基于神经网络的速度传感器幅频特性改进

Amplitude frequency characteristics improvement based on ANN for velocity transducer

为了降低磁电式振动速度传感器的下限测量频率,以实现超低频振动速度测量,提出改进其幅频特性的函数连接型人工神经网络(FLANN)方法。该方法以磁电式振动速度传感器动态试验数据为基础,通过FLANN训练来确定传感器动态补偿网络,以改善它的幅频特性。介绍了原理和FLANN权值调整的算法,给出用FLANN建立的磁电式振动速度传感器动态补偿网络的数学模型。结果表明:这种幅频特性的改进方法具有精度高、鲁棒性好,并能在线修正等优点,在工程测试领域有重要的实用价值。

The method of amplitude frequency characteristics improvement is presented to realize ultra-low frequency vibration measurement based on functional link artificial neural network （FLANN） for magnet-electrical vibration velocity transducer. In this method, a dynamic compensation network can he set up according to measurement data of dynamic response of magnet-electrical vibration velocity transducer. The dynamic compensation network parameters are trained by FLANN. The principle and algorithms of weight values of FLANN are introduced and the dynamic compensating mathematics model is given for magnet-electrical vibration velocity transducer. The results show that the proposed method has high precision, strong robustness, on-line correction ability practical value in engineering measurement field.