Research On Sensorless Control System Of Switched Reluctance Motor

The Switched Reluctance Motor is getting more and more attention due to its simple structure, reliable performance, high starting torque, wide speed range and high efficiency, and so on. SRM could work under certain bad circumstance, such as: extremely high speed, dusty, high temperature. But the position sensor, which is used to provide position message, is vulnerable to those factors and expensive. The sensorless control of SRM could promote its performance under bad circumstance, and reduce the cost and entire volume.For SRM, its flux linkage is determined by phase current and rotor position, so the position could be deduced from the flux and phase current. The Artificial Neural Network (ANN) is good at non-linear function fitting, so it could be used to estimate the rotor position on line. In order to reduce the DSP calculation time of ANN, we should simplify the ANN with good accuracy. The flux method depends greatly on the accuracy of flux. In order to get accurate flux curves, this paper use the finite element method, DC pulse method and AC method to analyze the prototype.In this paper, some trained neural networks are used to model the SRM in MATLAB/SIMULINK. After comparing the phase current between the simulation result and measurement result, the SRM modeling has been proved to be correct. On this basis, we can model the sensorless control of switched reluctance motor drive. At last, on the basis of the SRM senor control platform, the actual position provided by position sensor and actual trigger pulses are used to compare with the deduced position and trigger pulses, the result approves the method is correct.