It is difficult to identify the state of a marine gas turbine because of the complicated structure and poor working environment. This paper puts forward a new method combining the kernel principal component analysis (KPCA) and fuzzy integral. First, the KPCA method is adopted to extract nine state characteristics parameters from the kernel principal components, such as the high pressure rotor speed, low pressure rotor speed, turbine exhaust temperature and casing vibration, to create a feature vector space. Then, the nuclear principal eigenvector was created based on the generalized regression neural network (GRNN) for an Elman neural network identification model to identify the gas turbine condition. Finally, the fuzzy integral is used to calculate the gas turbine state according to the result of two kinds of state recognition for policy makers. Research shows that the proposed method can effectively identify the gas turbine health and fault state of the ships by combining the key components, and has very good practical application value.