摘要
采用数据融合方法研究了基于多地震波传感器的管道安全监测预警系统中目标特征提取及识别问题。用多个传感器及处理模块采集管道周围目标产生的震动信号,采用经验模态分解方法对地面震动信号进行处理,提取分解结果的归一化峭度,将其作为特征向量;利用主要频率区间的特征向量进行单一传感器的目标识别。由于监测系统由多个单独传感器采集模块组成,为了提高目标的识别率,采用D-S证据理论对识别结果进行了数据融合,得到最终识别结果。利用该方法对实验采集数据进行处理,验证了文中提出的方法。
Abstract： A method for extraction of target features in the pipeline security monitoring and pre-warning system based on multi-seis- mic sensors was studied with data fusion to improve recognition rate. Many sensors and processing modules were used to acquire the seismic signals generated by the ground targets. The non-stationary signal analysis method based on empirical mode decomposition was used to process the seismic signals. The normalized kurtosis extracted from the decomposed results was composed of the feature vectors. The decision of single sensor was made of the major normalized kurtosis in the main decomposed frequency bands. The D-S evidence reasoning was used to fuse these individual recognition results for improving recognition rate, and the final decision was obtained. The experiment data were acquired and analyzed. The analysis results verified the validity of the new method.
出处
《石油学报》
EI
CAS
CSCD
北大核心
2009年第3期465-468,共4页
Acta Petrolei Sinica
基金
国家自然科学基金重点项目(No.60534050)资助
关键词
管道安全监测
预警系统
目标识别
经验模态分解
地震波传感器
数据融合方法
特征向量
地震信号
pipeline security monitoring
pre-warning system
target recognition
empirical mode decomposition
seismic sensor
data fusion method
feature vector
seismic signal
