微型电脑应用

针对传统入侵检测方法在捕获复杂计算机危险行为方面存在的局限性，对此提出一种主成分分析—循环神经网络(PCA-RNN)的计算机危险因子入侵检测方法。所提出的方法通过采集Kubernetes架构当中的用户行为和状态，构建用户行为特征，并采用PCA对其进行降维归一化处理，利用RNN对计算机危险因子入侵进行检测。采用Podsecure监控系统对Kubernetes架构内部进程进行权限监测，可管理访问用户的各项权限，提高对异常进程的检测能力。使用可多级保护的K8Safe存储系统，采用独立和可自检的存储服务器、地址服务器和缓存服务器，为容器的运行和存储提供多层安全保障。实验结果表明，所提出的方法对计算机危险因子入侵检测的检测率能达到97.32%,召回率能达到91.04%,误报率为4.52%,内存和利用率均在正常范围内，为计算机危险因子的入侵检测提供一份可行方案。

Aiming at the limitations of traditional intrusion detection methods in capturing complex computer dangerous behaviors, a computer risk factor intrusion detection method based on principal component analysis-recurrent neural network(PCA-RNN) is proposed. The proposed method constructs user behavior characteristics by collecting user behaviors and states in Kubernetes architecture, and uses principal component analysis(PCA) for dimensionality reduction normalization processing, and uses RNN to detect computer risk factor intrusion. Podsecure monitoring system is used to monitor the rights of the internal processes of Kubernetes architecture, which can manage the permissions of users and improve the detection ability of abnormal processes. The K8Safe storage system can complete multi-level protection, independent and self-checking storage servers, address servers and cache servers, provide multi-level security for the operation and storage of containers. The experimental results show that the detection rate of the proposed method can reach 97.32%, the recall rate can reach 91.04%, the false positive rate is 4.52%, and the memory and utilization rate are all within the normal ranges, which provides a feasible scheme for the intrusion detection of computer risk factors.