Journal "Computational Technologies"

Article information

2025 , Volume 30, № 1, p.15-23

Zakrevskaya N.S., Kovalevskii A.P.

Detecting the lack of functioning of elevators using the change point detection method

We study the emergence of a linear Poisson regression model from the problem of statistical analysis of a time-inhomogeneous Poisson process and its application to elevator performance analysis. Methodology. We use a linear Poisson regression model, which assumes that the responses are independent and each is Poisson distributed with a parameter equal to the linear combination of the regressors. Findings. We studied the regression model that arises when observing a time-inhomogeneous Poisson process. We have shown that if the regressor matrix has full rank, then the maximum likelihood estimates have a simple explicit form. In this case, they do not involve a specific type of regression function and coincide with a known parameter estimate based on a sample from Poisson distribution. The elevator reliability index is calculated as the ratio of the number of successful activations of the elevator main drive multiplied by 100 to the estimate of the number of all activations requested by users. We carried out parameter estimation and stochastic simulation for the elevator operation. Based on the results of the analysis, it is recommended to calculate the elevator reliability index using data on weekdays with the exception of pre-weekends and holidays. Originality/value. The paper derives a linear Poisson regression model for the number of calls in fixed time intervals using assumptions of a linear regression model for the intensity of the Poisson flow. The case of periodic intensity is studied and conditions are found under which the estimates of the regression parameters have a simple explicit form. The developed evaluation algorithm is applied to determining periods of elevator lack of functioning and calculating the elevator reliability index. The calculation results were verified by direct stochastic modelling.

[link to elibrary.ru]

Keywords: Poisson process, change point, regression, elevator

doi: 10.25743/ICT.2025.30.1.003

Author(s):
Zakrevskaya Natalia Stanislavovna
Position: Senior Fellow
Office: Novosibirsk State Technical University
Address: 630073, Russia, Novosibirsk, 20, prospekt K. Marksa
Phone Office: (383) 3463226
E-mail: natali.erlagol@gmail.com

Kovalevskii Artem Pavlovich
Dr. , Associate Professor
Position: Leading research officer
Office: Novosibirsk State Technical University
Address: 630073, Russia, Novosibirsk, 20, prospekt K. Marksa
Phone Office: (383) 3463226
E-mail: artyom.kovalevskii@gmail.com
SPIN-code: 5702-8998

References:
1. Galton F. Natural inheritance. London: Macmillan; 1889: 266. Available at: http://galton.org/books/natural-inheritance/index.html (accessedon March 12, 2024).

2. Cook R.D., Weisberg S. Residuals and influence in regression. N.Y.: Chapman and Hall; 1982: 240. Available at: https://hdl.handle.net/11299/37076 (accessed on March 12, 2024).

3. Shorack G., Wellner J. Empirical processes with applications to statistics. N.Y.: Wiley; 1986: 956. DOI:10.1137/1.9780898719017.

4. Draper N.R., Smith H. Applied regression analysis. N.Y.: Wiley Series in Probability and Statistics; 1998: 716. DOI:10.1002/9781118625590.

5. Hardin J.W., Hilbe J.M. Generalized linear models and extensions. 3rd edition. UK: Taylor & Francis, Didcot, Oxfordshire; 2012: 479. Available at: https://econpapers.repec.org/bookchap/tsjspbook/glmext.htm (accessed on March 12, 2024).

6. Brown R.L., Durbin J., Evans J.M. Techniques for testing the constancy of regression relationships over time. Journal of the Royal Statistical Society. 1975; (37):19–192. DOI:10.1111/j.2517-6161.1975.tb01532.x.

7. MacNeill I.B. Limit processes for sequences of partial sums of regression residuals. Annals of Probability. 1978; (6):695–698.DOI:10.1214/aop/1176995491.

8. Zeileis A., Leisch F., Hornik K., Kleiber Ch. Strucchange: an R Package for testing for structural change in linear regression models. Journal of Statistical Software. 2002; 7(2):1–38. DOI:10.18637/jss.v007.i02.

9. BischoffW.On designs for recursive least squares residuals to detect alternatives. MODA—Advances in Model-Oriented Design and Analysis. 2016; (11):37–45. DOI:10.1007/978-3-319 31266-8_5.

10. Sakhanenko A.I., Kovalevskii A.P., Shelepova A.D. Remarks on invariance principle for one-parametric recursive residuals. Siberian Electronic Mathematical Reports. 2021; 18(2):1058–1074. DOI:10.33048/semi.2021.18.081.

11. Aue A., Horvath L., Huskova M., Kokoszka P. Testing for changes in polynomial regression. Bernoulli. 2008; (14):637–660. DOI:10.3150/08-BEJ122.

12. Aue A., Horvath L. Structural breaks in time series. Journal of Time Series Analysis. 2013; 34(1):1–16. DOI:10.1111/j.1467-9892.2012.00819.x.

13. Stute W. Nonparametric model checks for regression. Annals of Statistics. 1997; (25):613–641.
DOI:10.1214/aos/1031833666.

14. Bischoff W. A functional central limit theorem for regression models. Annals of Statistics. 1998; (26):1398–1410. DOI:10.1214/aos/1024691248.

15. Kovalevskii A.P., Shatalin E.V. Asymptotics of sums of residuals of one-parameter linear regression on order statistics. Theory of Probability and its Applications. 2015; 59(3):375–387. DOI:10.1137/S0040585X97T987193.

16. Kovalevskii A., Shatalin E. A limit process for a sequence of partial sums of residuals of a simple regression on order statistics. Probability and Mathematical Statistics. 2016; (36:1):113–120. Available at: https://www.math.uni.wroc.pl/~pms/files/36.1/Article/36.1.8.pdf.

17. Kovalevskii A.P. Asymptotics of an empirical bridge of regression on induced order statistics. Siberian Electronic Mathematical Reports. 2020; (17):954–963. DOI:10.33048/semi.2020.17.070.

18. Chebunin M.G., Kovalevskii A.P. Asymptotics of sums of regression residuals under multiple ordering of regressors. Siberian Electronic Mathematical Reports. 2021; (18:2):1482–1492. DOI:10.33048/semi.2021.18.111.

19. Aue A., Rice G., Sonmez O. Detecting and dating structural breaks in functional data withoutdimension reduction. Journal of the Royal Statistical Society:
Series B (Statistical Methodology). 2018; (80:3):509–529. DOI:10.1111/rssb.12257.

20. Algamal Z. Diagnostic in Poisson regression models. Electronic Journal of Applied Statistical Analysis. 2012; (5). DOI:10.1285/i20705948v5n2p178.

21. Khan A., Ullah M.A., Amin M., Muse A.H., Aldallal R., Mohamed M.S. Empirical examination of the Poisson regression residuals for the evaluation of influential points. Mathematical Problems in Engineering. 2022; Article ID 6995911. DOI:10.1155/2022/6995911.

Bibliography link:
Zakrevskaya N.S., Kovalevskii A.P. Detecting the lack of functioning of elevators using the change point detection method // Computational technologies. 2025. V. 30. № 1. P. 15-23