Due to the complex and changeable working environment of the diesel engine,the signal of the injector contains more noise,which leads to a low diagnosis rate.In order to reduce the impact of noise and improve the fault diagnosis rate,a signal denoising method combining sparrow algorithm optimized variational mode decomposition (SSA-VMD) and improved time-frequency peak filtering (ITFPF) was proposed.Aiming at the problem of VMD restricted by decomposition parameters,the energy decomposition factor was used as the objective function,and the signal was adaptively decomposed into a series of IMF through SSA,which could solve the mode aliasing problem caused by improper parameter setting.Improper selection of TFPF window length would affect its filtering effect.Since permutation entropy could measure the complexity of non-stationary signals,it was used as the fitness function to find the optimal window length,so that ITFPF could take into account both signal fidelity and noise suppression.The simulation and experiment results show that the noise reduction effect of the method combined with SSA-VMD and ITFPF is better than other methods,and the fault recognition rate is increased by 10.13% comparing with the original VMD.