Coal bump prediction is one of the key problems in deep coal mining engineering. To predict coal bump disaster accurately and reliably, we propose a depth neural network (DNN) prediction model based on the dropout method and improved Adam algorithm. The coal bump accident examples were counted in order to analyze the influencing factors, characteristics, and causes of this type of accidents. Finally, four indexes of maximum tangential stress of surrounding rock, uniaxial compressive strength of rock, uniaxial tensile strength of rock, and elastic energy of rock are selected to form the prediction index system of coal bump. Based on the research results of rock burst, 305 groups of rock burst engineering case data are collected as the sample data of coal bump prediction, and then, the prediction model based on a dropout and improved Adam-based deep neural network (DA-DNN) is established by using deep learning technology. The DA-DNN model avoids the problem of determining the index weight, is completely data-driven, reduces the influence of human factors, and can realize the learning of complex and subtle deep relationships in incomplete, imprecise, and noisy limited data sets. A coal mine in Shanxi Province is used to predict coal bump with the improved depth learning method. The prediction results verify the effectiveness and correctness of the DA-DNN coal bump prediction model. Finally, it is proved that the model can effectively provide a scientific basis for coal bump prediction of similar projects.
Genetic Optimization Method of Pantograph and Catenary Comprehensive Monitor Status Prediction Model Based on Adadelta Deep Neural Network
