用于步进电机细分控制的知识人工神经网络

摘要

The relationship between the control current and the angle displacement in the subdividing control of stepping motors is a kind of non-linear function, whose precise fitting is an important subject in the subdividing-control of stepping motor and a new attempt to simulation with artificial neural network. Because of the low convergence speed of BP ( back propagation) learning algorithm for feed forward neural network and lack of prior knowledge in the fitting nop-linear function, a KBANN ( knowledge-based artificial neural network) is put forward to determine the best subdividing current data in this paper. The simulation results show that the KBANN has the advantages of high learning speed and high accuracy in the subdividing of stepping motors, and can realize the uniform stepping subdividing control of stepping motors.%步进电机细分控制中电机绕组电流与电机角度输出是一种非线性函数,其精确拟合是步进电机细分控制中的一个重要课题,应用神经网络对其模拟是一种新尝试.针对前馈神经网络的反向传播(BP)学>-J算法在逼近非线性函数时收敛速度慢,没有先验知识的缺点,提出利用基于知识的人工神经网络(KBANN)来确定步进电机的最佳细分电流数据.仿真 结果表明,KBANN具有精度高、速度快的特点,能够实现步进电机均匀步距的细分控制.