Lower cut-off frequency of the magnetoelectric velocity sensor is limited by the structure and the volume due to its nature. In the engineering filed, the band of the sensor should be stretched to detect the velocity signals with lower frequencies. The dynamic characteristic model of magnetoelectric velocity sensor is derived and a dynamic compensation strategy based on FLANN (function link artificial neural networks) algorithm for velocity sensor is proposed. The dynamic compensation effect is compared between FLANN and the pole-zero placement method. The results show that the compensation error is smaller when using FLANN algorithm and the frequency bandwidth of the velocity sensor is expanded effectively, and thus the measurement of ultra-low frequency in engineering is satisfied.%磁电式速度传感器由于自身工作原理,其固有频率下限值受到结构和体积的限制.应用于振动测试时常需对其工作频带进行补偿扩展,以使其能检测固有频率以下的速度信号.本文针对磁电式速度传感器,建立了其动态数学模型,给出了一种基于函数链人工神经网络(FLANN)算法的动态补偿策略分析对比了采用传感器输入输出设计的FLANN算法补偿器与采用零极点配置法进行动态补偿的效果.结果表明,采用FLANN算法设计的补偿器具有更小的补偿误差,且有效扩展了速度传感器的使用频带,很好地满足了工程上超低频振动测量的要求.
展开▼