Adaptive fault diagnosis algorithm for multistage cloud computing system

The invention discloses a multilevel cloud computing system adaptive fault diagnosis algorithm, which comprises the following steps of: comparing with monitoring data in a previous monitoring period, and performing data pre-judgment by adopting an information entropy thought; defining the points with relatively large information entropy as suspicious nodes, and further confirming the suspicious nodes by utilizing a neighbor cooperation mode; and for the determined fault point, performing decision tree training to obtain a specific fault type. According to the invention, fault nodes can be effectively judged, corresponding fault type judgment can be made, and the timeliness of fault diagnosis is improved. 本发明公开了一种多级云计算体系自适应故障诊断算法，其步骤包括：与上个监测周期内监测数据进行比较，采取信息熵的思想进行数据预判断；将信息熵较大点定义为可疑节点，并利用邻居协作的方式对可疑节点进一步确认；针对已确定的故障点，进行决策树训练得到具体故障类型。本发明可以有效判断故障节点并作出相应的故障类型判断，并提高故障诊断的及时性。