“This is the 12th day of my participation in the Gwen Challenge in November. Check out the details: The Last Gwen Challenge in 2021.”
Introduction to the
By the previous two chapters, we learned the unmanned vehicles for threat assessment is a very complicated things, traditional threat assessment is not suitable for now to rapidly changing battlefield, only keep up with the trend of The Times to, through the use of computer technology, building dynamic bayesian network for real-time observation, real-time monitoring to achieve threat assessment requirements.
This article begins with an introduction to dynamic Bayesian network construction:
Dynamic Bayesian network construction
The key of threat assessment using Bayesian network is to construct network structure, that is, to establish assessment system according to target characteristics. This chapter first establishes the SBN structure and then extends it to DBN considering the time characteristics.
Static Bayesian network model construction
SBN is an Directed Acyclic Graph (DAG), a representation and inference model of uncertain knowledge based on probability theory and Graph theory. An SBN consists of two parts: network structure G and network parameter θ. Theta, namely B = (G). G is a directed acyclic graph in which directed edges represent conditional dependencies between variables. θ is the conditional probability distribution associated with each variable and is represented by a conditional probability table (CPT). A typical BN is shown in the figure below:
Let the reliability of node X be Bel(X), and we say X gets causal information from parent nodes U and V, and diagnostic information from child nodes Y and Z
Considering that in this SBN, node X isolates its parent nodes U and V from its child nodes Y and Z, and U and V are independent of each other, then
Type,
That is, A is a normalized factor, determined by the parent node and child node of X.
By comprehensive analysis of each characteristic attribute, the SBN based threat assessment model structure diagram G is obtained:
Dynamic Bayesian network model extension
SBN does not consider the influence of continuous time factors, so it is difficult to accurately evaluate the threat degree of dynamic combat targets. DBN expands SBN in time dimension and can evaluate the state of a specific moment according to the observation value of multiple moments. Therefore, the evaluation result is more reasonable. DBN is based on BN. Compared with BN, DBN needs to set the state transition probability between two time slices, that is, the prior probability of target threat degree is dynamically updated in each time slice. P (Wi + 1 | Wi) for bayesian time in the threat degree of adjacent nodes of transition probability matrix, P (w ‘I) for the current time in the threat degree of node a posteriori probability, the next time the prior probability in the threat degree of node can be expressed as P (Wi + 1) = P (‘ I) P (Wi + 1 | Wi). The threat assessment model based on DBN is shown in the figure:
remarks
This paper mainly introduces how to realize the threat assessment of unmanned vehicle target through dynamic Bayesian network. Combined with the content of the former three chapters, it systematically explains the content of “Unmanned vehicle Target Threat Assessment Based on Dynamic Bayesian Network”, so that people can have a preliminary understanding of dynamic Bayesian network.
Present past content
- Dynamic Bayes (1) – Nuggets (juejin. Cn)
- Dynamic Bayes (2) – Nuggets (juejin. Cn)