Abstract: This paper discusses the order and parameter identification of hydraulic system by time eries ARMAX. The cooperation of extended least-square and BTC criterion proposed here has higher accuracy than other method. The suborder and delay of the model are determined. According to the asymptotically Gaussian distribution of estimated parameter. This method is applied to identifying the order and parameter of hydraulic position system and its noise model. The effectiveness of identified model is illustrated by simulation step response in comparison with measured step response.

Key words: 时间序列, 系统辨识, 液压位置系统, 增广最小二乘法