基于人工免疫算法的数据压缩技术在滑坡异常提取中的应用研究

滑坡异常信息提取对滑坡预测预报至关重要。以四川丹巴滑坡为例，应用人工免疫算法对滑坡镜2、镜6和镜9的位移监测数据进行压缩，分别获得最优压缩率为0．821，0．819，0．829，保真度为0．95。并在置信水平为0．95的条件下，对压缩后数据进行异常提取，分别获得了3个监测点的4个异常信息。这些异常信息所反映的滑坡变形发展演化规律与滑坡声发射等其他监测信息所反映的规律基本保持一致，说明应用人工免疫算法对滑坡位移监测数据进行压缩处理可以在保持原有数据的特性与规律的同时，有效除去监测数据中的冗余与不相关信息，增加有效数据密度，有利于准确提取滑坡异常信息。

Application of data compression, based on AIS to the extraction of landslide anomaly

The extraction good reflection to lands landslide from complex immune algorithm is algorithm is used to of anomaly is a very impor lide. Therefore, how to data information is a key proposed compress rant technique for landslide forecasting owing effectively problem. extr act the anomaly really relat applied method-- the art for data compression in this paper. In ca the displacement data of the monitoring to its ed to ificial se study, the artificial immune point 2, 6 and 9 of the Danba landslide in Sichuan, China. And three data sets got rid of the impacts of redundancy and irrelevancy are obtained, whose optimal compression rates are 0. 821, 0. 819, 0. 829 respectively and fidelity is 0.95. Then, the anomaly recognition method is used to the compressed data, and four obvious anomalies are obtained under the confidence level of 0.95. The anomalies reflecting to landslide are in accordance with the other monitoring information such as acoustic emission of rocks. So, it can be easily concluded that the AIA algorithm could be used to compress the monitoring data with higher data density and lower redundancy and irrelevancy.