基于测井资料与优化通用向量机的煤层气含量预测模型

Prediction model of CBM content based on logging data and optimized general vector machine

为实现煤层含气量的高精度评价,合理地进行生产布局及高效勘探开发,以沁水煤田柿庄南区3号煤层含气量密闭取心实验数据为样本,提出了基于弹性网络优选测井曲线的改进的量子粒子群优化通用向量机混合预测模型EN-IQPSO-GVM。模型在煤层含气量测井响应特征和敏感性分析基础上,将弹性网络用于通用向量机模型特征输入参数的优选;提出了一种改进的量子粒子群算法优化GVM网络权值阈值,构建了基于弹性网络-改进量子粒子群算法的通用向量机煤层含气量预测模型;将该模型用于靶向区盲井煤层含气量预测,与相同优化策略下的支持向量机和BP神经网络模型及传统多元回归模型进行对比,分析该模型性能及适应性。结果表明:新模型盲井预测精度从21.83%减小到4.25%,具有更强的泛化能力,更适用于非均质性强的煤储层含气量高精度评价。

In order to achieve high-precision evaluation of CBM gas content, reasonable production layout and efficient exploration and development, taking the gas content of No.3 coal seam in Shizhuang south area of Qinshui Coalfield as a sample, an improved quantum particle swarm optimization general vector machine hybrid prediction model(EN-IQPSO-GVM) based on elastic network optimization logging curve is proposed in this paper. Firstly, based on the response characteristics and sensitivity analysis of coal seam gas content logging, the Elastic Network(EN) is used to optimize the feature input parameters of the general vector machine model. Then, an improved quantum particle swarm optimization(IQPSO) algorithm is proposed to optimize the GVM network weight threshold, and a general vector machine coal seam gas content prediction model based on elastic network and improved quantum particle swarm optimization algorithm is constructed. Finally, the model is used to predict the coal seam gas content of blind wells in the target area, and compared with the support vector machine, BP neural network model and the traditional multiple regression model under the same optimization strategy to analyze the performance and adaptability of the model. The results show that the blind well prediction accuracy of the new model is reduced from 21.83% to 4.25%, which has stronger generalization ability and is more suitable for the high-precision evaluation of gas content in highly heterogeneous coal reservoirs,and can lay a geological foundation for the exploration and development of coalbed methane target areas.