Improved Clustering Algorithms Used in Diesel Engine Vibration Fault Diagnosis Based on Bayesian Networks

The work presented in this chapter focuses on diesel engine vibration fault diagnosis. The vibration fault of diesel engine has the property of randomness and layers, and its fault information has the features of uncertainty and non-integrality. The chapter studied the method of typical diesel engine vibration fault diagnosis based on Bayesian networks (BNs), which used expert knowledge to determine conditional probability, converted fault information into numeric data, and then established the Bayesian network model. The clustering algorithms were improved to reduce calculation work and enhance the accuracy in the diagnosis by way of optimizing correlation value between Bayesian network nodes. Simulation and experimental results verified the effectiveness of the improved clustering algorithms.