RECURSIVE LEAST-SQUARES ROTOR TIME CONSTANT IDENTIFICATION FOR VECTOR CONTROLLED INDUCTION MACHINE

This paper deals with on-line rotor time constant identification for vector controlled induction machine based on measurement of the stator voltages, currents and speed. Nowadays, field-oriented control (FOC) is a promising induction machine control method which leads to excellent motor dynamic performances. It is well known that in this FOC scheme, the induction machine parameters change widely during the operation of the drive, especially the value of rotor time constant which varies with rotor temperature and flux level of the machine. So, the quality of the drive system decreases if no means for compensation or identification is applied. Our research involved an on-line identification of rotor time constant by applying recursive least squares (RLS) approach in order to implement a robust control law for industrial application. A convenient formulation is proposed in order to compute physical parameters of the machine. The Bierman recursive least squares algorithm is briefly presented and tested with different configuration of machine excitation. The influence of state formulation and noise are also controlled to provide the user experimental datas which are fundamental during the implementation. In our application, the parameters of the induction machine are slowly variable which permits to perform linear identification method updating inputs and outputs by means of variable forgetting factor.