Multi-path Error Correction Method for Slope Monitoring Based on BP Neural Network

In order to solve the problem of multi-path error caused by complex environment in BDS slope monitoring, a method of multi-path error correction using Fourier transform and BP neural network is proposed to solve the problem of periodic fluctuation of monitoring results. Firstly, Fourier transform is used to analyze the spectrum of monitoring results and extract the period and frequency of periodic signals. Then BP neural network is used to model the results of multiple continuous cycles to generate the multi-path error model, which is used to correct the follow-up monitoring results to weaken the impact of multi-path error. The experiment shows that the error in the corrected monitoring results is reduced from ±3.2 mm to ±1.4 mm, and the time series of the monitoring results is stable without obvious periodic error, and the displacement trend of the monitored objects can be preserved.