Modeling method for predicting multi-factor induced landslide based on deep learning

Abstract

In order to improve the prediction accuracy of rainfall type landslide and overcome the problem that an existing prediction method is difficult to process a multi-factor nonlinear relation, the invention discloses a modeling method for predicting multi-factor induced landslide based on deep learning, a traditional DBN algorithm is improved, a momentum learning rate, Dropout and Softmax technologies are introduced, convergence difficulty or local optimum is avoided. The problem of overfitting is reduced, and nonlinear classification optimization and prediction of multiple influence factors causing rainfall-type landslide are realized. A simulation experiment result verifies the accuracy of the model provided by the invention, and a new thought is provided for rainfall type landslide prediction by using a deep learning method.

Description

technical field [0001] The invention relates to a modeling method for multi-factor induced landslide prediction based on deep learning. Background technique [0002] my country has a vast territory and complex and diverse geographical environment. Common geological disasters include six types of landslides, collapses, and debris flows. Among them, the number of landslide disasters accounts for more than 50% of the total number of natural disasters. Complex, with non-linear dynamic characteristics, the study of the factors that cause landslides and the establishment of predictive models are still hot topics in academic circles all over the world. Although there are many inducing factors that cause landslides, rainfall is the most basic triggering factor. The use of artificial intelligence technology to carry out research on rainfall-type landslide prediction is of great significance. [0003] The early research on landslide prediction was a model proposed by Japanese scholar ...

AI Technical Summary

Problems solved by technology

[0003] The early research on landslide prediction was a model proposed by Japanese scholar Saito Ditaka based on the creep failure theory in 1968. With the rapid application of mathematical theory in modeling, some scholars began to carry out the research object of debris flow velocity and flow. Research on computational modeling of dynamic characteristics, and with the rise of artificial intelligence technology, research on nonlinear models has gradually become a research hotspot. Biswajeet Pradhan used the backpropagation training method of neural networks to determine the sensitivity index of landslides. ChenJ et al. used genetic algorithms and neural networks. network to model the trend of geological disasters and realize the prediction of geological disasters. Poonam Kainthura et al. introduced geographic information system (GIS) into geological disaster prediction, and used K-means algorithm to create clusters that define different rainfall levels. ID3 decision tree for forecasting and early warning, Alvioli M studied the time and distribution of rainfall-induced shallow landslides, and constructed a grid-based regional slope stability model, and proposed a method for rainfall-triggered landslide early warning, Liu Yanhui made a study based on statistics Based on the analysis of the regional geological disaster early warning model, it is believed that the existing models have problems such as relatively single early warning indicators or complicated model calculations. network, most of these methods are highly subjective and focus on qualitative analysis rather than quantitative analysis, making it difficult to fully model and predict the multiple influencing factors and their complex nonlinear relationships that cause landslides

[0004] Deep learning is an important research direction in the field of artificial intelligence. Among them, deep belief network (deep belief network, DBN) was proposed by Hinton in 2006. It is formed by multiple restricted Boltzmann machine (RBM) stacks, which can be more It can well reflect the complex relationship between the factors that cause landslides. At present, DBN has been widely used in many fields such as speech recognition and image recognition. However, there are few studies on applying DBN to landslide disasters, mainly focusing on landslides. in the susceptibility study of

Images