基于噪声特征的地震动信号目标预警和识别算法研究

利用MEMS加速度传感器采集到的人员、车辆目标在地面运动产生的地震动信号，由MSP430单片机采集和处理，完成对地面目标的预警、分类识别. 预警提出了基于噪声方差的3倍作为阈值的方法，目标识别采用一种改进的过零分析方法. 结果表明，该算法能对8 m以内的人员、20 m以内的轮式车辆进行预警和分类识别，正确识别率达到85%以上. 此算法对于识别数字输出式传感器采集到的信号非常简单、实用. 

Research on Early-Warming and Target Identification Algorithm of Seismic Signal Based on Characteristics of Noise

Seismic signal generated on ground by moving person and vehicle target was sensed by MEMS accelerometer, then collected and processed by MSP430, achieving early-warming and target identification for ground targets. A method based on three times of the noise variance as the threshold value was proposed as to early-warming. A modified zero-crossing analysis was used to identify different kind of targets. The results show that person within 8 meters and vehicles within 20 meters away from the accelerometer can be detected and identified effectively. The correct identification rate can reach 85% or more. This algorithm is very simple and practical when processing signal collected by sensors with digital output.