Tomás Sánchez, José Enrique de, Tomás Marín, Sergio de, Peiro, Victoriano
FESAEI: a fuzzy rule-based expert system for the assessment of environmental impacts
Environmental Monitoring and Assessment. 2018, 190:528. doi:10.1007/s10661-018-6907-9
URI: http://hdl.handle.net/10045/79234
DOI: 10.1007/s10661-018-6907-9
ISSN: 0167-6369 (Print)
Abstract: 
Currently, the method mostly used by practitioners of environmental impact assessment (EIA) is the “crisp numbers” method. Nevertheless, this arithmetic method is far away of giving correct values due to its rigidity and the lack of consideration of important aspects as the imprecision and incompleteness of data and the uncertainty that usually pervade our knowledge of environment. A more flexible model that considers uncertainty of knowledge and imprecision of data is necessary. Among the different approaches for the assessment of environmental impacts, the fuzzy logic-based one takes account of the aspects said before; this was our primal assumption. On this paper, we explain the structure and performance of the fuzzy rule-based inference model we built, how it works, and what can be obtained when used to assess environmental impacts. Our fuzzy expert system for the assessment of environmental impacts (FESAEI) is built as the combination of five subsystems, using a total of 120 fuzzy rules, and being the output and input for the next subsystem. We assessed the parameters of rarity, robustness, quality, recoverability, intrinsic value, extension, intensity, persistence, impact_character, cumulativeness, transmissivity, and impact prevalue in four subsystems. The fifth subsystem gives the definitive impact value corresponding to the impact type of “compatible,” “moderate,” “severe,” and “critical.” The model is verified and statistically validated. Weighted Cohen’s kappa shows an almost perfect concordance among experts and FESAEI’s evaluations.
Keywords:Fuzzy logic, Environmental impact assessment, Fuzzy inference, Crisp numbers
Springer International Publishing
info:eu-repo/semantics/article