The invention discloses a
satellite power supply
system fault diagnosis method based on multi-strategy fusion
deep learning, and belongs to the technical field of fault diagnosis. The defects of poor precision and low stability of a fault diagnosis model caused by complex working conditions of an existing
satellite power supply
system are overcome. The method is based on DETamp; according to the
satellite power supply
system topology of the S4R, Dymolamp is used;
modelica software carries out modeling
simulation, and data sampling points are set as output and input ports of each device; faults of all levels are injected into the model, sampling
voltage and current during fault
simulation are recorded, and a fault sample
data set is constructed; sample features are extracted by using a
convolutional neural network, a low-dimensional sample
data set is constructed, and the calculation complexity is reduced; a bidirectional gating circulation unit is improved based on the combination of an adaptive lifting
algorithm and multi-head attention enhancement, hyper-parameters and network parameters of a BIGRU are solved by using a white shark optimization
algorithm WSO, so that the precision and training efficiency of a diagnosis model are improved, and a strong learner multi-classification fault diagnosis model based on multi-strategy fusion
deep learning is established. And finally, fault diagnosis is carried out, and diagnosis results including fault diagnosis precision, diagnosis time, fault positions, fault types and the like are output. The method is suitable for fault diagnosis of the satellite power supply system.