Rotor flux observation accuracy is a key factor to affect control performance of multi-motor closed-loop system
working on rotor flux field-oriented vector control mode, but the precision of flux observation is usually difficult to meet
the requirement because of motor parameters variation. Aiming at three-motor synchronous system, the model of rotor flux
is built based on local model network. In this model, variability of machine parameter is taken as interference term, so the
accuracy of flux identification is improved. Then, based on the neural network inverse control method, linear subsystems of
speed, rotor flux linkage and tension can be obtained by combing the inverse neural network with the three-motor synchronous
system. Thus, the speed, flux and tension control of three-motor system can be decoupled. The simulation results show that
the flux model has satisfactory accuracy, and control system has good dynamic and static operation performance.