Abstract: At present, China's IoT network card users are developing rapidly, but operators lack of security supervision of IoT network card, which makes IoT network card abuse and embezzlement. This paper introduces the Internet of things card monitoring technology based on machine learning, which can use fuzzy c-means algorithm to audit online business, and use naive Bayesian algorithm to classify online content and SMS. Based on the above technology, this paper implements the Internet of things security risk monitoring system, and implements it in China Mobile Liaoning company. Under the premise of ensuring efficiency and accuracy, the system can find a large number of illegal IoT cards and effectively guarantee the security of IoT.

Key words: IoT card, security monitoring, business audit, fuzzy c-means, naive Bayes

摘要： 目前中国的物联网卡用户发展迅速，但运营商却缺乏对物联网卡的安全监管，使得物联网卡出现违规滥用、盗用等情况。文本介绍了基于机器学习的物联网卡监测技术，该技术可以使用模糊C均值算法进行上网业务稽核，使用朴素贝叶斯算法对上网内容以及短信进行分类。基于以上技术，本文实现了物联网安全风险监测系统，并且在中国移动辽宁公司进行实施。该系统在保证效率与准确率的前提下，可发现大量违规物联网卡，有效的对物联网安全进行保障。

关键词: 物联网卡, 安全监测, 业务稽核, 模糊C均值, 朴素贝叶斯