基于改进MSPCA的交通检测器故障诊断模型

Traffic Detector Fault Diagnosis Model Based on Improved MSPCA

针对交通检测器故障诊断过程中,噪声掩盖了部分故障信息以及故障信息分布的多尺度性,提出了一种改进的多尺度主元分析(MSPCA)模型。模型首先将交通检测器数据进行分段处理,再加入改进的小波阈值除噪,对滤除噪声后的小波系数进行主元分析,最后利用二维贡献图完成故障的定位。模型应用于线圈检测器的故障诊断实验,与MSPCA及自适应主元分析相比,该模型减小了误报率和漏报率,准确率更高,抗噪能力更强。

During the process of data fault information , it is found that the fault information has a property of multi -scale, and sometimes parts of them are covered by random high frequency noise , so a model of data fault diagnosis based on improved multiscale principal component analysis （MSPCA） was presented.Firstly, an improved wavelet threshold method was joined , which was used to remove most of random high frequency noise and then improved the data confidence .Secondly , the model de-composed the reconstructed signals by principal component analysis .After this, the model finished the mission of the fault isola-tion by two-dimensional contribution plots .Finally, a case study of loop detector data fault diagnosis showed that the model had many advantages such as lower fault and failing rate , and stronger anti-noise ability .