基于隐马尔可夫模型的智能化无线传感监控平台研究

Research of Intelligent Wireless Sensor Monitor Platform Based on Hidden Markov Model

随着无线传感器监控网络系统的广泛使用,作为监控系统重要组成部分,后台监控软件单纯地作为可视化工具已经不能满足使用需求。"智能化"成为了后台监控软件所要满足的重要需求,针对这一需求,本文设计并实现了采用"层次化"预警机制的后台监控软件;并结合机器学习模型,设计了基于隐马尔可夫预测模型(HMM)的监控数据预测方法。为了验证本文所设计的预测方法的准确性,结合从柑橘种植园中采集到的真实的监控数据,与其他几种流行的机器学习预测模型进行了一系列的对比。对比试验的结果表明,通过本文所设计的预测方法,能够实现对监控数据在短期过程与长期过程中的可靠预测,从而改善了后台监控软件的智能化水平。

Sensor Network-Based monitor systems are widely applied in multiple areas currently. Background monitor software,which is definitely an important part of monitor system, is no longer a simple tool for visualization.Intelligent becomes a primary de-mand that background monitor softwares need to meet.To satisfy this demand, this paper designs and implements a backgroundmonitor software firstly, in which a hierarchical alarming method involved. Then a prediction solution of monitor data based on Hidden Markov Model(HMM) is introduced.For justifying the performance of the proposed prediction solution, a series of contrast experiments between the proposed solution and other machine learning modelsare conducted. The contrast results show that thepre-dictionperformance of proposed prediction solution is better than the other models', no matter in short term prediction or long termprediction, which means the intelligence of background software is effectively improved.