Marine Diesel Engine Bushing Fault Pattern Recognition Based on the Sensor Data and Improved Density Peaks Clustering Algorithm

Bushing fault pattern recognition is crucial for extending the lifespan of marine diesel engines. Density peak clustering (DPC) is widely used as an unsupervised learning method for fault pattern recognition. However, the DPC algorithm faces the problems of uneven local density distribution of data and sensitivity to parameter selection when dealing with axial tile fault diagnosis. To address the above problems, this manuscript introduces an unsupervised approach using the improved density peaks clustering (IAO-HDPC) for sensor data clustering and further applies this method to the data collected by the sensors for the bushing fault pattern recognition. Specifically, the proposed method first revises the allocation strategy of the clustering algorithm to address the issue of the DPC algorithm’s sensitivity to the local density of the data. Subsequently, the improved aquila optimizer (IAO) algorithm is employed to determine the optimal parameters for the clustering algorithm, thereby solving the challenge of parameter selection in the DPC algorithm. Finally, the experimental results demonstrate that the method achieves an average fault identification accuracy of 98%. Compared with the other four unsupervised algorithms, the proposed method achieves the best recognition results.