Manned spacecraft on-orbit oxygen consumption state automatic discrimination method based on unsupervised learning

The invention relates to a manned spacecraft on-orbit oxygen consumption state automatic discrimination method based on unsupervised deep learning, and the method comprises the steps: 1) transmitting original data into an encoder through a data encoding module 1-1, and obtaining corresponding hidden features; 2) sending the hidden features into a decoder through a data decoding module 1-2 to obtain reconstructed data; 3) clustering the hidden features output by the encoder by using a K-means algorithm through pseudo tag generation modules 1-3, and using a clustering result as a pseudo tag for automatically judging the on-orbit oxygen consumption state of the manned spacecraft; 4) updating network parameters of the encoder and the decoder by using a loss minimization function through parameter updating modules 1-4; wherein the loss function comprises similarity loss and reconstruction loss. According to the method, the dichotomy judgment problem of any pair of data is solved through unsupervised deep learning,