Flexibility Discrete Dynamic Bayesian Networks modeling and Inference algorithm

Directly applying Discrete Dynamic Bayesian Networks to time-varying environment is highly complex, it mainly dues to: application environment with mutant characteristics; network structure and parameters needing to have the variation; adapt to the uncertainty of sensor observations. To meet the above requirements, proposing the concept of Flexible Discrete Dynamic Bayesian Network, designing a mechanism of flexible model based on muti-model for the discrete-time systems under mutant environment. Based on the above, applying the Forward algorithm to fulfill Flexible Discrete Dynamic Bayesian Network probabilistic inference thus can be able to use uncertainty observations information to obtain a reliable state estimation.