Research and Application of Deformation Prediction Model for Deep Foundation Pit Based on LSTM

Algorithm 6

Adam algorithm.

Require: set the initial learning rate to , and the exponential decay rate of moment estimation of the first and second orders is set to and in turn, and the value range of and is in the interval .

Require: set the initial parameter to , and initialized first and second orders are denoted as and , respectively, initialize time step

While stop condition not satisfied do

Randomly extract samples from the training set, and is the true value corresponding to .

Calculate the average gradient of samples:

Biased first-order moment estimation update:

Biased second-order moment estimation update:

First-order moment error correction:

Second-order moment error correction:

Update gradient value: (element operation)

Parameter updating:

End while