中文摘要：

基于输入输出信号趋势分析，提出基于形态学-小波的传感器故障检测与诊断的新算法。由不同宽度结构元素构成的改进型广义形态滤波器对输入输出信号进行滤波；采用小波多分辨分析对滤波后的信号进行分析，对故障的突变点进行准确定位；利用小波变换模极大值在多尺度上的表现与李普西兹（Lipschitz）指数的关系，对传感器各个类型故障进行识别。该文针对传感器死区、恒偏差、恒增益及漂移故障进行了仿真，仿真结果证明了该算法的有效性。

英文摘要：

A new algorithm for sensor fault detection and diagnosis basing on trend analysis of input and output signals was proposed. Generalized morphological filter with multi-structure elements was designed to filter the random noise and. impulse noise in sensor＇s input and output signals. Wavelet transform was used to decompose and analyze the filtered signals. By the multi resolution analysis （MRA）, the fault could be located accurately. The type of abrupt and incipient fault concerned could simultaneously be distinguished by using Lipschitz exponent, according to the fault＇s point of sudden change. The typical sensor faults such as fix, gain, bias, drift faults were studied. The simulation results show that this scheme is capable of locating accurately and diagnosing effectively.