Short-term Elevator Traffic Flow Estimation with Hybrid Long Short-Term Memory Network

In modern life, elevator plays an important role in our daily life. It transports citizens and goods in the high buildings and shopping malls. However, sometimes there is some tiny malfunction with the elevator, which means the maintenance worker will fix it in the next hours. During the repair, citizens cannot use the elevator. Thus, estimating the elevator traffic flow in the next hours to choose a suitable time for repairing is important, which means it will make less trouble to the citizens. Up to now, there are some researches about elevator traffic flow monitoring. However, few of them make an estimation of the future flow. This paper presents an elevator traffic flow estimation method by designing a hybrid long short-term memory neural network. Firstly, the traffic flow is denoted by the running frequency of the elevator per hour. Then, a hybrid long short-term memory neural network is designed which comprehensively considered the trend, period and fusion of the data. Finally, experiments on several elevators show the effectiveness of the proposed method.