Year: 2019
Type: dissertation
Abstract: The use of electric machines can be found in many applications such as household appliance, machine tools, vehicles and railways, due to their indispensable role in converting energy.Most recently, permanent magnet synchronous machines have been increasingly employed in electrical/hybrid electrical vehicles, industrial servo drives and wind power generators for their high power density and good efficiency.There is a growing trend towards the inclusion of thermal management in permanent magnet synchronous motors by monitoring their internal temperatures during real-time operation, because high temperatures can significantly shorten the lifetime of the motor components.Whilst temperature sensors are suitable for measuring stator temperatures, fixing them on rotating permanent magnets is difficult in practice.As a result, model-based temperature estimation methods are preferable.A practical and computationally efficient system for the estimation of the critical temperatures in permanent magnet synchronous machines is introduced, based on a low-order lumped parameter thermal network which represents stator iron, stator winding and permanent magnet.The parameterization of the network requires an accurate rotor temperature measurement, which is provided by a PWM-based estimation algorithm predicting rotor temperature via permanent magnet flux linkage.The proposed temperature estimation system is validated in simulation, including offline simulation in Matlab/Simulink, and online simulation utilizing the Hardware-in-the-Loop technique, which performs the emulation of motor and control in two real-time platforms.Comprehensive experimental validation is also conducted on a three-phase surface-mounted permanent magnet servo motor, with motor temperature estimation error less than 6℃.The main contributions of the research work include: a) A three-node thermal network for motor temperature estimation, which is simple to implementdetailed knowledge of motor dimensions, material properties is not needed, as the thermal parameters are derived from a measurement-based recursive parameter identification procedure, based on the recursive Kalman Filter, b) A simplified and accurate PWM-based rotor temperature estimation method without using signal injection, which is a commonly-employed approach for temperature estimation and disturbs motor operation.It is also insensitive to practical implementation errors, such as inverter nonlinearity, c) The integration of the rotor temperature estimation method and the thermal network.As a result direct rotor temperature measurement which can be expensive and troublesome is avoided.Prof. Jiabin Wang of the Department of Electronic and Electrical Engineering, the University of Sheffield, for their continuous encouragement and guidance in the course of my study.I would not have been able to complete this project without their support.My thanks also to the colleagues of the Electrical Machines and Drives Research Group, for helping me solve the technical difficulties I encountered throughout the years, and providing a lively and friendly working atmosphere.I also would like to express my gratitude towards my family, particularly my parents, for their unconditional financial and moral support, without which this journey would not have been possible.
Author: Shuai Xiao
Institution: University of Sheffield
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Keywords: permanent magnet synchronous motor, thermal management, lumped parameter thermal network, rotor temperature estimation, permanent magnet flux linkage, recursive Kalman filter, recursive parameter identification, surface-mounted permanent magnet motor, servo motor, hardware-in-the-loop, electric vehicles, wind turbine generator, inverter nonlinearity
Open Access status: green
Funder: University of Sheffield
OpenAlex ID: W2982075748
