ANN interior PM synchronous machine performance improvement unit

Abstract EN

This paper propose improve of the permanent magnet machine efficiency by introducing two neural network units, using the back propagation (BP) learning algorithm due to its benefits.

These units will drive the motor at highly performance and optimum efficiency.

The first unit using voltage reduction technique, and presents performance improvement with maximum torque per ampere control technique.

The second in the manner of V / f control, at various frequencies values.

For each one of them, the goal is to pick the peaks of efficiencies at various torque values.

For second one, it can be generate set of curves which govern the operation of optimum efficiency by giving suitable voltage at each frequency value.

Also, deduction for one function connects between input and output for neural network, to can use with each static and dynamic model to improve its performance.

The validity of these units comes from comparisons between techniques values obtained in MATLAB environment with corresponding from Neural Networks.