Sensorless Control of Permanent Magnet Synchronous Machines based on oversampling, FPGA based aquisition of measured currents - II

• Applicant: Professor Dr.-Ing. Axel Mertens
• Subject Area: Electrical Energy Systems, Power Management, Power Electronics, Electrical Machines and Drives
• Term: since 2017
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 329209868

In the previous project, an encoderless, highly dynamic position and speed estimation for PMSM in the lower speed range including standstill was found and investigated, which allows a very good accuracy of the position estimation in the closed speed control loop and can also compensate for load torque changes in a highly dynamic manner. Only the inductance matrix of the machine is required for this. In contrast to most existing methods, the new method can also be parameterized analytically and does not require a large computational effort. The basis is an oversampling of the measured currents at a rate of approx. 106/s, which allows a precise analysis of the current waveforms over the duration of a PWM period. From this, the gradients of the currents, and thus the rate of change in flux, can be determined and compared with expected values based on the applied voltages. From the analytically available error function, a gradient can be determined according to the desired variable and calculated online with little effort. An online optimization of the error function in the manner of a gradient descent leads to the desired rotor position and speed within a sampling time. So far, however, this new approach could only be investigated with a small machine with surface magnets and almost linear behavior. In the follow-up project, an extension to heavily saturating machines is pursued, such as those typically used for higher power levels, e.g. in the field of traction. In such machines, saturation effects lead to operating ranges with vanishing or reversing anisotropies. In our own preparatory work, the procedure developed shows the potential of being able to operate within these areas when the mutual inductances are included in the models. Therefore, in the follow-up project, the developed procedure shall be extended for use with a higher-power machine with buried magnets, including operation in the regions with vanishing or reversed anisotropy. Side effects, such as currents distorted by eddy current effects or by long motor cables, must also be addressed. For an application in industrial drives, the current-dependent inductance matrix should also be able to be determined fully automatically and without a position sensor. Finally, the resulting procedure should be compared systematically with other processes according to the state of research in order to classify it in the existing body of knowledge.

• DFG Programme: Research Grants