Many manufacturing industries are facing with the pressure to plan production projects that could satisfied the complicated and fickle marketing economy. Manufacturing industry plays a pivotal role in the Chinese economy and social change. At the main time, production forecasting remains a crucial position in Enterprise Resource Planning (ERP) system. With the complexity of ERP system, improving the accuracy of the production forecasting is the best way to keep an efficient production chain. However, making such a prediction is challenging because the output of production is influenced by many factors, such as fluctuating marketing demand, rapidly progressive of production techniques, investment capital. The goal of this study is overcome these constraints and establish a high-precision forecasting model. Due to environmental and traffic problem, the bicycle becomes a favorite transportation and fitness tool for many people in the world. Simultaneously, owing to v the complicated bicycle industry with the uncertain economic structure of the country, an advanced accurate forecasting method for the bicycle demanded is more important than before. However, it is challenging to forecast the demand of bicycle products with the low-quality information data and rapid innovation of manufacturing industry. Nevertheless, this study proposed an advanced forecasting model combined use of grey system, differential evolution algorithm, Taguchi method and Rolling Mechanism. The first method contributes to forecast the outputs of manufacturing industry and the other three methods are applied to optimize the parameters of a forecasting model based on the minimization of forecasting error. To demonstrate the superiority of this improved method, Chinese bicycle industry as a case study for this thesis. The experimental results show that the Rolling-TDE-GM(1,1) can significantly improve the prediction precision when compared to the traditional models.