Artificial Intelligence in ISES Measureserver® for Remote Experiment Control
The paper deals with the area of Internet School Experimental System (ISES) remote experiments in general and its core module called ISES Measureserver®. In particular ISES Measureserver® is, in fact, a finite state machine, serving for the measured data accumulation, processing and providing communication in the server-client system. Recently, we replenished ISES Measureserver® by a new functionality, namely diagnostics of the connected to the RE physical hardware, using the artificial intelligence solutions.
In the introduction, the state of the art of ISES remote experiments is described. In the next chapter a consideration for the applying of proper artificial intelligence method to improve the Measureserver® reliability is made. We focused on the cognitive Fault Diagnosis System (FDS) intended for distributed sensor networks. FDS makes advantage of spatial and temporal relationship among sensors connected to RE physical hardware to give the information for reduction of the influence of failures, ill effecting the Measureserver® functioning. The lower layer uses Change Detection Test (CDT) based on Hidden Markov models (HMM) configured to detect variations in the relationships among couples of sensors. Changes in the HMM are detected by inspecting the corresponding likelihood. The output information provided by the CDT lower layer is then passed to the cognitive higher layer collecting information to discriminate among faults, changes in the environment and false positive.
The intended improvement is the increase of the reliability, monitoring of the state and the fast remedy of the functioning of remote experiments in case of malfunction.Proposed diagnostics solution will contribute to improvement to remote experiments reliability and to a wider acceptance of this new ICT technology.
KeywordsISES Measureserver® Fault Diagnosis System Change Detection Test Hidden Markov model remote experiment sensor network
Unable to display preview. Download preview PDF.
- 1.Zeman, P.: Software environment for control of remote experiments. Ostrava: VŠB-Technical University of Ostrava (2011)Google Scholar
- 2.Alippi, C., Ntalampiras, Roveri Manuel, S.: A Cognitive Fault Diagnosis System for Distributed Sensor Networks. IEEE Transactions on Neural Networks and Learning Systems 24(8), 1213–1226 (2013),http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=06502725 (read on April 8, 2014), ISSN 2162-237X. doi: 10.1109 / TNNLS.2013.2253491
- 3.Kadous, M.: Hidden Markov models. Australia: School of The University of New South Wales, Computer Science and Engineering (2002), http://www.cse.unsw.edu.au/~waleed/phd/html/node34.html (read on April 8, 2014)
- 4.Kadous, M.: Advantages of HMMs. Australia: School of The Uni-versity of New South Wales, Computer Science and Engineering (2002), http://www.cse.unsw.edu.au/~waleed/phd/html/node35.html (read on April 8, 2014)
- 5.Venkatasubramanian, Rengaswamy, Yin, Kavuri: A review of process fault detection and diagnosis: Part i: Quantitative model-based methods, Computers Chemical Engineering (2003), http://www.sciencedirect.com/science/article/pii/S0098135402001606 (read on April 8, 2014)
- 6.Ljung, Caines: Asymptotic normality of prediction error estimators for approximate system models. In: 1978 IEEE Conference on Decision and Control including the 17th Symposium on Adaptive Processes, vol. 17, pp. 927–932 (January 1978)Google Scholar
- 7.Rabiner, Juang: An introduction to Hidden Markov models. IEEE ASSP Magazine, 4–15 (January 1986)Google Scholar
- 8.Durbin, Eddy, Krogh, Mitchison: Biological Sequence Analysis: Probabilistic Models of Proteins and Nucleic Acids. Cambridge University Press (1998), http://f3.tiera.ru/2/Cs_Computer%20science/CsBi_Bioinformatics/Durbin%20R.,%20Eddy%20S.R.,%20Krogh%20A.,%20Mitchison%20G.%20Biological%20sequence%20analysis%20%28CUP,%201998%29%28ISBN%200521629713%29%28O%29%28371s%29_CsBi_.pdf (read on April 8, 2014)