Identification of Shipboard Power Quality Disturbances Based on MTF-EfficientNet

A power quality disturbance identification method based on the combination of Markov transition field（MTF） and Efficient Net is proposed to address the problem of increasingly deteriorating power quality in ship power systems caused by the increased use of nonlinear loads, leading to difficulties in disturbance identification. Firstly, based on the basic principle of Markov transition field, the one-dimensional time series of the disturbance signal is reconstructed and transformed into a visualized two-dimensional image. Then the visualized image is input into the EfficientNet network for feature extraction. Finally, a linear classifier Softmax is used to classify the features and output the recognition results. Compared with different disturbance identification methods, the proposed method has the advantages of simple dimensional transformation and small model parameter quantity. In the absence of noise and with the addition of three different signal-to-noise ratios of noise, good disturbance classification and recognition characteristics can still be ensured, proving that the proposed method is suitable for complex operating conditions in ship power systems.