Fault States Diagnosis of Marine Diesel Engine Valve Based on a Modified VGG16 Transfer Learning Method

Abstract

The marine diesel engine is an important power machine for ships. Traditional machine learning methods for diesel engine fault diagnosis usually require a large amount of labeled training data, and the diagnosis performance may decline when encounters vibrational and environmental interference. A transfer learning convolutional neural network model based on VGG16 is introduced for diesel engine valve leakage fault diagnosis. The acquired diesel engine cylinder head vibration signal is first converted to time domain, frequency domain, and wavelet decomposition images. Secondly, the VGG16 deep convolutional neural network is pretrained using the ImageNet dataset. Subsequently, fine tuning the network based on the pretrained basic parameters and image enhancement methods. Finally, the well-trained model is adopted to train and test the target dataset. In addition, the cosine annealing learning rate setting method is used to make the learning rate close to the global optimal solution. Experimental results show that the proposed method has higher accuracy and better robustness against noise with a small sample dataset than traditional methods and deep learning models. This study not only demonstrates a novel view for the diagnosis of marine diesel engine valve leakage, but also provides an applicable diagnosis method for other similar issues.

Transfer Diagnosis Model of Internal Combustion Engine With Embedded Vibration Signal Impact Decomposition

“…This study adopts a DSAN framework to achieve transfer fault diagnosis, embedding the proposed signal decomposition layer and FVL constraint. The feature extraction module of the framework consists of the classical VGG structure [28], while the DA module employs LMMD as the distance loss function. For the diagnostic module, a classical softmax classifier is applied.…”

“…The parameters of the modified VGG module are shown in Table II. The model structure and parameters refer to the commonly used VGG structure [28]. It should be noted that for calculation convenience, the length of the input signal in this article was uniformly interpolated to 3072, and the training method used was train-on-batch, which randomly extracts a batch of samples from the dataset for one-step training.…”