为提高煤层气含量预测的能力，提出了一种基于细菌觅食优化广义回归神经网络（ BFA⁃GRNN）的煤层气含量预测算法。利用已有煤层资料，通过神经网络建立回归模型，采用细菌觅食算法对模型参数进行优化，减少人为因素在网络训练中的影响。据此算法，在聚类分析及灰色关联分析的基础上，选取密度等共7个影响因素，建立煤层气含量预测的BFA⁃GRNN模型，通过实例分析验证该方法的可行性。结果表明：BFA⁃GRNN模型预测值与实测值之间相对误差小于6％，采用该模型预测煤层含气量具有较好的应用前景。%Coalbed methane is an important part of the natural gas energy, and determination of coal seam gas content is the key to the study of exploration and development of coal seam. In order to improve the capability of coal bed gas content prediction, this paper puts forward a kind of bacteria foraging optimization algorithm and generalized regression neural network ( BFA⁃GRNN) of the coalbed gas content prediction algorithm. Well logging data and core data of coal seam are used by neural network to establish regression model, bacterial foraging algorithm is used to optimize the model parameters, and artificial factor influences on determining the network struc⁃ture and the process of spreading factor are reduced. According to this algorithm and on the basis of clustering analysis and gray correla⁃tion analysis, seven main factors of coal bed gas content are chosen, which include density, resistivity, ash content etc. BFA⁃GRNN model is set ip by using the data of coal seam, and through the example analysis, the feasibility of this method is verified. The results show that the BFA⁃GRNN model is a true reflection of the nonlinear relationship between the coal seam gas content and the main control factors, and the relative error between predicted values and the measured values is less than 6%, suggesting that using the model to predict coal bed gas content has a good application prospect.
展开▼
