Traffic Demand Forecasting for EGCS with Grey Theory Based Multi-Model Method

Elevator traffic demand forecasting is the essential prerequisite for effectively implementing elevator group control system (EGCS). Considering that there exists lots of abnomal information in elevator traffic caused by subjectivity and occasionality in human behaviour and that observing traffic information continuously is costly and difficult, an improved grey forecasting based method using multi-model to forecast future elevator traffic demand of EGCS is proposed, the abnomal information which refers to outliers is processed, based on which a smoothing technique on original traffic data is conducted to transform the raw data into an increasing sequence, to further reduce the randomness of the observed traffic data and to make full use of regularity information. The proposed method not only avoid the theorrtical error of grey model per se, but also improved the forecasting accuracy, which is suitable for short period forecasting for elevator traffic demand. Simulation experiments show the validity of the proposed method.