Study on Blind Source Separation of Vibration Signals of IC Engine
Yuan Tao · Transactions of Csice · 2007
This paper studied the algorithm of multi-channel blind least mean square(MBLMS) based on the nongaussianity of sources by using Gray's variable norm as the cost function.To meet the practical requirement in vibration sources separation in IC engine,three simulated convolutive-mixture signals were generated to make the blind source separation.Compared with the sources,the separated signals keep all information except scale variation and time delaying.Ten vibration signals were simultaneously measured from the 4135 diesel engine,and reach three signals were chosen to make the blind source separation.Three separations attained different independent components.Analyzing these components showed that they reflected the vibration sources of piston slapping,exhaust valve closing and inlet valve closing.The separated results of the simulated signals and the actual IC engine vibration signals indicate that MBLMS is an effective algorithm for blind source separation.