To improve the computing efficiency of the traditional ant colony clustering algorithm, from data combination mechanism, an abstraction ant colony clustering algorithm was proposed. By analyzing the data of rock burst samples, from the engineering analogy thinking by the abstraction ant colony clustering algorithm, a new method for forecasting of rock burst in deep underground engineering was given. Some rock burst engineering applications of deep mines prove that compared with traditional ant colony clustering algorithm, for similar computing difficulty and complexity, the validity of the new algorithm whose accuracy is 92%, is higher, and the computing velocity can increase 40%. So the abstraction ant colony clustering algorithm is practical for rock burst forecasting.%为了提高仿生聚类算法的计算效率,对传统的蚁群聚类算法进行改进,采用数据合并机制提出了一种筛选蚁群聚类算法.采用筛选蚁群聚类算法,在岩爆工程实例资料分析的基础上,基于工程类比的思想判断岩爆的发生状态,提出了一种深部地下工程岩爆预测的新方法.通过多个煤矿深部地下工程岩爆实例的应用研究证明,和传统的蚁群聚类算法相比,在不显著增加计算难度和复杂度的情况下,新算法对工程实例的判断准确率更高,其准确率达92％,计算效果更好；而且,计算速度有较大幅度的提高,其计算时间缩短近40％,说明新算法的计算效率更高.因此,筛选蚁群聚类算法是一种比较实用的岩爆预测新方法,值得在深部地下工程岩爆研究领域推广应用.
展开▼
