Digital sequence control method based on LFR

The invention discloses a digital sequence control method based on a location forgetting and re-encoding method (LFR). The method comprises the following steps of: encoding a composition part and a neural network part; for a given IQ digital sequence V, the digital code in each V is a new number which is coded according to a neural network input rule; the LFR is applied to each part of sequence Sfrom the first digit to the Nth digit, and the forgetting factor alpha is used for transmitting the previously forgotten digit to the next input variable by multiplying the forgetting factor alpha bythe factor; the sequence is recoded into L1, L2 and L3 through a formula with a forgetting factor, and finally all the sequences are converted into an input format of the neural network; The method uses a first model to represent LFR re-encoding of an input sequence of a training set of a neural network, uses a second model to represent LFR re-encoding of an input sequence of the training set of the neural network, and use