Weighted belief function of sensor data fusion in engine fault diagnosis

Abstract

Fault diagnosis (the process of finding out whether system or equipment is in fault and where the corresponding fault is by using various inspection and testing method) on the engine is a typical information fusion (the process of integrating multiple data sources to produce more consistent, accurate, and useful information than that provided by any individual data source) problem where the information can be obtained from engine vibration, temperature, pressure, etc. Due to the efficiency of data fusion, Dempster–Shafer evidence theory is widely used in fault diagnosis. One key step to using evidence theory is to obtain the so-called basic probability assignment (BPA), or belief function. In this article, a new mathematical framework is presented to determine weighted BPA (WBPA). This WBPA function is obtained by weighting the distance between sample data and empirical data. With the assumption that the empirical data are normally distributed, the weighting factor can be determined. Then, the WBPA can be combined with D–S evidence theory to determine the status of the engine. Finally, a case in fault diagnosis and comparison with Song and Jiang (Adv Mech Eng 8(10):1–16, 2016) method illustrate the efficiency of the proposed method.