基于空气质量数据特征，在B-样条基底拟合曲线的基础上，将曲线本身信息、曲线变化信息引入分析，构造加权曲线深度指标，探索一种异常曲线探测方法。与现有仅考虑离散点信息和曲线本身信息的方法相比较，该探测方法更加符合空气质量数据特点，具备缺失值处理能力及整体异常和局部异常的识别能力。将该方法应用于兰州市空气质量数据采集点的二氧化氮水平曲线异常情况分析，结果表明该方法具有更好的异常情况识别效果。%According to the characteristics of air quality data ,this paper fits the curves using B -spline basis ,constructs a weighted curve depth index using the information of curve and its derivative ,and gives an approach of outlier curves detection .Compare with the approach used currently which consider only the information of discrete point or only the information of curve itself ,our approach has the ability to handle missing values ,identify local and overall outlier ,which is more in line with the characteristics of air quality data .As an application ,NO2 emission levels measured by an environmental monitor station in Lanzhou were analyzed in this paper ,which shows our approach has better performance than the approach only consider the information of curve itself .Finally ,the further applications of outlier detection were discussed .
展开▼
