Abstract: In this paper, the intelligent sprinkler engine is taken as the research object, and the neural network method is used for engine fault diagnosis.The random optimization algorithm is introduced to optimize the fault diagnosis mode of neural network, and the simulation experiment is carried out on the engine exhaust data, which is compared with the initial fault information code to determine the engine fault.The simulation results show that the optimized neural network fault diagnosis method can significantly reduce the error value in the diagnosis process, and the simulation results are highly consistent with the prediction information, which shows that the neural network fault diagnosis method optimized by stochastic optimization algorithm has high reliability.