In recent years, bicycle races, along with the crest of the high technology continues to increase. Because of this increased races, performance of bicycles, in both biological and mechanical terms, is extraordinarily important and efficient. In terms of the ratio of cargo weight a bicycle can carry to total weight, it is also a most efficient means of cargo transportation. In spite of advanced technology, there are still some problems on bicycles during working conditions and road roughness such as on the mountain from tire and mechanical parts. In this investigation, a extraordinary designed with fiber-carbon body and light bicycle is tested on mountain road conditionswith prescribed trajectory on the mountain for different elevation, speed, hearth rate, bike cadence and average temperature. The real time measured parameters are predicted with proposed two types of neural networks for approaching real time neural network predictors. The results of the proposed neural network have shown that neural predictor has superior performance to adopt the real time bicycle performance.