A Fault Diagnosis Method for Diesel Engine Cylinder Liner Wear Based on CWT-CNN and BKA-LSSVM

The accurate diagnosis of diesel engine cylinder liner wear is crucial for equipment operational safety. However, under complex operating conditions with strong noise interference, traditional diagnostic methods often suffer from insufficient accuracy and reliability. To address this issue, this paper proposes a novel fault diagnosis framework combining CWT-CNN and BKA-LSSVM. The method first utilizes CWT’s multi-resolution analysis capability to extract time-frequency features and designs a deep CNN network with multi-scale feature fusion for automatic deep feature extraction. Subsequently, an improved BKA algorithm is introduced to optimize the LSSVM classifier, along with a new adaptive parameter adjustment strategy. Validation results on an actual diesel engine test platform demonstrate that this method exhibits superior noise resistance and generalization capability compared to existing methods, achieving an average diagnostic accuracy of 99.6% under complex operating conditions, significantly outperforming traditional approaches.