Identification of sensor block model using Volterra series and frequency response function

This paper proposes a kind of identification method of block models to describe the nonlinear dynamic characteristics of sensors. The Volterra series are employed to analyze the block models and separate the higher-order Volterra kernels of the block models. The estimation method of frequency response function in the non-parametric form is used to identify the model of linear dynamic subsystem. And then coefficients of the nonlinear static subsystem are determined according to the nonlinear outputs of various orders. The identification method only needs the step or impact signal with different amplitudes as the input. These experimental data can be obtained in the calibration experiments of sensors easily. Considering the random noise or Gaussian noise in calibration experiments, the correlation function and bispectrum are used to reduce the influence of noise and improve the calculation precision. The method is applied to the hot-film mass air flow sensor for building its nonlinear dynamic model.