Kalman filtering for temperature estimation of electric motors

Due to their high power density and good efficiency, permanent magnet synchronous machines (PMSM) have been increasingly employed in high-power applications such as vehicular propulsion (electrical/hybrid vehicles), industrial drives and power generation. Since high temperatures can significantly shorten the lifetime of the motor components and cause non-reversible demagnetization of the permanent magnets, there is a growing trend towards real-time monitoring the internal temperatures of such machines during operation. Therefore, to guarantee optimal utilization of the machine, maximizing its efficiency while assuring safer operation modes, the temperature at some key points within the motor needs to be measured. While the temperature at the stator can be easily accessed by embedding thermal sensors, rotor temperatures are difficult to measure in practice. As an alternative to conventional direct/indirect measurement approaches, model-based methods have been investigated in the past decades. In this work, the feasibility of using the Kalman algorithm, as a thermal observer for temperature estimation, is investigated. This model-based approach starts from a simplified linear time invariant finite element model through which the performance of such strategy is evaluated to be able to apply it to more complex PMSM models.