ABSTRACT: In this paper the separation and identification of diesel engine noise is investigated using independent component analysis (ICA) and time frequency analysis (TFA). The separation and identification of engine real noise is important for noise reduction and faults diagnosis. This study can be considered as a guide for engine real noise separation and identification. A diesel engine noise is mathematically modeled and a single cylinder engine noise is simulated. The simulated noise sources are exhaust valve close, inlet valve close inlet valve close, fuel injection, combustion, piston slap, exhaust valve open and inlet valve open. The individual simulated engine noise sources signals are considered as acquired signals by seven channels microphone array. The mixed sources signals are separated using ICA. The separated independent components (ICs) are analyzed by short time Fourier transform (STFT) in time-frequency plane. The time is then converted to crank angle in order to identify engine noise sources. As predicted, the results show that the separated engine noise sources well correlated with the engine design specifications.