Online detection anomaly identification method based on data flow density incremental learning

The invention relates to the technical field of online detection, and discloses an online detection anomaly identification method based on data stream density incremental learning, which comprises the following steps: automatically acquiring hyper-parameters according to a data stream; clustering the known data set by using a density incremental learning method; and calculating the local density of new input data and a clustering-based outlier factor value and further updating the data. According to the method, the concept of iLOF and an incremental density-based noise clustering algorithm is combined, the local abnormal factor of each VOCs data point is calculated by using the core k nearest neighbor, the method does not depend on the overall distribution of the data, and the outliers can be effectively detected under the condition of different data distributions. 本发明涉及在线检测技术领域，公开了一种基于数据流密度增量学习的在线检测异常识别方法，包括：根据数据流自动获取超参数；使用密度增量学习方法对已知数据集进行聚类；计算新输入数据的局部密度和基于聚类的离群因子值并进一步更新数据。本发明结合了iLOF和增量式基于密度的带噪声聚类算法的概念，利用核心k