Research on artificial neural network-time series analysis of slope nonlinear displacement

As an explicit behavior of the complicated dynamic system, the displacements of slop are characterized with randomness and indetermination involving many uncertain factors; and the physical-based modeling is very difficult to fulfill prediction function. As an alternative, it was proved by many practical engineering cases that a set of displacement time series to predict the future displacement can be used. Based on the principles of artificial neural network and time series analysis, the BP network is established by zero mean method, standard deviation preprocess, regularization energy function, and Bayes-regularization to extract the trend term of displacement time series. After the extraction, the displacement time series becomes a balance series, which could be processed by normal ARMA model. In addition, combined with the real-time tracing algorithm, the artificial neural network-time series analysis(united modeling)for nonlinear displacement in geotechnical engineering was proposed. As a test, this modeling was used in displacement prediction of Geheyan Hydraulic Power Station intake slope and Dayantang slope in Shuibuya Hydraulic Power Station project. The results of engineering case indicate that it is reliable with high precision. It is proved that this modeling can be used to practical engineering.