基于K2算法的发动机冷试贝叶斯网络模型研发

Development of Bayesian network model for engine cold test based on K2 algorithm

发动机冷试测试较传统热试方法可减少燃油消耗及排放,为提高冷试测试故障诊断精度,提出使用K2算法构建贝叶斯网络故障诊断模型。选取多台柴油机的相关冷试测试数据,分别构建基于专家知识和基于K2算法的贝叶斯网络故障诊断模型进行比较,结果表明,使用K2算法构建贝叶斯网络优于通过专家知识构建的贝叶斯网络,所提方案可以优化目前的故障诊断模型。

Engine cold test can reduce fuel consumption and emissions compared to traditional hot test methods.In order to improve the accuracy of fault diagnosis in cold test,a Bayesian network fault diagnosis model using K2 algorithm was proposed.Firstly,relevant cold test data from multiple diesel engines were selected to construct Bayesian network fault diagnosis models based on expert knowledge and K2 algorithm for comparison.The results showed that using K2 algorithm to construct Bayesian network was better than using expert knowledge to construct Bayesian network,and the proposed solution could optimize the current fault diagnosis model.