Research On Monitoring Method Of Landfill Pollution Diffusion Based On LSTM-BP

Landfill site contamination may cause secondary environmental problems such as groundwater and air pollution,threatening human health.Pollution diffusion monitoring is an important measure for risk management and control of landfill sites.Through site monitoring,contaminated areas can be found in time,providing an important basis for subsequent accurate site restoration and governance.The resistivity method has been widely used in the exploration of contaminated sites because of its advantages of non-destructiveness,high efficiency,and high spatial and temporal resolution.The inversion process of the resistivity method is actually a complex nonlinear solution problem.Conventional inversion methods are prone to fall into local extrema when solving problems,and the inversion results depend on initial model selection.BP（Back Propagation）neural network has excellent nonlinear mapping ability,which can effectively solve the problems encountered by conventional inversion methods.Long Short-Term Memory（LSTM）neural network solves the long-term dependence problem of ordinary Recurrent Neural Network（RNN）,and has achieved great success in time series forecasting in recent years.In this paper,the LSTM neural network and the BP neural network are combined to construct the LSTM-BP neural network,which realizes the monitoring of the occurrence,development and change trend of the pollution diffusion process.The main work done in this paper is as follows:（1）The high-density resistivity method is applied to the exploration of contaminated sites in landfills,and the Wenner device method is used to observe the apparent resistivity data;the finite element method is used to build a resistivity forward model to generate sample data for neural network training.（2）The two-dimensional inversion of resistivity is realized by using BP neural network,and the inversion results are compared with the results of least squares method,which verifies the feasibility of BP neural network for resistivity inversion.（3）The LSTM neural network is used to predict the apparent resistivity time series data,and the BP neural network is combined to invert the predicted apparent resistivity to monitor the pollution diffusion.（4）Using LSTM-BP neural network to predict and invert the long-term monitoring data of a hazardous waste landfill in Yancheng,it is confirmed that the LSTM-BP neural network is effective in monitoring pollution diffusion.The research in this paper shows that the use of LSTM-BP neural network can realize the pollution diffusion monitoring of landfill sites,and this method has certain engineering reference and application value.