基于Wi-Fi探测数据的公共场所客流预测方案

Passenger flow forecasting scheme for public places based on Wi-Fi detection data

在大型活动现场或商场、营业厅等公共场所中,通过部署具有Wi-Fi信号探测软件的智能Wi-Fi设备,可以实时地探测到具有Wi-Fi功能的智能设备数量,进而分析出客流相关数据,从而对现场活动计划调整提供有价值的帮助。本文针对Wi-Fi探测数据进行客流分析预测中的需求特点和现有时间序列趋势,以及预测相关算法存在的问题,提出了改进的基于滑动窗口二次指数平滑客流预测算法。同时,通过对客流进行实地统计并进行回归统计分析,建立了Wi-Fi探测客流量和实际客流量在不同时间区间的回归方程。试验结果表明,该方案对客流的预测效果良好,预测误差率比原算法降低了19%~45%,并具有步骤简单、计算量小等优点。

In the scene of the large-scale activities or shopping malls, business halls and other public places, we can monitor the number of smart devices with Wi-Fi function in real-timeby deploying intelligent Wi-Fi equipment with Wi-Fi signal detection software, and then analyze the passenger-related data, provide valuable help to plan adjustments.Based on the characteristics of passenger flow forecasting through Wi-Fi detection data and the existing time series trend forecasting algorithm, this paper proposes an improved integrated forecasting scheme based on sliding window quadratic exponential smoothing algorithm combined with trend and periodicity. At the same time, the regression equation was established by manually counting the passenger flow and analyzing the actual traffic and the number of devices detected by Wi-Fi. The experimental results show that the proposed scheme has a good prediction effect on the passenger flow,and the prediction error rate is 19% ~45% lower than that of the original algorithm, and has the advantages of simple steps and small computation.