SVM-based Elevator Traffic Flow Prediction

During the peak traffic period, the performance of elevator group control system heavily depends on the prediction of elevator traffic flow (ETF). This paper proposed a novel approach to predict future ETF demand of each floor by using support vector machine (SVM), which is a novel type of learning machine based on statistical learning theory. The basic principle of SVM was firstly reviewed, and then the approach to elevator traffic prediction based on SVM was outlined. The comparison is made between SVM method and other three traditional methods, i.e. the artificial neural network prediction, the historical exponential smoothing prediction and the simple mean prediction. The simulation results demonstrate that support vector regression method outperforms the others in predicting ETF