An online intelligent modeling method for rate-dependent hysteresis nonlinearity

In this paper, an online intelligent modeling method, which is based on an improved online least squares support vector machines(IOLS-SVM), is presented for identifying rate-dependent hysteresis nonlinearity, and is used to online real-time training. In order to obtain the data which is adapted to IOLS-SVM, the original one-dimension input space is first projected on a high dimension input space, then the multi-valued function in one-dimension space can be converted into the single value function in the high dimension space. The data measured in the experiment are used for modeling. The numerical simulation shows the presented method can accurately describe the rate-dependent hysteresis in giant magnetostrictive actuator (GMA).