Download Nonlinear Homogeneous Order Separation for Volterra Series Identification This article addresses identification of nonlinear systems represented by Volterra series. To improve the robustness of some existing methods, we propose a pre-processing stage that separates nonlinear homogeneous order contributions from which Volterra kernels can be identified independently. The proposed separation method exploits phase relations between test signals rather than amplitude relations that are usually used. This method is compared with standard separation process. Its contribution to identification is illustrated on a simulated loudspeaker with nonlinear suspension.