摘要
针对超声波电机时变、强耦合使得数学模型难以建立的问题,以超声波电机转速的非线性逆控制为应用背景,本文给出了超声波电机神经网络逆模型的辨识建模方法。基于实验测得的足够样本数据,通过反复测试确定模型形式为三层非线性DTNN网络。进行串-并联辨识,建立了以电机转速为输入、驱动频率为输出的超声波电机神经网络逆模型。所得模型的输入-输出关系与实测数据接近,表明了所建模型的有效性。
Ultrasonic motor (USM) has time-varying nonlinearity and strong coupling, which causes a great deal of difficulties in establishing the mathematical model of the motor exactly. To solve this problem, aiming at the nonlinear inverse control of speed for ultrasonic motor as application background, this paper worked out an inverse model identification method of ultrasonic motor using neural network. Based on enough measured data, the model form was determined as nonlinear DTNN network with three layers after re peated testing. Later, the paper conducted series parallel identification, worked out the neural network inverse of ultrasonic motor using the motor speed as input and driving frequency as output. Calculation results obtained model are close to measured data, it indicates that the modeling method is reasonable and the is effective. model of the model
出处
《微电机》
北大核心
2013年第4期75-77,共3页
Micromotors
关键词
超声波电机
逆模型
神经网络
系统辨识
ultrasonic motor
inverse model
neural network
system identification