Sensorless speed detection of induction machine drives using Goertzel algorithm
Sensorless speed detection of induction machine drives is a goal that has seen efforts towards in generations. Traditional methods of speed detection include shaft - mounted tachometer, etc. However, since most induction machines work in harsh conditions and maintenance is not expected to be usual, shaft-mounted-tachometers generally lower the strength of the shaft and reduce the lifetime of the induction machine. It has been found that eccentricity and rotor slot harmonics in the induction machine stator current contain accurate speed information. Many efforts have been made to extract ihformation from eccentricity and rotor slot harmonics. However, they usually end up with extremely complicated analog circuits and unsatisfying performances. It is only after the development of Digital Signal Processing theory and DSP microprocessor that the extraction of speed information from eccentricity and rotor slot harmonics became feasible. Since then, algorithms have been developed to perform sensorless speed detection. However, they generally turned out to be very MIPS (million instruction per second) intensive and required a large number of RAM. And this prevents the application of sensorless speed detection in the motor industry. This thesis intends to investigate and develop techniques that utilize the unique characters of Goertzel algorithm, namely, very low RAM requirements and lower MIPS requirements. Such techniques would make the sensorless speed detection more feasible and applicable to the motor industry. In the process of implementing Goertzel algorithm, some new phenomena were identified and some previous work in the area was questioned. It was found that it is very possible to design a sensorless speed detection algorithm based on Goertzel algorithm with very low RAM and MIPS requirements while achieving the desired resolution.