机构地区: 西南石油大学机电工程学院,成都610500
出 处: 《传感技术学报》 2017年第7期1076-1082,共7页
摘 要: 在管道运输中传感器的正常使用至关重要,为了防止因传感器故障而导致的数据采集失效,造成误报警和漏报警,对传感器本身的故障诊断和失效分析已经成为当前研究的重要方向。通过对各传感器采集数据进行组合分析,以各传感器之间的数据关联特性作为研究内容,运用K-最近邻算法对管道上传感器所采集数据进行相似性拟合,提出采用C4.5算法定义各传感器所采集数据对目标传感器的支持度以决定目标传感器数据的有效性,对故障传感器运行状态进行分析与定位,进而判断传感器的数据可靠性和输差出现位置。实验结合西南某管道流量传感器数据进行分析,结果表明,该算法能够准确判断目标传感器数据的有效性和故障传感器在时域中发生的位置。 The normal use of the sensor in the pipeline transportation is very important. In order to prevent the data acquisition failure which caused by the sensor fault. The fault diagnosis and failure analysis of the sensor has be- come an important direction of the current research. Through the combination analysis of the data collected by the sensors ,the data association characteristics between the sensors are used as the research content. The K-nearest neighbor algorithm is used to fit the data that collected by the sensors of the pipeline. The C4.5 algorithm is pro- posed to define the support degree of each sensor to determine the target sensor data. To analyze and locate the fault sensor's running state, and then to determine the data reliability and determine the location of the sensor. The exper- iment analysis the data of the pipeline flow sensors. The results show that the algorithm can accurately determine the effectiveness of the target sensor and the location of the fault sensor in the time domain.