A Heuristic Fuzzy Grey Model And Application in Taiwan’s Water-Power Forecasting
Two barriers occurred in grey theory models over the decades. Firstly, the dynamic weights of background value will produce the different forecasting errors. Secondly, the different data models needed the different forecasting methods to adjust their predicting situations. This approach presents an integrated testbed to reduce the forecast error of grey model (GM). This forecasting testbed adopts the concept of a new heuristic algorithm to adjust the ability of GM model (HFGM (1,1)). This study also combines a GA-fuzzy controller with the normalization algorithm of control chart theory to modify the data structure to increase GM forecasting efficiency. This study also verified the proposed model by the water-power demand of Taiwan’s government statistics. The experimental results revealed that the average error of (GM(1,1)/HFGM(1,1) models are the MAPE is 11.24% and 9.87%) respectively, which effectively reduce the error level of GM(1,1) model.
Key words- Forecasting, Heuristic Fuzzy Grey theory, Genetic algorithm