Application of the Improved Fuzzy K-Modes Algorithm in Evaluating the Ground Collapse Risk in Covered Karst Region of Wuhan City

摘要

The influence factors of karst collapse are very complex, the kinds of which are various and inter-restrictions. The influence factors, such as lithology, karst development degree, groundwater fluctuations and human engineering activities, usually are expressed by categorical data. Although the mathematical theory of quantitative forecasting is quite mature, it has many limitations in processing large scale qualitative data. Therefore, it is necessary to introduce and apply the improved fuzzy k-modes algorithm processing qualitative data to evaluate the ground collapse risk in covered karst region. In this method the frequency of each attribute value of each attribute is taken as the clustering center. The distance of objects can be modified by the weight of each attribute. A non-random method is applied to choose the clustering centers, which can ensure the clustering results of complex qualitative data accuracy and reliability. In this study we choose 13 mainly influence factors as evaluating indexes at the basis of the geological environment and inducing factors of the ground collapse in covered karst region of Wuhan. The improved fuzzy k-modes algorithm is applied to forecast the ground collapse risk in covered karst region of Wuhan city. The predicting results tally with the facts. That indicates that the improved fuzzy k-modes algorithm is effective in predicting the ground collapse risk in covered karst region of Wuhan city. It also provides us the scientific decision-making basis of disaster prevention and reduction of Wuhan city.