基于Dempster-Shafer证据理论的柴油机故障诊断

Fault Diagnosis Approach for a Diesel Engine Based on Dempster-Shafer's Evidence Theory

在简述多传感器信息决策层融合暨Ｄｅｍｐｓｔｅｒ－Ｓｈａｆｅｒ证据理论的基础上，研究了决策层信息融合 的实现方法和算法，利用柴油机表面振动信号与高压油路压力信号所提供的特征信息进行融合处理，使 用决策规则对柴油机供油系统工作过程多种故障进行了诊断识别。通过分析、比较基于融合信息进行诊 断识别的结果与单传感器信息诊断识别的结果，说明了多传感器信息融合的诊断识别方法具有良好的 稳定性、精确性和容错性，能够有效地提高柴油机故障诊断的准确性和可靠性。

The conception incorporated with multisensor data fusion and the Dempster-Shafer' s evidence theory is discussed in brief. The features are extracted from vibration signals of the surfance of diesel engines and the pressure waveform of the fuel supply pipes. A fusion algorithm is stressed that uses the Dempster-Shafer's model to fuse the information from different sensors at the decision level. The theory and the fusion algorithm are used to recognize and classify the faults of the fuel injection system of a diesel engine. Successful results have been obtained in practice. The final results indicate that the multisensor data fusion has the stability and the ability of fault tolerance,and can improve the accuracy and relibility of fault diagnosis of diesel engines efficiently.