Induction machines air gap flux prediction with artificial neural network

Direct air gap flux measurement is one approach to induction machine torque measurement. It has limitations and inconvenience for implementation, however. Instead of using the partial phase winding as a sensing coil to measure the air gap flux, imaginary sensing coils are used here. The voltage induced in the imaginary sensing coil is the same with its corresponding partial phase winding coil. The air gap flux linkages are predicted by a trained artificial neural network according to line voltage measurements and are used to calculate the air gap torque. The results are quite close to expected torque values. This method has been implemented in the laboratory on different induction machines and the results are quite satisfactory for both steady state and transient state torque measurement.