Abstract:
This paper presents a sensor less scheme to obtain an optimum operating period for each load variation in the 8/6 switched reluctance motor. The drawbacks of using a position sensor in dusty environment are eliminated in this approach. Here, the operating parameters are fed back to the Adaptive Neural Network (ANN) controller. This model is developed efficiently and trained under supervised learning method in which flux-linkage and phase-current are fed as input and rotor position is estimated as output. Multi objective particle swarm optimization (MOPSO) based optimization technique is used in determining the optimum operating angle for the SR machine. The finalized model consist of ANN based position prediction loop, MOPSO based optimum angle selection loop with rigid current controller. The main objective of the model concentrates on maximizing the average torque with minimum torque ripple in the optimized operating region under various loading conditions with removal of position sensor. This system has the advantages of robustness, simple construction, reduced manufacturing cost with the absence of position sensor etc. The simulation is carried out in MATLAB/SIMULINK and Hardware is implemented using dsPIC.
