Initialization of the HMM-based delay model in networked control systems

Ge, Yuan; Xiaoxin, Zhang; Chen, Qigong; Jiang, Ming

doi:10.1016/j.ins.2016.05.013

Cited by 8 publications

(6 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Firstly, we need to initialize the DCHMM-based delay model λ 0 (λ 0 � (N, M, π 0 , A 0 , B 0 )). By using the optimal stabilization methods in [33], we can obtain that the optimal initial values of N and M are 4 and 6, respectively, which means that there are four different network states…”

Section: Illustrative Examplementioning

confidence: 99%

“…e uniform initialization method is no longer suitable for obtaining the optimal initial value of the observation matrix (B 0 ). In this section, the segmental K-means clustering algorithm proposed in [33] is used to optimally initialize the observation matrix (B 0 ) as follows based on the first 200 CA delay values in Figure 5. With N � 4 and M � 6, we can rewrite the entry of the observation matrix B as follows according to equation ( 3):…”

Section: Illustrative Examplementioning

confidence: 99%

“…Based on the first 200 CA delay values in Figure 5, the optimal initial values (B 0 ) of the observation matrix can be obtained as shown in Figure 7 by using the segmental K-means clustering algorithm proposed in [33].…”

Section: Illustrative Examplementioning

confidence: 99%

“…To illustrate the superiority of the DCHMM-based delay model over the DTHMM-based delay model in the prediction accuracy, we carry out again the experiment under the DTHMM-based delay model in [33]. Without considering the dependence between adjacent CA delays, the observation matrix B 0 degenerates into a two-dimensional matrix, in which the rows represent the network states and the columns represent the intervals of CA delays.…”

Section: Illustrative Examplementioning

confidence: 99%

“…According to the delay characteristics, there are mainly three kinds of HMMs to model the random delays of NCSs: discrete-time HMM (DTHMM) [28,30], continuous-time HMM (CTHMM) [29,31], and semicontinuous HMM (SCHMM) [32]. Moreover, how to optimally initialize the parameters of HMM-based delay models has been discussed in [33].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Modelling and Prediction of Random Delays in NCSs Using Double-Chain HMMs

Zhang

Gao

et al. 2020

Discrete Dynamics in Nature and Society

Self Cite

View full text Add to dashboard Cite

This paper is concerned with the modelling and prediction of random delays in networked control systems. The stochastic distribution of the random delay in the current sampling period is assumed to be affected by the network state in the current sampling period as well as the random delay in the previous sampling period. Based on this assumption, the double-chain hidden Markov model (DCHMM) is proposed in this paper to model the delays. There are two Markov chains in this model. One is the hidden Markov chain which consists of the network states and the other is the observable Markov chain which consists of the delays. Moreover, the delays are also affected by the hidden network states, which constructs the DCHMM-based delay model. The initialization and optimization problems of the model parameters are solved by using the segmental K-mean clustering algorithm and the expectation maximization algorithm, respectively. Based on the model, the prediction of the controller-to-actuator (CA) delay in the current sampling period is obtained. The prediction can be used to design a controller to compensate the CA delay in the future research. Some comparative experiments are carried out to demonstrate the effectiveness and superiority of the proposed method.

show abstract

Section: Illustrative Examplementioning

confidence: 99%

Section: Illustrative Examplementioning

confidence: 99%

Section: Illustrative Examplementioning

confidence: 99%

Section: Illustrative Examplementioning

confidence: 99%