Identification of nonlinearity in knocking vibration signals of large gas engine by deep learning

Nonlinear knocking combustion analysis based on vibration measurements of a Deutz MWM 8 cylinder large gas engine will be presented. The knocking nonlinearity is identified by K-entropy and deep learning technique. A framework for determining the knocking nonlinearity of the vibration signals is discussed using deep belief network (DBN) of the knocking probability sequences.