ANN- and GIS-Based Regional Prediction of Cover-Collapse Probability: A Case Study in West Part of Guilin City
-
摘要: 岩溶地面塌陷是岩溶区常见的一种地质灾害, 塌陷区域预测是进行国土规划、资源开发与灾害防治的必要工作.由于岩溶塌陷的影响因素众多且相互作用, 发展过程复杂, 加之各评价因子的数值获取困难, 致使长期以来塌陷区域定量预测成为一个难以解决的课题.现行的区域预测模型不能描述塌陷形成模式的非线性特征, 也难以克服评价因子权重确定过程中人为经验因素的影响.神经网络技术的自学习、自适应与高度非线性映射特点显示了其在塌陷区域预测领域中应用的前景.根据研究区内地面塌陷空间聚集分布的特征, 提出了不同因子组合条件下塌陷发生可能性的定量化方法, 结合选定的评价因子类别确定了神经网络预测模型的结构, 利用312个塌陷点样本中的292个进行网络训练, 余下的20个样本的校验结果表明该模型具有较高的可信度.运用GIS技术将研究区进行评价单元划分, 并获取各评价因子的取值, 输入到训练好的网络中进行预测.将各单元的输出值进行归并处理后得到研究区岩溶塌陷的稳定级分区图.Abstract: Cover-collapse is one of frequent geological hazards in karst zone. Predicting the probability of cover-collapse occurrence is a requisite task for territorial planning, resource exploiting and hazard harnessing. Regional quantitative prediction of cover-collapse probability has become very intricate problem for the following reasons: (1)interactions between many influencing factors; (2)a sophisticated developing process; (3) difficulty associated with the value acquisition of factors. Some recent prediction models can not display the nonlinear characteristicsof collapse development pattern, nor can they eliminate the impact of empiricism during the course of weights allocation. Three major characteristics of artifical neural network(ANN)technology, i.e. self-learning, self-adapting and nonlinear mapping, indicate a powerful application potential in collapse prediction field. This paper reports the methodology of developing an ANN model to predict cover-collapse occurrence probability. An approach was established to measure the relative probability of the collapse corresponding to certain factor combinations, and some impact factors were specified. Consequently, the structure of ANN prediction model was created.292 stochastic collapse samples from the sample aggregate, of which the size was 312, were used to train the ANN model. The testing results of the 20 remaining samples show that this model has a good precision. Evaluation grids division and their value acquisition of every factor were accomplished with the aid of GIS software tools. The unstable probability of each grid was calculated through the trained ANN model, which enables us to delineate the different stable zones in the study area.
-
Key words:
- cover-collapse /
- artifical neural network(ANN) /
- nonlinear /
- prediction model /
- GIS
-
图 1 研究区塌陷点的模式分析
P代表任一塌陷点周围L m邻域范围内存在另外n个塌陷点的平均概率
Fig. 1. Pattern analysis of cover-collapse plot in study area
图 3 岩溶地面塌陷预测分区
1.稳定区; 2.基本稳定区; 3.中等稳定区; 4.不稳定区; 5.极不稳定区; 6.非评价区(基岩裸露区)
Fig. 3. Zones of cover-collapse probability prediction
表 1 区域岩溶塌陷预测评价因子选择
Table 1. Selected factors for prediction of regional cover-collapse probability
表 2 评价因子对于塌陷的相对影响程度
Table 2. Relative contribution of factors to cover-collapse probability
表 3 岩溶塌陷发生危险性分级
Table 3. Target value classes of cover-collapse probability
-
