Research On Fault Diagnosis Method Of DC-DC Circuit Based On Feature Fusion

DC-DC circuits play a key role in the stable operation of smart grid,new energy vehicles,and medical equipment as the basic devices,so the operational reliability of DC-DC circuits is becoming more and more demanding.Accurate DC-DC circuit fault diagnosis can effectively reduce the loss and maintenance cost caused by circuit failure.In this paper,we study a feature fusion-based fault diagnosis method for DC-DC circuits,using Buck circuits and Boost circuits as objects to explore the parametric faults of DC-DC circuits due to electrolytic capacitor degradation.The research in this paper is as follows.Briefly analyze the basic principles of Buck and Boost circuits,build Buck simulation circuits and Boost physical circuits according to different fault types of capacitor parameter degradation,collect the output voltages under different fault modes as fault signals,and explore the research of three DC-DC circuit fault diagnosis methods based on feature fusion.VMD-MHHO-BPNN-based fault diagnosis method for DC-DC circuits.The method uses the VMD algorithm to decompose the input signal according to a preset decomposition scale to obtain multiple IMF components;for the spurious component problem generated after the VMD decomposition,the correlation coefficient between the IMF components and the original signal is calculated,the components below the threshold are removed and the remaining components are reconstructed to extract the fusion features in the time and frequency domains;the fusion features are input to the BPNN for training.To improve the classification capability of BPNN,the MHHO algorithm is used to find the optimization of the weights and threshold parameters of BPNN,and the fault diagnosis is performed using the optimized BPNN.Through experiments on two typical DC-DC circuits,Buck circuit and Boost circuit,it is verified that the method can effectively solve the spurious component problem of VMD algorithm,and the MHHO algorithm can improve the classification ability of BPNN.A dual-channel fusion-based fault diagnosis method for DC-DC circuits.The method constructs a dual-input single-output 1DCNN and BiGRU dual-channel fusion model,performs adaptive feature extraction of the input signal,and forms a new feature vector through the feature fusion layer,which is input to the Softmax layer to complete the DC-DC circuit fault diagnosis.The experiments on Buck circuit and Boost circuit show that the method can effectively explore the feature connection between the deep levels of the input signals,thus improving the fault diagnosis.A DC-DC fault diagnosis method based on IGWO optimized 1DCNN-BiGRU network.The method integrates the special features of 1DCNN and BiGRU networks that are good at extracting the spatial and temporal characteristics of the input signal,and constructs a single-input,single-output 1DCNN-BiGRU network model.For the hyperparameter problem of the network model,it is converted into an objective function optimization problem,and the optimal hyperparameter combination of the model is quickly found using the optimization capability of IGWO.The optimized 1DCNN-BiGRU network is used to train the input signal and input to the Softmax model for fault identification.The experimental results show that the method can effectively solve the problem of selecting hyperparameters in the network model,and the method also shows good diagnostic performance in a noisy environment.Figure [30] Table [40] Reference [78]...