基于多特征融合的云平台异常点检测方法研究

Research on Cloud Platform Anomaly Detection Based on Multi-Feature Fusion

针对传统云平台异常点检测方法存在准确率低和运算速度慢的问题,提出基于多特征融合的云平台异常点检测方法,构建云平台不同子系统的特征空间优化模型,实现不同子系统之间的特征自动融合。首先设计基于多特征融合的云平台异常点检测算法,对模型约束进行监测分析;然后通过试验验证方法不仅能够进行云平台异常点检测,而且比其他传统方法有更高的准确率和更快的计算速度。

In order to solve the problem of low accuracy and low computing speed in anomaly detection of cloud platform,a multi-feature fusion method is proposed for anomaly detection of cloud platform.The feature space optimization model of different subsystems of cloud platform is built to realize the automatic feature fusion among different subsystems.An outlier detection algorithm is designed based on the above multi-feature fusion of cloud platform,and the model constraints are monitored and analyzed.Then experiments are designed to verify that the algorithm based on multi-feature fusion is indeed able to detect the abnormal points of cloud platform.It has higher accuracy and faster running speed than other traditional algorithms.