Abstract
A method is considered for detecting and predicting the abnormality in operation of power unit equipment by an example of a gas-turbine unit (GTU). A problem of detecting abnormality in operation is formulated as the mathematical problem of modeling an abnormality criterion taking the values from 0 to 1. It has been assumed that the predictive analytics methods can be effective for predicting the future state of process equipment based on the existing scope of measurements without any increase. It is assumed that, even when each individual measurement is within the range taken as the range of normal functioning, their cumulative dynamics enables us to judge a developing defect, i.e., about the transition of the diagnosed process equipment (DPE) to the zone of abnormal operation. To solve this problem, an approach is proposed based on calculating the value of the “abnormality indicator,” which can be interpreted as a conditional potential created by points in a multidimensional space of indicators that characterize the state of equipment at the given time. By learning the model against the indicators that set the regions of states (the state of normal operation and the state for various kinds of fixed defects), one can then apply the trained model to determine the type of state: the closer the value of the abnormality indicator to the values inherent in a particular region of functioning, the greater the probability that the state of DPE corresponds to this region. It is shown that, due to certain objective circumstances, there is no practical possibility of training the model against the data obtained during abnormal operation with specific types of defects in DPE. This reduces the problem to adapting the method to the case when we have only the region of normal operation for learning the model. The proposed model was trained and tested during normal operation of the equipment. The test results indicate that the proposed method is consistent (i.e., it does not yield false positive response).
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
REFERENCES
E. Siegel, Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die (Wiley, Hoboken, N. J., 2016). https://doi.org/10.1002/9781119172536
M. A. Waller and S. E. Fawcett, “Data science, predictive analytics, and big data: A revolution that will transform supply chain design and management,” J. Bus. Logist. 34, 77–84 (2013). https://doi.org/10.1111/jbl.12010
S. S. Khan and M. G. Madden, “One-class classification: Taxonomy of study and review of techniques,” Knowl. Eng. Rev. 29, 345–374 (2014). https://doi.org/10.1017/S026988891300043X
A. G. Trofimov, K. E. Kuznetsova, and A. A. Korshikova, “Abnormal operation detection in heat power plant using ensemble of binary classifiers,” in Neuroinformatics 2018: Advances in Neural Computation, Machine Learning, and Cognitive Research II, Ed. by B. Kryzhanovsky, W. Dunin-Barkowski, V. Redko, and Y. Tiumentsev (Springer-Verlag, Cham, 2019), in Ser.: Studies in Computational Intelligence, Vol. 799, pp. 227–233. https://doi.org/10.1007/978-3-030-01328-8_27
A. A. Korshikova and A. G. Trofimov, “Model for early detection of emergency conditions in power plant equipment based on machine learning methods,” Therm. Eng. 66, 189–195 (2019). https://doi.org/10.1134/S0040601519030042
A. A. Korshikova and A. G. Trofimov, “Predictive model for calculating abnormal functioning power equipment,” in Cyber-Physical Systems: Industry 4.0 Challenges, Ed. by A. Kravets, A. Bolshakov, and M. Shcherbakov (Springer-Verlag, Cham, 2020), in Ser.: Studies in Systems, Decision and Control, Vol. 260, pp. 249–259. https://doi.org/10.1007/978-3-030-32648-7_20
H. Lingjun, R. A. Levine, J. Fan, J. Beemer, and J. Stronach, “Random forest as a predictive analytics alternative to regression in institutional research,” Pract. Assess., Res. Eval. 23, 1 (2018). https://doi.org/10.7275/1wpr-m024
T. G. Dietterich, “Ensemble learning,” in The Handbook of Brain Theory and Neural Networks, 2nd ed., Ed. by M. A. Arbib (Massachusetts Inst. of Technology Press, Cambridge, Mass., 2002), Vol. 3, pp. 110–125.
M. Lipatov, “New reality dictates new rules,” Prana, Nov., May 25 (2020). https://prana-system.com/novosti/ novosti/novaya-realnost-diktuet-novye-pravila/. Accessed November 23, 2020.
Author information
Authors and Affiliations
Corresponding author
Additional information
Translated by T. Krasnoshchekova
Rights and permissions
About this article
Cite this article
Korshikova, A.A., Idzon, O.M. Model of Emergency Conditions’ Early Detection in Power Plant Equipment Based on the Least Potentials Method. Therm. Eng. 68, 763–769 (2021). https://doi.org/10.1134/S0040601521090032
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1134/S0040601521090032