Safety evaluation of rock slope construction based on nested dominance relation rough set and fuzzy theory

In order to reduce the risk of engineering construction and solve the problems of complex evaluation methods of rock slope construction safety, this paper studied the evaluation method of rock slope construction safety based on nested dominance rough set and fuzzy theory. First of all, the impact factors of rock slope construction safety were analyzed. Through literature research and expert investigation, the impact factors of rock slope construction safety were classified into four categories: survey and design risk, construction risk, engineering geological risk, and natural environment risk, including 19 secondary impact factor indicators such as survey data accuracy, slope form, support structure, and design parameters. The three-phase membership function was constructed according to the quantitative index; Second, it introduced the specific methods of weight determination and index grading fuzzy evaluation; Third, the operation flow of establishing evaluation decision table, data discretization, multi-level attribute reduction, and generation of preference rules were described; Finally, the reliability of the method was verified by a case study of fifteen representative slopes. It was proved by practice that the risk assessment and analysis method based on nested dominance rough set and fuzzy theory is feasible, which can provide a new idea for risk assessment and law analysis of slope stability. The nested dominance relation rough set theory reduces the attributes of all levels of indicators of slope stability risk from top to bottom, reduces the complexity of the original system, and takes into account the preference information contained in the indicators, which is comprehensive and has high reliability of rule analysis. The number of sample sets and the selection of reduction have a great impact on the evaluation and decision-making based on the nested dominance relationship rough set and fuzzy theory. The larger the sample set, the higher the credibility of the rules generated, and the more reasonable the correct selection of reduction preference rule generation. © 2023, Central South University Press. All rights reserved.