一种基于加权隐马尔可夫的自回归状态预测模型

Research on Condition Trend Prediction Based on Weighed Hidden Markov and Autoregressive Model

针对电子系统状态趋势预测问题,提出了一种加权隐马尔可夫模型的自回归趋势预测方法.该方法以自回归模型作为隐马尔可夫的状态输出,利用加权预测思想对马尔可夫链中的隐状态进行混合高斯模型的加权序列预测,并利用最大概率隐状态下的自回归系数计算模型输出.通过对实际的复杂混沌序列和电子系统BIT状态数据进行趋势预测,并针对不同模型参数下的预测结果进行实验分析,结果表明该方法对系统状态变化的趋势具有较好的预测性能.

A novel trend prediction approach based on weighed hidden Markov model（HMM）and autoregressive model（AR）is presented in order to solve this problem of trend prediction for complex electronic system.This approach regards the autoregressive model as the output of HMM,uses weighted prediction method and mixed Gaussianin model to predict the hidden state of Markov chain,and calculates the output of model by using the regression coefficient of the maximum probability hidden state.This approach is applied to the trend prediction of complex chaotic time series and typical electronic equipment＇s BIT data,and the effects of various model parameters on trend prediction precision are discussed.The experiments based on condition trend prediction for electronic equipments demonstrate the effectiveness of the method.