Erliang Zhang (Enseignant-Chercheur, School of Mechanical Engineering, Zhengzhou University, Henan, China) - Séminaire du LVA

Identification of linear dynamic EIV system excited by arbitrary input

Representations where measurement noises are present on both the inputs and outputs are called errors-in-variables models and play an important role when the identification purpose is the determination of the inner laws that describe the dynamics. They have been investigated in the past decade from different points of view. In this presentation, a novel EIV identification method will be presented for linear dynamic system, which has the following advantages,1. Allow arbitrary input signals, including nonstationary and harmonic components;2. Handle all kinds of measurement noise models (white, colored, (un)correlated);3. Return efficient estimates for plant model parameters, and consistent estimate for the input-output noise covariance matrix.   The basic idea and technical details will be explained, simulation examples are also provided. As it goes beyond the classicl EIV identification framework, some chanllenging problems will be opened, such as starting values of the colored noise model.