Data Generation for Effective Prediction and Evaluation

Big data analytics is a powerful tool for maintaining stable quality, but due to the fierce global competition and short product life cycle, it is hard to collect sufficient data for practical analysis, especially in the early stage of development. To resolve the above problems, this study proposes a forecast system that combines Grey Generation to increase sample data and fuzzy regression for prediction. To make the best use of data over time, the generated sample data will be calibrated accordingly to improve the prediction accuracy. Then, a fuzzy regression model is adopted to relax the strict limitations of traditional regression. By providing an interval of forecasting trend, the result would give more information for decision making. A case of the capacity decay prediction of Lithium-Ion batteries is employed for illustration and comparison.