Validation of simple thermal model of permanent magnet synchronous induction machine

By increasing requirements on efficiency of electrical drives, especially in road and rail vehicles, and by the improving multi-physical modelling environments the simulation received an important role in design of electrical drives. It is an advantage to simulate both, the machine's electro-mechanical and thermal behaviour. Thermal models of electrical machines exist in different levels of complexity. One approach is to create a thermal model, based on known geometrical parameter for each modelled machine. Another approach is to create a simple thermal model, which consists of a few thermal capacitances and resistances. Therefore, it is possible to describe this thermal network by equations with low number of parameters. These parameters can be identified for each investigated machine. The advantages of simple thermal model are obvious, when no geometrical data are available or several machines have to be modelled by reasonable effort. The simple thermal model was built in this project by multi-physical modelling language Modelica. Beside the simple thermal model of Permanent Magnet Synchronous Machine, further model components were created for processing the reference measurement data and for modelling thermal losses in the machine. The used reference data are recorded in the laboratory and represents the machine's real behaviour. The unknown parameters of the thermal model were identified by optimization of the minimum of temperature deviation to reference data, by an optimization program. Then the thermal model's behaviour was analysed by setting the identified parameter values. Different methods are presented for improving the accuracy of the simple thermal model. The achieved accuracy of thermal model in described process shows the simple thermal models' opportunities in design and operation of electrical drives. Other machines can be modelled with the same modelling environment by simple repeating the described steps for parameter identification and model validation, if measurement data for that specific machine are available.