103 Disturbance Discrimination of Elevator Door by a Support Vector Machine : Pattern recognition by machine learning from model data

Abstract

A Support Vector Machine (SVM) is applied to discriminate disturbances during operation of an elevator door such as draft and a caught passenger. The SVM with a soft margin is designed with priority on detection of passengers for safety. Moreover, Fast Fourier Transform (FFT) is also utilized in the SVM for robust discrimination of a disturbance with an unknown time of occurrence. As a result, the SVM achieves nearly 100 % discrimination rate of human factors while achieving about 80 % discrimination rate of draft.