Discrete wavelet transform based data trend prediction for marine diesel engine

In this paper, a multi-model data trend prediction method is proposed for marine diesel engine to the prognosis of faults. According to the data characteristics, the discrete wavelet transform is used to process the data, which can eliminate the noise of the high-frequency and retain the low-frequency signal. The auto-regression, the gray model, the BP neural network and the radial-based neural network methods are employed to trend prediction and the results are compared. In terms of convergence speed, the autoregressive model has the best performance of the fault prognosis. In terms of fitting error, the neural network model has the best accuracy.