State Estimation for Probabilistic Boolean Networks Via Outputs Observation
State Estimation for Probabilistic Boolean Networks Via Outputs Observation
复制标题
通过输出观察进行概率布尔网络的状态估计
DOI：
10.1109/tnnls.2021.3059795
复制
发表时间：
2022
影响因子：
10.4
通讯作者：
Jianquan Lu
中科院分区：
文献类型：
--
作者：
Jie Zhong;Zongxi Yu;Yuanyuan Li;Jianquan Lu
本文研究了通过观测输出序列来估计概率布尔网络的状态。可检测性描述了观测器唯一估计系统状态的能力。通过定义观测输出序列的概率，提出了可检测性度量的新概念。可检测性度量定义为当输出序列的长度为无穷大时，所有可检测输出序列的概率之和的极限，它可被视为状态估计的定量评估。通过定义相应的非确定随机有限自动机，结合状态估计信息和输出序列的概率，设计了一种随机状态估值器。提出的可检测性度量概念进一步对可检测性进行了定量分析。此外，通过定义马尔可夫链，将可检测性度量的计算转化为马尔可夫链中特定状态概率和的计算。最后，给出了数值算例来说明所得的理论结果。
This article studies the state estimation for probabilistic Boolean networks via observing output sequences. Detectability describes the ability of an observer to uniquely estimate system states. By defining the probability of an observed output sequence, a new concept called detectability measure is proposed. The detectability measure is defined as the limit of the sum of probabilities of all detectable output sequences when the length of output sequences goes to infinity, and it can be regarded as a quantitative assessment of state estimation. A stochastic state estimator is designed by defining a corresponding nondeterministic stochastic finite automaton, which combines the information of state estimation and probability of output sequences. The proposed concept of detectability measure further performs the quantitative analysis on detectability. Furthermore, by defining a Markov chain, the calculation of detectability measure is converted to the calculation of the sum of probabilities of certain specific states in Markov chain. Finally, numerical examples are given to illustrate the obtained theoretical results.