Abstract:

Deformation monitoring is an important part of the safety engineering construction and management, and it runs through the design, construction and operation of the project. It is of great practical significance to process the monitored settlement data, predict the settlement amount, and make early warning of the safety of the project. Based on the GM (1,1) grey model, wavelet analysis and neural network combination of related theories, using Matlab programming, this paper establishes a gray-wavelet neural network deformation prediction network model. Combined with engineering examples, the established deformation prediction network model is applied to the accumulated settlement observation data. The results show that the combined model has a very stable forecasting effect and is more accurate than the single GM(1,1) gray model. The more training samples, the better the fitting effect and the prediction is more in line with the actual situation.

Key words: deformation monitoring, GM(1,1) gray model, wavelet neural network, deformation prediction, subway settlement