融合电控柴油机传感器数据的贝叶斯综合故障诊断网络研究

Bayesian integrative fault diagnostic network integration with sensor data of electrical diesel engine

针对电控柴油机故障源多样性和不确定性的问题,对故障现象、状态数据提取及处理、贝叶斯故障诊断网络、故障源的先验概率获取、故障源的确定等方面进行了研究,在故障现象出现的情况下对如何有效确定故障源进行了分析和归纳,构建了融合传感器数据的电控柴油机贝叶斯网络综合故障诊断模型,提出了使用传感器数据技术检测电控柴油机工作状态,并结合经验法等估算出各类故障源的先验概率。通过运用贝叶斯网络技术推断查找到故障源的方法,并以丰田1KZ电控柴油发动机为实验对象,使用Hugin Expert工具对该诊断网络进行了推理验证。研究结果表明,该诊断网络充分发挥了传感器数据诊断技术的实时性和贝叶斯网络技术的判断决策能力,有效提高了电控柴油机故障诊断的正确率和实效性。

Aiming at the uncertainty of the problem and electronically controlled diesel engine fault source diversity of symptoms, data extraction and processing, a priori probability Bayesian network fault diagnosis, fault source of acquisition, such as failure to determine the source areas were studied, in case of failure phenomena on how to effectively determine the source of the problem was analyzed and summarized, the diesel engine Bayesian network integrated fault diagnosis model was constructed, it was proposed to detect the use of sensor data electronically controlled diesel engine operating state. Combined with experience method, a prior probability of various types of fault sources was estimated. Using a Bayesian network inference techniques, the source of the fault approach was found and Toyota 1KZ electronically controlled diesel engine as experimental subjects, using Hugin Expert tool for the diagnosis of network were reasoning verified. The results indicate that the diagnosis of network decision-making ability into full play the sensor data to determine real-time diagnostic techniques and Bayesian network technology,to improve the accuracy of diesel engine fault diagnosis and effective.