Soil stress-strain relation determination method based on improved LSTM deep learning method

Abstract

The invention discloses a soil stress-strain relationship determination method based on an improved LSTM deep learning method. The determination method comprises the following steps: preparing soil samples with different physical and mechanical parameters; establishing an original data set of stress and strain; performing normalization processing on the original data set; establishing a four-layer LSTM deep learning network, and determining a hidden layer node number Nh, an activation function and a cost function J; determining an initial weight matrix and a vector of the LSTM deep learning network and an initial parameter of a hybrid activation function; updating the weight matrix by using a modified Adam momentum gradient descent algorithm, determining a cost function descent amplitude Jd, and updating an activation function parameter; and repeating iteration until the cost function J of the LSTM deep learning network is smaller than a preset value or reaches a preset number of iteration training times Iter. According to the determination method, the nonlinear relationship between the stress and the strain is extracted and determined from the experimental data, and the time-dependent characteristics of the stress-strain behavior of the soil body can be considered. The determination method is simple, practical and convenient to popularize, and has great application value.

Description

technical field [0001] The invention relates to the field of soil constitutive relations, in particular to a method for determining the stress-strain relationship of soil based on the improved LSTM deep learning method. Background technique [0002] Soil is the carrier of geotechnical infrastructure. Determining the nonlinear mechanical response of soil under load conditions is of great significance to the design and construction of infrastructure. Due to the complex internal structure and various components of the soil, the soil has complex nonlinear deformation characteristics under the action of external loads. To determine the nonlinear mechanical behavior of soil, the current mathematical model describing the nonlinear mechanical behavior of soil is called constitutive model. According to different research methods, soil constitutive models can be divided into traditional theoretical models and neural network constitutive models. The traditional constitutive model usu...

AI Technical Summary

Problems solved by technology

However, the above methods do not consider the time-dependent characteristics of soil stress-strain behavior, and cannot describe the influence of historical stress-strain on current stress-strain, so there are large deviations in the results

At present, there is no neural network stress-strain determination method that can consider the long-term time characteristics of stress-strain

Images