The invention discloses a QKD
decoy state and error
correction code rate joint optimization method based on
machine learning, and belongs to the technical field of
quantum communication, and the method comprises the steps: collecting or generating key data under QBER fluctuation, carrying out the
processing to obtain an original key
feature extraction value, inputting a
machine learning model for training, and carrying out the optimization of a QKD
decoy state and an error
correction code rate by taking the minimization of a
loss function as a target. And iteratively adjusting parameters to obtain a
machine learning predictor. Optimizing a
decoy state and an error
correction code rate by using a predictor; initializing a QKD
system, and optimizing a
signal state and the
decoy state according to an initial QBER predicted value; secret
key distribution is carried out, and original secret key data are collected and preprocessed to select an optimal error correction code; generating an expansion key, performing
error correction coding, and obtaining a statistical distribution value sequence; and inputting the QBER prediction value into a
machine learning predictor to obtain a QBER prediction value, and performing joint optimization on a
decoy state and an error correction
code rate of the
system by using the QBER prediction value. The QBER prediction is introduced in the information coordination process, the QBER prediction value is used for
decoy state optimization, and the security key rate of the QKD
system is improved.