Estimation of temperature in turbo-molecular pump based on motor resistance online identification

Abstract

The temperature of the turbo-molecular pump (TMP) system has to be monitored in real-time for long stable operation in a high vacuum environment. Since the high-speed motor is the major heat source of the TMP system, the thermal sensors such as Pt100 are widely used to measure the temperature via being embedded in the motor stators. However, these thermal sensors require extra wires and signal conditioning circuits, which increase the complexity of installation. In order to solve this problem, a real-time online estimation method for the temperature of the TMP driven motor is proposed in this paper. First of all, the resistance identification model of the high-speed motor used in TMP system is built. After that, an improved recursive least squares (RLS) algorithm, which introduces a discount factor to overcome the problem of data saturation and weak dynamic tracking, is used to identify the resistance of the motor stator windings. Then, the temperature of the TMP could be estimated indirectly based on the linear relationship between the motor stator resistance and the temperature. At last, the proposed method is tested in a self-developed magnetically levitated turbo-molecular pump (MLTMP) system, and the experimental results verify its validity and reliability.