基于自适应LMD和SVM的电传系统传感器故障诊断  

Fly-by-wire Sensor Fault Diagnosis Based on Adaptive LMD and SVM

作　　者：吴学钊[1] 王小平[1] 林秦颖[1] 王发威[1] 

机构地区：[1]空军工程大学航空航天工程学院,西安710038

出　　处：《计算机测量与控制》2013年第6期1487-1490,共4页Computer Measurement &Control

摘　　要：提出了基于自适应LMD分解和SVM的方法对电传系统法向过载传感器进行故障诊断;首先建立电传系统纵向通道模型;然后针对故障信号的非线性和非平稳性,采用LMD方法进行特征向量提取,考虑到电传系统存在多频段、强噪声干扰,设计自适应滤波器,去除特定频带噪声信号,分解得到的频率带保留了反映传感器工作状态的特征能量段;最后针对故障信号小样本特点,采用SVM进行故障识别;仿真结果准确率高达98%,说明文中所用方法能够有效诊断传感器故障,对电传系统的多频段、强干扰噪声有很强的鲁棒性,提高了故障诊断准确率。In order to achieve to diagnose the fault of overloading sensor in the Fly--By--Wire system, a method based on Local Mean Decomposition （LMD） and Support Vector Machine （SVM） is presented. First, the Fly--By--Wire model is built. Considering the non linear and non--stationary of the fault signal, the LMD method is applied to Signal Processing. Since there exists multi--band, strong noise disturb in the Fly--By--Wire system, adaptive filter is designed to eliminate given noisiness to make sure that the production function in- cludes the exact characteristic which reflects the state of the Sensor. At last, inputting the fault characteristic to SVM, the fault type is rec- ognized precisely. The simulation results show that this fault diagnosis approach is robust to multi--band, strong noise disturb in the Fly- By-Wire system.

关 键 词：电传系统 局域均值分解(LMD) 自适应滤波 支持向量机(SVM) 故障诊断 

分 类 号：TH873.7[机械工程—仪器科学与技术]