1. School of Earth and Environment,Anhui University of Science and Technology,Huainan 232001,China 2. Laboratory of Unconventional Gas,Institute of Exploration,Anhui Coalfield Geology Bureau,Hefei 230088,China 3. School of Resources and Earth Sciences,China University of Mining and Technology,Xuzhou 221116,China
Conventional coal seam gas content prediction methods are mostly based on logging constrained seismic attribute inversion and linear mapping model,which makes the prediction accuracy difficult to control,and severely limits the universality of the method.This paper starts with two aspects:seismic attribute optimization and BP neural network prediction model improvement.Using the Q-type clustering analysis method,the seismic attributes of the extracted target reservoirs are classified and optimized,and four kinds of seismic attributes with good correlation with the geological targets are obtained.The particle swarm optimization algorithm is further used to BP neural network algorithm.The connection weights of the input layer and the hidden layer and the threshold of the hidden layer are optimized,and the PSO-BP prediction model is constructed.The PSO-BP model is trained by using the preferred seismic attributes and gas content data of the well location.Based on the trained PSO-BP model,the coal seam gas content prediction in the study area is carried out with the preferred seismic attributes of the entire work area as input.The comparison between the predicted gas content of the well position and the measured results shows that the prediction method has high accuracy.Therefore,it can be considered that the PSO-BP prediction model and the corresponding prediction method flow can be effectively applied to the prediction of coal gas content in coal reservoirs.
Zhang R, Chen G, Pan B Z, et al. Prediction of coalbed methane content based on generalized regression neural network optimized by bacterial foraging[J]. Geophysical and Geochemical Exploration, 2016,40(2):327-332.
Xu K, Chen Z G, Zhao J H, et al. Layout design method of star sensor based on particle swarm optimization algorithm[J]. Journal of Jilin University:Engineering and Technology Edition, 2019,49(3):972-978.