Abstract In order to avoid the low-frequency vibration of the stepping motor, and to improve control precision, often using subdivision control technology. Due to non-linear relationship of the stepper motor winding current, output torque and angle changes, it is difficult to find a precise mathematical model to achieve uniform subdivision step angle. In this paper, using the characteristic of BP neural network is possible to approximate the characteristics of any precision complex nonlinear mapping algorithm and does not require the specific case of the mathematical model. It is able to achieve uniform subdivision step angle through training the relationship of "controlling the current duty cycle - output step angle", without measuring the winding current and voltage value.
2015, Vol. 16 
