论文部分内容阅读
为满足数字矿山系统应用中对海量数据高效存取的技术要求,提出一种块段模型海量数据存储新方法。该存储技术满足数据库的特点:ACID兼容、并发支持、数据共享及高效访问;采用线性八叉树的方法组织块段模型,并存将其储于LMDB(快速内存映射数据库)中;通过定位码比较算法及从几何空间地址码到存储空间定位码的转换算法,可高效地对三维空间任意点的地质属性进行查询;采用三维到二维的转换及内外点二维网格扫描筛选法,使地质属性查询的三维问题转化为二维问题,其性能和鲁棒性得到显著提高。实验结果表明,这种方法能够高效存取块段模型海量数据,并满足数据库的特点;相比于采用etree方法,采用LMDB方法至少快3倍,特别是在读取数据时效率更高,且数据量越大,效果越明显。
In order to meet the technical requirements for efficient access to massive data in digital mining system applications, a new method for massive data storage of block models is proposed. The storage technology meets the characteristics of the database: ACID compatible, concurrent support, data sharing and efficient access; linear octree organized block model, stored in the LMDB (fast memory-mapped database); by positioning code comparison Algorithm and transformation algorithm from geometry space address code to storage space locating code, which can efficiently query the geological attributes at any point in the three-dimensional space. Using the three-dimensional to two-dimensional transformation and the two-dimensional grid scanning screening method of internal and external points, Attribute query three-dimensional problem into two-dimensional problem, its performance and robustness has been significantly improved. The experimental results show that this method can efficiently access massive data of the block model and satisfy the characteristics of the database. Compared with the etree method, the LMDB method is at least 3 times faster, and in particular, it is more efficient in reading data The greater the amount of data, the more obvious the effect.