微信平台中网络设备在线故障预测方法仿真

摘要

对微信平台中的网络设备在线故障进行准确预测,可以对故障进行及时检修,使网络设备正常运行.进行故障检测时,应获取网络设备在线故障预测先验条件概率和约束条件,组建微信平台的网络设备在线故障预测模型,但是传统方法通过对故障样本对概率神经网络进行迭代训练完成检测,但是不能准确计算在线故障预测先验条件概率,无法获取想对应的约束条件,降低了故障检测的准确度.提出一种马尔科夫链的网络设备在线故障预测方法.上述方法先采集微信平台网络设备在线故障状态训练样本集,构造设备在线故障状态线性回归函数,将设备在线故障状态训练问题转变为在线故障状态求解问题,组建设备在线故障状态线性方程组,然后结合马尔科夫链理论计算出网络设备在线故障预测先验条件概率,给出在线故障预测绝对概率的约束条件,依据在线故障预测初始概率向量和预测下一步转移概率矩阵,组建微信平台的网络设备在线故障预测模型.仿真结果表明,所提方法可以提前检测出微信平台系统的异常状态,且预测精确度较高.%An online fault prediction method of network devices is proposed based on the Markov chain.Firstly,the training sample set for the online fault state of network devices in the WeChat platform is collected and the linear regression function of online devices fault is built.Then,the training problem of online fault state is converted into the solving problem and the system of linear equations of devices online fault state is built.Moreover,the priori conditional probability of network devices online fault prediction is worked out and the constraint condition of online fault prediction absolute probability is given out.Finally,the next step of transition probability matrix is predicted according to the initial probability vector of online fault prediction and the prediction model for online fault of network devices in the WeChat platform is built.The simulation results show that the method mentioned above can predict the abnormal state of WeChat platform system ahead of time.It has good prediction precision.