中文摘要：

鉴于运用基本粗糙集理论获取与更新经验性知识时,进行增量式更新过程中数据库时空开销大的问题,提出一种基于改进差别矩阵的增量式规则获取方法。首先对原有的差别矩阵进行优化,使其具有比较小的空间复杂度;然后在分析新增对象与原决策系统之间关系的基础上,设计了求核及其增量式更新算法以及基于差别函数的属性约简与增量式更新算法,提高了核的更新效率;通过计算规则的精度,设定规则的阈值,形成了规则提取算法;最后以改模方案和故障诊断知识的获取与增量式更新作为实例,验证了该方法的正确性与可行性。

英文摘要：

The acquisition and incremental updating of the experience knowledge could be obtained by using basic rough sets theory, but the time and space costs of the algorithm would be very large. Aiming at these problems, a rules incremental updating algorithm based on discernibility matrix was built. The original discernibility matrix was optimized to make its space complexity become small. By analyzing the relationship between new object and old deci- sion table, the incremental updating algorithm for finding the core was built, and the incremental updating algorithm for attribution reduction based on discriminent function was designed. The updating efficiency of finding the core was increased. The incremental updating algorithm of rules was formed by calculating the accuracy of the rules and setting threshold of the rules. The validity and feasibility of the algorithm was verified by the acquisition and incre- mental updating of the mould repair schemes and fault diagnosis knowledge.