LAUSR

Abstract Conductive gussasphalt mixture can melt snow on the bridge deck, but it may corrode steel bridge deck and have an impact on traffic environment and safety when the power is on. To solve this problem, five conductive gussasphalt mixtures were prepared, and the effects of mixture type, working conditions and environmental factors of conductive gussasphalt mixture on corrosion of steel bridge deck was studied systemically. Based on the extreme learning machine optimized by genetic algorithm, the corrosion degree prediction model of steel bridge deck was established. The results indicated that number of times on power, mixture type and temperature had significant effects on the corrosion of steel deck, and their contribution rates were 58.47%, 24.62% and 15.40%, respectively. After optimization by genetic algorithm, the error of extreme learning machine model was 0.40–9.25%. Compared with the traditional extreme learning machine model, they decreased by 5.31–10.63%.