基于噪声特征的地震动信号目标预警和识别算法研究

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

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8m以内的人员、20m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.

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.