Skip to Main Content
Both soft computing (SC) and hard computing (HC) techniques are often successful for solving real-world control problems. In cases where problems could be solved by either or both methodologies, an important research problem is to find what are the advantages for fusing SC methods together with HC methods, rather than using the HC method alone. Hence, in this paper, a performance comparison is detailed for a fused soft control/hard observer type controller (where a classical or HC type observer is fused with an adaptive fuzzy or SC type controller) and a hard control/hard observer type controller (where both the observer and feedback linearization controller are classical HC types). The domain in which this comparison is made is for the sensorless speed control of switched reluctance motors (SRMs). This is because this type of motor has highly nonlinear characteristics, and the HC type controller can often be detrimentally affected by modeling inaccuracies, as well as noise. Simulation and experimental results are illustrated to show the performance comparison of the soft control/hard observer type controller and the hard control/hard observer type controller under a wide range of identical operation conditions including transient speed and torque, SRM model parameter variations, and measurement noise. It can be seen from the results that the soft control/hard observer type exhibits a better performance than the hard control/hard observer type controller.