An Optimized Vibration Signal Compressed Sensing Based on Phase Blocking K-SVD Algorithm for Marine Diesel Engine Cylinders

The condition of the marine diesel engine cylinder can be monitored by analyzing the vibration collected from the cylinder head. However, the large amount of vibration data brings a series of storage and transmission problems. Compressed sensing is an effective way to address these issues, but the classical K-SVD compressed sensing algorithm has a low compression ratio (CR) and poor reconstruction accuracy in this application. Considering the steady working phase of the marine diesel engine, a compressed sensing algorithm based on phase blocking K-SVD (PB-KSVD) is proposed. First, the vibration signal is divided into blocks based on the phase information extracted from the top dead center (TDC) signal. Then, the K-SVD algorithm is employed to train the corresponding dictionaries for each cylinder. Finally, the trained dictionaries are used to reconstruct the compressed data according to the phase blocks. Compared with the classical K-SVD compressed sensing method, PB-KSVD achieves high-precision reconstruction in both the time and frequency domains under a low CR.