[1] Goodings D J, Abdulla W A. Stability charts for predicting sinkholes in weakly cemented sand over karst limestone [J]. Engineering Geology, 2002, 65: 179-184. doi: 10.1016/S0013-7952(01)00127-2 [2] 陈星, 黄润秋. 岩溶塌陷的地质概化模型[J]. 水文地质与工程地质, 2002, (6): 30-34. https://www.cnki.com.cn/Article/CJFDTOTAL-SWDG200206009.htmCHEN X, HUANG R Q. Geological conceptive models of karst collapse[J]. Hydrogeology and Engineering Geology, 2002, (6): 30-34. https://www.cnki.com.cn/Article/CJFDTOTAL-SWDG200206009.htm [3] 张发旺, 贾秀梅, 赵华. 灰色统计方法及其在岩溶塌陷预测分析中的应用[J]. 河北地质学院学报, 1996, 19(2): 144-150. https://www.cnki.com.cn/Article/CJFDTOTAL-HBDX602.005.htmZHANG F W, JIA X M, ZHAO H. Application of gray system statistic method in the field of karst collapse prediction analysis[J]. Journal of Hebei College of Geology, 1996, 19(2): 144-150. https://www.cnki.com.cn/Article/CJFDTOTAL-HBDX602.005.htm [4] 陈学军, 陈植华, 陈先华, 等. 桂林西城区岩溶塌陷模糊层次综合评价[J]. 桂林工学院学报, 2000, 20(1): 112-116. https://www.cnki.com.cn/Article/CJFDTOTAL-GLGX200002003.htmCHEN X J, CHEN Z H, CHEN X H, et al. Fuzzy-hierarchy prediction of karst collapse in the west area of Guilin city[J]. Journal of Guilin Institute of Technology, 2000, 20(1): 112-116. https://www.cnki.com.cn/Article/CJFDTOTAL-GLGX200002003.htm [5] Kaufmann O, Quinif Y. Geohazard map of cover-collapse sinkholes in the" Torunaisis" area, southern Belgium[J]. Engineering Geology, 2002, 65: 117-124. doi: 10.1016/S0013-7952(01)00118-1 [6] 贺玉龙, 杨立中, 黄涛. 人工神经网络在岩溶塌陷预测中的应用研究[J]. 中国地质灾害与防治学报, 1999, 10 (4): 86-90. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGDH199904013.htmHE Y L, YANG L Z, HUANG T. Research on application of artificial neural network in the prediction of karst collapse[J]. Chinese Journal of Geological Hazard and Control, 1999, 10(4): 86-90. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGDH199904013.htm [7] 包惠明, 胡长顺. 岩溶地面塌陷神经网络预测[J]. 工程地质学报, 2002, 10(3): 299-304. doi: 10.3969/j.issn.1004-9665.2002.03.014BAO H M, HU C S. Neural network prediction of karstic ground collapse[J]. Journal of Engineering Geology, 2002, 10(3): 299-304. doi: 10.3969/j.issn.1004-9665.2002.03.014 [8] 陈守余, 周梅春. 人工神经网络模拟实现与应用[M]. 武汉: 中国地质大学出版社, 2000. 1-36.CHEN S Y, ZHOU M C. Realization and application of ANN models[M]. Wuhan: China University of Geosciences Press, 2000. 1-36. [9] 阎平凡, 张长水. 人工神经网络与模拟进化计算[M]. 北京: 清华大学出版社, 2000. 1-32.YAN P F, ZHANG C S. ANN and simulating calculations of evolution[M]. Beijing: Tsinghua University Press, 2000. 1-32. [10] 朱寿赠, 周健红, 陈学军. 桂林市西城区岩溶塌陷形成条件及主要影响因素[J]. 桂林工学院学报, 2000, 20 (2): 100-105. https://www.cnki.com.cn/Article/CJFDTOTAL-GLGX200002001.htmZHU S Z, ZHOU J H, CHEN X J. Analysis of forming conditions and main influential factors of karst collapse in west urban district, Guilin city[J]. Journal of Guilin In-stitute of Technology, 2000, 20(2): 100-105. https://www.cnki.com.cn/Article/CJFDTOTAL-GLGX200002001.htm [11] Chen X J, Chen Z H, Hu C. The dominant factors and spatial-temporal distribution of karst collapse in the west area of Guilin city[A]. Proceedings of international symposium on hydrogeology and the environment[C]. Beijing: China Environmental Science Press, 2000. 502-507. [12] Cooley T. Geological and geotechnical context of cover collapse and subsidence in mid-continent US clay-mantled karst[J]. Environmental Geology, 2002, 42: 469-475. -
下载:
点击查看大图
图(3) / 表(3)
计量
- 文章访问数: 3517
- HTML全文浏览量: 82
- PDF下载量: 6
- 被引次数: 0
