一种基于BP神经网络的状态评估模型

A State Evaluation Model Based on BP Neural Network

在体系结构复杂、计算规模较大的集群中,内部计算节点容易出现故障,主动冗余技术是保障系统可靠运行的常用方式。提出在主动冗余技术中构建一种基于BP神经网络的状态评估模型,该模型根据计算节点的历史和当前运行状态对系统进行评估,给出其运行健康状况。实验表明,该模型能够兼顾不同运行特征之间的非线性关系,提高状态评估的准确率,减少误报率。

In the cluster with complex structure and large scale, the internal nodes are prone to failure, and the active redundancy technology is a common way to ensure the reliable operation of the system. Puts forward an evaluation model based on BP neural network in active redun- dancy technology, evaluates the history and current condition of the system based on the model of computing nodes, and gives its operational health. Experimental results show that the model can take into account the nonlinear relationship between different run-time indica- tors, improve the accuracy of state assessment, and reduce the false alarm rate.