基于自适应线性模型的环境数据预测算法

Environmental data prediction algorithm based on adaptive linear model

针对环境大数据在智慧城市应用中的实时性和准确性问题,提出一种基于自适应线性模型的环境数据预测算法。根据气象数据的实时变化情况对模型进行训练,自适应调整训练窗口大小,并在训练态与预测态之间动态实时切换,使模型具有较强的适应环境的能力。该算法具有较低的时延和较小的计算开销,可以在传感器节点上直接部署,满足数据预测的实时性需求。在真实环境数据集的基础上构建仿真试验,相比固定窗口模型,该算法数据预测误差降低17.4%以上,环境数据采集能耗降低80%以上,平均时延降低超过50%;相比已有的机器学习算法,训练及预测时间降低37%以上。

To address the issues of real-time performance and accuracy in the application of environmental big data in smart cities,an environmental data prediction algorithm based on an adaptive linear model was proposed.The model was trained according to the real-time changes in meteorological data,with the training window size being adaptively adjusted.A dynamic and real-time switch between training and prediction states was implemented,enhancing the model's adaptability to environmental changes.The algorithm featured lower latency and reduced computational overhead,allowing for direct deployment on sensor nodes to meet the real-time requirements of data prediction.Simulation experiments constructed on real environmental datasets showed that,compared to fixed-window models,the proposed algorithm reduced data prediction error by more than 17.4%,decreased the energy consumption of environmental data collection by over 80%,and reduced the average latency by more than 50%.When compared to existing machine learning algorithms,the training and prediction time of the proposed algorithm was reduced by more than 37%.