Genetic Algorithms Based Logic-driven Fuzzy Neural Networks for Emergency Capability Assessment of Hydropower Engineering

Abstract

Emergency capability assessment of hydropower engineering is researched by using two fuzzy neural network (‘FNN’) models which are constructed with the aid of AND OR fuzzy neurons, namely: (i) the genetic algorithm-based fuzzy neural network (‘GAFNN’); and (ii) the hybrid genetic algorithm-based fuzzy neural network (‘HGA-FNN’). The GA-FNN model employs a basic genetic algorithm (‘GA’) to optimize its structure and skeleton and HGA-FNN model is designed as the extension of GA-FNN which is involved in a conditional local search method. The performances of the two proposed models are tested and further validated using a big experimental data set of expert estimation by questionnaire investigation. The results indicate that HGA-FNN has a better predictive performance than GA-FNN and that both of them have good potential in evaluating emergency response ability of hydropower engineering.