Drive of senseless Switched Reluctance Motor (SRM) depending on Artificial Neural Networks

The Competition between the various types of electric motors has not been dependent on the development principle of the performance. Instead of that the total cost of the drive system (Motor-Inverter-Control system) has become an important element which cannot be ignored in the market of automated working mechanisms and modern drive systems. It was necessary to get rid of the position sensor which was the basic unit in the drive system of the Switched Reluctance Motor (SRM), and to replace it by a digital estimation system of the position depending on artificial intelligent technology to reduce the cost of the drive system in one hand and to qualify this motor to work in very high-speed range on the other without causing mechanical problems. This appears clearly in low-power applications like pumps and fans. Therefore, a study has been applied on a drive system of three phase (SRM) of 6/4 type without external position sensor. It was based on Artificial Neural Networks (ANNs) technique with only measuring stator phase currents and fluxes. Whereas the resulting drive system includes neural network estimator acting on closing the motor control loop. This system is described by its performance strength and its high ability to estimate the rotor angle of the SRM after completing the training of the neural network on the data set of (the stator fluxes, stator currents and the rotor position), including different cases which the motor may experience in practical conditions of the environment operation of the studied system, and depending on the neural network method. The results of simulation of the drive system in the presence of neural network estimator have shown a strong performance of the estimator and a high ability to estimate the SRM rotor angle in the speed range for the motor operation in both the linear and non-linear regions and with/without load.