基于卡尔曼滤波预测的电动汽车分时租赁监测数据去人为干预技术研究

发布时间：2018-11-15 13:19

【摘要】：随着社会环保意识的增强,大力推广采用清洁能源的电动汽车,电动汽车分时租赁业务随之不断拓展,多级管理平台纷纷建立。总管理平台所采集的电动汽车参数日益积累,为大数据分析研究奠定了坚实的基础。预处理技术在数据采集过程中起着至关重要的作用。特别地,从各运营商子平台所采集的监测数据存在人为干预风险。为此,文章提出采用人为干预概率曲线量化监测数据与协方差比间关系,将人为干预概率曲线区间和观测量关系作为输入,建立了回归对象决策树并引入传统卡尔曼滤波算法,从而提出基于决策树分析的卡尔曼滤波预测方法,从而减少人为篡改的干预影响,以达到对运营总平台所采集的数据去人为干预的目的。将该算法应用于车速预测领域,得到可信度更高的车速预测数据,为有效地实现大数据分析奠定坚实的技术支撑。
[Abstract]:With the strengthening of social awareness of environmental protection, electric vehicles with clean energy are popularized vigorously, and the time-sharing leasing business of electric vehicles expands continuously, and multi-level management platforms are established one after another. The general management platform collects the electric vehicle parameter to accumulate day by day, has laid the solid foundation for big data analysis research. Preprocessing technology plays an important role in the process of data acquisition. In particular, there is a risk of human intervention in monitoring data collected from operator subplatforms. Therefore, the relationship between the quantitative monitoring data and the covariance ratio of the artificial intervention probability curve is proposed. The interval of the artificial intervention probability curve and the observational relation are taken as the input, and the decision tree of the regression object is established and the traditional Kalman filtering algorithm is introduced. A prediction method of Kalman filter based on decision tree analysis is proposed to reduce the influence of artificial tampering and achieve the purpose of eliminating human interference to the data collected by the general platform of operation. The algorithm is applied to the field of speed prediction to obtain more reliable speed prediction data, which lays a solid technical support for the effective implementation of big data analysis.
【作者单位】：国网重庆市电力公司电力科学研究院;
【基金】：国家科技支撑计划项目(2015BAG10B00)
【分类号】：U491

【相似文献】