Details

Autor(en) / Beteiligte

• Li, Juzhi
• Chen, Haiwei
• Zhou, Haonan

Titel

Improved BP neural network-based subsurface displacement prediction method

Ist Teil von

• 2023 IEEE 3rd International Conference on Power, Electronics and Computer Applications (ICPECA), 2023, p.1648-1653

Ort / Verlag

IEEE

Erscheinungsjahr

2023

Quelle

IEEE Xplore

Beschreibungen/Notizen

• Natural disasters such as landslides and mudslides are extremely harmful to society. In order to better monitor the subsurface displacement, this paper proposes to apply the Seagull optimization algorithm (SOA) to the Back Propagation Neural Network (BPNN), and use the SOA algorithm to optimize the parameters in the neural network algorithm to improve the accuracy and efficiency of its fitting prediction. The SOA algorithm is used to optimize the parameters of the neural network algorithm to improve the accuracy and efficiency of its fitting prediction. The subsurface displacement data provided by the landslide detection platform are cleaned and interpolated by three splines to expand the data set and weighted by the entropy method to enhance the differentiation ability among the indicators. After that, the data were brought into the SOA-BP model for prediction. The results show that the accuracy of the fit is improved by 33.07% compared with the conventional BP neural network, and the mean horizontal and vertical errors are kept within ±1mm. In terms of implementation efficiency, the displacement is obtained faster than the existing segmented linear interpolation method. It can be shown that the SOA-BP model is a better method for predicting subsurface displacement data and satisfies the requirements for subsurface displacement monitoring.