基于ART-并行BP神经网络的柴油机故障诊断研究

Research on Fault Diagnosis of Diesel Engine Based on ART-BP Neural Network

造成柴油机故障的因素十分复杂,既存在单一类的故障,也存在多故障并存的现象,而且还会出现新型故障,仅仅依靠单一神经网络技术的故障诊断已经很难满足对柴油机的有效诊断要求。本文在信息决策层融合的基础上,以自适应谐振理论ART和误差反向传播并行BP两种神经网络为基础,建立了用于柴油机故障诊断的新型神经网络模型,以对柴油机系统工作过程多种故障进行诊断识别。通过与单一神经网络诊断识别结果的分析和比较,验证了该神经网络诊断模型的可行性,它能够进行多传感器信息综合诊断,既能识别单故障和并发故障,又具有识别新型故障的能力,可有效地提高对柴油机故障诊断的准确性和可靠性。

Factors causing diesel engine faults are very complicated and it is difficult to carry out effective diagnosis using the diagnosis technique that relies only on neural network because there may be both single fault anti multiple faults as well as new faults in a diesel engine. This paper established a neural network model for diagnosing diesel engine faults using the adaptive resonance theory（ART） and the back propagation（Be） neural network in order to diagnose and identify the multiple faults that occur during the operation of a diesel engine. The feasibility of the new neural network model was verified by analysing and comparing the diagnosis results gained from a single neural network model. The new neural network model can perform synthetic diagnosis of multi-sensor information because of its ability to identify a single fault, multiple faults and new faults, thus enhancing the accuracy and reliability of diesel engine fault diagnosis.