Coupled Identification Based Approximate Least Absolute Deviation for Multivariable System
【摘要】：Based on approximate least absolute deviation criterion and Newton's search principle, a full coupled identification algorithm is proposed for multivariable system whose model parameters are fully coupled. To solve the non-differentiable problem of the least absolute deviation, an approximate least absolute deviation objective function is established by introducing a deterministic differentiable function to replace the absolute residual. The proposed method can overcome the disadvantage of large square residual of least square criterion when the identification data is disturbed by the impulse noise which obeys symmetrical alpha stable(SaS) distribution. Moreover, a differentiable objective function avoids the operation of solving matrix inversion, which makes it easier to calculate. Simulation experiments show that the proposed full coupled identification algorithm can effectively identify the model parameters of multivariable system and restrain the impact of impulse noise.Satisfactory performance and strong robustness indicate the proposed multivariable identification algorithm could be well applied to practical industrial systems.