地质结构的多点统计学三维重构是揭示地质过程、开展资源和能源预测等领域研究的重要基础和框架。多元数据集成是困扰高精度复杂地质结构多点统计学随机重构的瓶颈之一。以复杂地质结构三维重构为研究对象，针对多点统计学模拟中多元数据融合问题，本课题拟依托反演思想和交叉熵理论，从交叉熵函数和数据事件相似度两个角度建立多点统计学中多元数据融合的核函数，探讨面向多元数据融合的地质结构多点统计学三维随机模拟理论和方法体系，着重研究基于交叉熵损失函数的MPS目标函数构建及其对多元数据集成的响应特征，厘定多元数据耦合中制约地质结构三维随机模拟的关键影响因子，以揭示多元数据对地质结构多点统计学三维重构的响应规律，突破随机模拟中融合多元数据的理论和技术瓶颈，从而为大比例尺、高精度三维模型的构建提供新思路。

Three dimensional reconstruction of geological structures based on the multiple point statistics (MPS) method is a vital foundation and framework in many fields such as understanding geological process, energy and mineral resources prospective. In the MPS-based simulation, multivariate data integration is one of the bottlenecks in the research on reconstructing complex geological structures precisely with a high degree of accuracy. Taking 3D simulation of complex geological structure as the study subject, this project addresses in solving the problem of multivariate data integration in MPS-based simulation. From the viewpoints of cross-entropy and similarity of data event, this study will construct the kernel functions for integrating multivariate data in MPS simulation on the basis of inversion theory and cross-entropy theory. The theoretical study and framework of the MPS-based stochastic modeling of 3D geological structures will be carried out base on multivariate data integration. The MPS objective function that is based on the cross entropy loss function and its response to multivariate data integration are specifically focused in the study. The key factors restricting 3D stochastic modeling of geological structures in multivariate coupling will be determined. The response rule of MPS-based modeling of 3D geological structures to multivariate data integration will be revealed. The aim of this project is to break through the bottleneck theory and techniques of multivariate data integration in stochastic modeling, and to provide a new approach to build 3D complex geological structures precisely on large scale.