Abstract:To achieve precise prediction and timely stability assessment of slope deformations in opencast mines, this study utilizes time series data of slope deformations, employing the mutual information method and the cross-correlation (C-C) method to determine the optimal embedding dimension and time delay for phase space reconstruction of the time series. Through phase space reconstruction, the dynamic evolution law of the nearest neighbor phase points in the time series of slope deformations is thoroughly analyzed. On this basis, the Long Short-Term Memory (LSTM) artificial neural network is combined to predict the time series of slope deformations. The results indicate that, under the condition of phase space reconstruction, an increase in the average distance between the nearest neighbor phase points signifies more prominent chaotic characteristics of slope deformations and a more complex underlying dynamic system. The LSTM deep learning network demonstrates high accuracy in short-term predictions, effectively predicting slope deformations. Based on the evolution law of the nearest neighbor phase points and the deformation prediction results, the stability state of the slope can be accurately identified, providing new ideas and methods for slope stability assessment during opencast mining.