volterra series ، nonlinear system modeling ، wavelet packets ، best basis algorithm ، entropy criterion