Power balance control of multi-motor driving belt system using fuzzy neural network

Multi-motor driving belt conveyor system has the characteristics of nonlinear, big delay and multivariable. To achieve its power balance control, a T/2 parameter forecast method was introduced, and a multi-variable fuzzy controller was built. It adopted currents as control variables and belt speed as the reference. For its too many rules a modified CPN (Counter Propagation Network) was analyzed. It can realize fuzzy algorithm by executing pattern matching and weight average. Meanwhile membership functions and rules can be adjusted on-line. Simulation was made and compared with BP and conventional CPN algorithms. Results show that it has faster learning convergence speed and stronger generalization than the other two, and has faster response speed, smaller overshoot and steady-state error than fuzzy algorithm. The trained network was tested on a three-10kw-motor driving system, and power kept balance within its 4%, which donates that the modified CPN is suitable for this system.