基于混合机器学习模型的物联网网关状态预测  

IoT gateway state prediction based on hybrid machine learning model

作　　者：刘涛 LIU Tao(Xi’an Traffic Engineering Institute,Xi’an 710000,China)

机构地区：[1]西安交通工程学院中兴通信学院,西安710000

出　　处：《信息技术》2025年第2期86-91,共6页Information Technology

摘　　要：由于传统方法物联网状态预测准确率低、预测时间长,因此,文中提出基于混合机器学习模型的物联网网关状态预测方法。利用有监督机器学习方法采集过滤数据样本,通过信息增益算法对分类数据进行特征评估,经相关性判断公式删除冗余数据,提高数据参数精准度。利用多元线性回归分析构建混合机器学习预测模型,模拟训练得到数据运行规律,有效提高状态预测的精准度。实验分析得出:混合机器学习预测误差值在0.1MB左右,误差率最小可达2%,迭代50次的单位预测速度为0.3min,准确率和预测速率明显优于传统方法,具有良好的可行性与有效性。The traditional method of IoT state prediction has low accuracy and long prediction time,so an IoT gateway state prediction method based on hybrid machine learning model is proposed.Supervised machine learning methods are used to collect and filter data samples,feature evaluation is performed on classified data through information gain algorithms,redundant data is deleted by correlation judgment formula,and the accuracy of data parameters is improved.Multiple linear regression analysis is used to build a hybrid machine learning prediction model,and the operation law of the simulated training data is effectively improved.Experimental analysis shows that the prediction error value of mixed machine learning is about 0.1MB,the error rate can be as small as 2%,and the unit prediction speed of 50 iterations is 0.3min,and the accuracy and prediction rate are significantly better than traditional methods,which has good feasibility and effectiveness.

关 键 词：混合机器学习 物联网网关 状态预测 多元线性回归 信息增益 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]