Abstract: The evaluation standard of flood disaster loss is difficult to determine due to its randomness, fuzziness and uncertainty. In this paper the internal distribution characteristics of historic flood sample set in Sichuan Province as a case study were extracted through fuzzy clustering iteration calculation using the weighted fuzzy kernel clustering algorithm (WFKCA). The boundary values between different flood loss levels were calculated as the reference standard in flood loss assessment. The evaluation indexes were selected specifically to make the assessment universal, and two kinds of standards (absolute value and relative value) of data set were discussed, respectively. In view that index weights are crucial to the assessment, a flexible coupling weighting method (FCWM) is proposed. Firstly, the fuzzy analytic hierarchy process (FAHP) algorithm was adopted to calculate the subjective judgment matrix for deducing the expert guidance range of each index weight; then the guidance range was applied to constrain the objective weight search range of WFKCA. The FCWM could get the optimized cluster effect as much as possible on the premise of meeting the subjective decision-making preferences. Experimental simulations show that the assessment standard calculated by FCWM is relatively reliable and valuable for the decision-making reference.

Key words: flood disaster assessment standard, weighted fuzzy kernel clustering algorithm (WFKCA), fuzzy clustering iterative calculation, fuzzy analytic hierarchy process (FAHP), flexible coupling weighting method (FCWM)