The Application Research Of DNN Algorithm In Tailing Pond Safety Evaluation

Along with the development of social industry and economy, the demand for mineral resources is increasing rapidly, which make more tailing ponds are needed. Because total storage capacity and dam height will be rose during operations, tailing pond is changed into high-potential dangerous source, which poses a enormous threat to the safety of mine, people living downstream and natural surroundings. So an efficient and reasonable evaluation method of tailing pond is motivated to propose which could be helpful to predict the trends of tailing pond status.In order to learn the hidden information and internal relationship among the factors which have big influence on tailing pond safety, a new trend prediction method is designed based on deep neural network. The main results and achievements are summarized as follows:1) The statement and summary of the research of tailing pond safety evaluation methods are provided. According to comparing the advantage and weakness, methods in this domain are analyzed.2) The principle of deep neural network is introduced. Further more, this paper described the design solution about deep stacked auto-encoders including greedy layer-wise pre-training, LM-BP algorithm.3) In terms of the accidents statistic analysis and structural characteristics of tailing pond, the critical factors are proposed which come from the main cause of tailing pond accidents.4) The simulation experiments are carried out. Experiment results show that deep neural network has advantages for feature abstraction, representation learning and prediction accuracy rate.5) A safety evaluation software is designed in the end of this paper for the purpose for implementing the algorithm more conveniently.This paper validates that deep neural network is a reliable safety trend prediction method and provides a new idea and theoretical support to tailing pond safety evaluation.