Purpose–This paper proposes a robust modeling method of a giant magnetostrictive actuator which has a rate-dependent nonlinear property.Design/methodology/approach–It is known in statistics that the Least Wilcoxon learning method developed using Wilcoxon norm is robust against outliers.Thus,it is used in the paper to determine the consequence parameters of the fuzzy rules to reduce the sensitiveness to the outliers in the input-output data.The proposed method partitions the input space adaptively according to the distribution of samples and the partition is irrelative to the dimension of the input data set.Findings–The proposed modeling method can effectively construct a unique dynamic model that describes the rate-dependent hysteresis in a given frequency range with respect to different single-frequency and multi-frequency input signals no matter whether there exist outliers in the training set or not.Simulation results demonstrate that the proposed method is effective and insensitive against the outliers.Originality/value–The main contributions of this paper are:first,an intelligent modeling method is proposed to deal with the rate-dependent hysteresis presented in the giant magnetostrictive actuator and the modeling precision can fulfill the requirement of engineering,such as the online modeling issue in the active vibration control;and second,the proposed method can handle the outliers in the input-output data effectively.
