基于PGM的态势预测技术

Situation Prediction Technology based on PGM

概率图模型采取概率论相关方法,利用图来表达变量相关关系,近年来在机器学习和人工智能等领域应用广泛。提出基于概率图模型(PGM)的两种系统状态预测方法。首先介绍了通过隐马尔可夫模型(HMM)可以计算出一个系统每一时刻处于各种状态的概率以及这些状态之间的转移概率。然后详细描述了可以通过每个时刻的状态转移矩阵,计算出任意时刻系统状态取值概率。最后在已知系统历史状态序列,利用概率图模型算法,在状态转移矩阵随机时间变化和不随时间变化的两种情形下,预测未来系统的状态。

In recent years,probability graph models is widely used in machine learning,artificial intelligence and other fields.Two methods of system state prediction based on PGM(probability graph model)are proposed.This paper first introduces that the HMM(hidden Markov model)model is used to calculate the probability of each system being in various states at every moment and the transition probability between these states.Then,it describes in detail that through the state transition matrix at each time,the probability of the value of the system state at any time is calculated.Finally,the paper treats that when the historical state sequence of the system is known,the PGM algorithm is used to predict the state of the future system in two situations of random time change and no change with time.