可变模糊识别方法在滑坡稳定性评价中的应用

Application of variable fuzzy pattern recognition method on landslide stability assessment

针对滑坡发生的复杂性,该文把可变模糊集理论模型应用到滑坡稳定性评价中。选取4个层次17个因素建立了滑坡稳定性的多层次多指标评价单元系统。采用三级权重和二元比较模糊量化相结合的方法确定各影响因素权重。利用可变模糊识别模型对丹巴滑坡、老金山滑坡、印江岩口滑坡进行综合评价,评价结果与实际情况相吻合;并得到丹巴滑坡从20世纪80年代到2007年的稳定性先减弱后增强,在2004年3月到2005年3月处于最不稳定状态。可变模糊评价法与模糊综合评价法分别对丹巴滑坡a时期(20世纪80年代至1998年)、d时期(2005年3月到2007年)进行评价结果都为III级,验证了滑坡可变模糊识别评价模型的可行性。可变模糊模型利用动态可变的相对隶属度和级别特征值对丹巴滑坡a时期、d时期的特征值评价结果分别为3.27和3.44,评价出丹巴d时期稳定性比a时期稳定性差,进一步说明了可变模糊识别模型的优越性。滑坡稳定性评价可为滑坡灾害的防治和预警提供可靠的科学依据。

Landslides occur frequently and lose greatly, especially in China, and always have the characteristics of large scale, complex mechanism as well as serious destruction. It is important to assess landslide stability for understanding the evolution process of the landslide. In this paper, the recognition method of variable fuzzy pattern was applied to evaluate landslide stability. The method was based on the theory of variable fuzzy set, relative membership degree function and multi-level and multi-index evaluation unit system on landslide stability which included both the qualitative and quantitative factors. In order to overcome the limitations of the fuzzy comprehensive evaluation method which had a maximum membership degree, the relative membership degree function and the grade eigenvalue were used in the recognition method of variable fuzzy pattern. For the complexity and fuzziness of landslide system, the multi-level and multi-index evaluation unit system was set up by selecting 4 levels and 17 factors. In the purpose of confirming the fuzzy weights, the comprehensive evaluation of the three-level weights was adopted and the method of dualistic factor contrast combined with fuzzy mood operator was also used. The recognition method of variable fuzzy pattern was adopted to appraise the stability of Danba Landslide, Laojinshan Landslide and Yankou Landslide, and the results of the evaluation were consistent with the investigation. Depending on the different evaluation results of the Danba Landslide in 4 periods, the impact of the stability of Danba Landslide was semi-quantitatively evaluated. The stability declined at first and later improved from 1980s to 2007, and was in the most unstable state from March 2004 to March 2005. By comparing the stabilities of Danba Landslide during 1999-2004 and from March 2004 to March 2005 with that during 1980s-1998, the stabilities declined by 6.4% and 10.1%, respectively. And the stability of the landslide in period d （from March 2005 to 2007） improved by 4.7%compared with period c （from March 2004 to March 2005）, and decreased by 5.2%compared with period a （from 1980s to 1998）. In order to compare present method with fuzzy comprehensive evaluation method, two cases of Danba Landslide were studied in period a （from 1980s to 1998） and d （from March 2005 to 2007）. The results evaluated by variable fuzzy evaluation method and fuzzy comprehensive evaluation method were both in Level three. The stability eigenvalues of Danba Landslide in period a and d were evaluated by variable fuzzy evaluation method, and the two eigenvalues were 3.27 and 3.44, respectively. Stability of Danba Landslide in period d was worse than in period a, although they belonged to the same grade. In this study, the average eigenvalue which could make the error smaller was obtained. Actually, the eigenvalue was obtained by a series of different parameters. Therefore, the recognition method of variable fuzzy pattern is more feasible and reliable, and it is much stricter in the superior grade. Just as any other traditional evaluations, the recognition method of variable fuzzy pattern has a difficult problem that the qualitative factors cannot be completely quantified. But the qualitative indices can be converted into semi-quantitative indices by the recognition method of variable fuzzy pattern. Besides, the reliability of evaluation results can be guaranteed because of reasonable selection of evaluation indices and the accuracy of specific numerical indicators. The recognition method of variable fuzzy pattern is a promising evaluation approach in Landslide stability assessment, and it can provide scientific basis for landslide disaster prevention and control.