摘要
磁浮控制系统采用涡流传感器检测其运行状态。由于控制系统对精度和实时性有较高要求,涡流传感器输入输出特性的非线性和模型随环境的变化需要快速准确的补偿。为此,采用径向基函数网络建立涡流传感器逆模型,使加入逆模型后涡流传感器输入输出映射单位线性化。采用本文提出的简化自适应隐层结构和中心学习算法能够快速准确得到网络结构和参数。实际测量说明,该方法在精度和实时性方面满足要求,校正误差小于0.7%,能够补偿涡流传感器模型的变化。该方法能够用于精度和实时性要求高的其他传感器校正中。
The eddy current sensor is used in the maglev control system to detect the state of the system. Because of the requirements of real time and precision, the nonlinear input-output characteristic of the eddy current sensor and its model variation with environment should be calibrated accurately and quickly. And then, the Radial Basis Function （RBF） neural network is used to construct the inverse model of the eddy current sensor. The input output mapping is unity after the adding the RBF network. The proposed simplified adaptive algorithm for hidden layer structure and center value can compute the structure and parameter of RBF network quickly and accurately. In the practical measurement, the calibration error is less than 0.7% and the variation of the model can also be calibrated. The method can satisfy the requirements of real time and precision in the maglev control system. It can also be used to calibrate other types of sensors used in high precision and real time system.
出处
《电工技术学报》
EI
CSCD
北大核心
2006年第3期71-75,88,共6页
Transactions of China Electrotechnical Society
基金
北京交通大学科技发展基金资助项目(DQJ05007)。
关键词
涡流传感器
RBF网络
非线性校正
简化自适应学习算法
Eddy current sensor, radial basis function, nonlinear calibration, simplified adaptive algorithm
