A method for predicting seepage pressure of earth-rock dam based on dual-driven mechanism and data

Abstract

The present invention proposes a method for predicting seepage pressure of earth-rock dams based on dual mechanism and data drive, comprising the following steps: generating seepage pressure data under multiple working conditions, optimizing the radial basis function neural network through the Aurora optimization algorithm to construct an efficient proxy model; based on the prediction results of the proxy model, fitting with measured data to establish a seepage pressure prediction mechanism model based on a hysteresis effect function; using the predicted values of the mechanism model as labels, supervising the training model to learn the physical laws of the seepage field; freezing the physical feature extraction layer in the trained model, and using measured data to fine-tune the time series modeling layer to achieve approximation to the measured values. The present invention organically combines mechanism drive and data drive, retaining the interpretability of physical laws while using measured data to supplement complex factors not considered by the mechanism model. While ensuring the model's prediction accuracy, it significantly improves the generalization ability of the deep learning model under extreme working conditions.

Images