Application Of BP Neural Network Revise Kalman Filter In Predicting The Deformation Monitoring Of High Slope

Mining open-pit coal mine is a kind of extremely dangerous project, the high slope will be more and more high with the construction, and the internal stress is more and more unbalanced, therefore these will make high slope unstable. At the same time, there are rain, sun, weathering and other adverse factors, so the landslide would happen in the future. In order to forecast landslide, a lot of scholars study the high slope monitoring and have produced a lot of prediction models. But if the error of the input value is big, the prediction effect will be not very good. In order to solve such problem, we need to filter the data.Due to the instability of the statistical characteristics of the kalman filter, discrete phenomenon will happen. In order to solve the problem, in this paper we propose to use BP Natural Netwrok to revise kalman filter to filter the data. The trained BP Natural Netwrok applied to the kalman filter to smooth the data. At last, we predict the data.For the special geographical environment of the project of the slope monitoring in Shanxi Province, this paper designs construction scheme. And this paper presents the operation of the system, the submission of data.Finally through the test data, using the RMSE to compare BPKF algorithm and the standard KF algorithm. The result shows that the filtering results of BPKF algorithm is more smooth, it is more conducive to predict. And this paper combines with the scheme, the BPKF and the forecasting model, the project successfully predicted five landslides which includes one bigger and four smaller.