Sparsity Adaptive Compression Identification Algorithm for Volterra Series Model

Amount of computation will increase sharply in identification of Volterra series with the increase of order and memory length of nonlinear system. A sparse kernel coefficient identification model based on sparsity of coefficients in polynomial expansion and method of sparse signal reconstruction in compressed sensing is constructed in this study. Measurement matrix form of the second-order Volterra series is derived and sparse adaptive matching pursuit algorithm is used to reconstruct kernel coefficients of each order. Simulation results showed that this algorithm overcomes the influence of excessively high and excessively low estimated signal sparsity on reconstruction accuracy compared with orthogonal matching pursuit algorithm.