Dynamic Model for Assessing the State of the Occupational Health and Safety Management System: Application of the Bayesian Approach


Annotation:

The article is a continuation of the analytical approach to assessment of condition of the occupational health and management system. The risk-oriented approach supposes decision-making and taking measures to ensure labor protection according to the risk value. Such approach is based on the constant analysis of hazards and current risks and regular adjustment of the occupational health and management system.
The model uses the Bayesian approach which, on the basis of a priori value of a composite indicator of assessment of occupational health and management system, determines the potential changes of this indicator according to such factors as the hazard level of units of occupational health and management system and complexity of fires or accidents. The value of composite indicator of assessment of the condition of occupational health and management system is found via the method of maximization of a posteriori density of probability. The level of hazard of units of occupational health and management is assessed for groups of units homogenous by their economic activities and functional hazard classes, by the values of injury (fatality) risk for employees. The complexity of fires (accidents) is assessed in accordance with the quantity of equipment units engaged in fire extinguishing or accident elimination. 
In order to assess the efficiency of occupational health and management system, the Harrington function’s intervals are applied. The boundary values of the function complying with a good, satisfactory and poor efficiency of occupational health and management system are determined.
As an example, calculation of a composite indicator of assessment of condition of the occupational health and management system for a fire department is considered. It is demonstrated that considered complexity of fires and the hazard level of fire units may significantly change the assessment of efficiency of the occupational health and management system.
The general scheme of adjustment of the assessment of efficiency of occupational health and management system is provided.

References:
1. Poroshin A.A., Bobrinev E.V., Udavtsova E.Yu., Kondashov A.A. Dynamic Model for Assessing the State of the Occupational Health and Safety Management System. Bezopasnost Truda v Promyshlennosti = Occupational Safety in Industry. 2021. № 6. pp. 28–33. (In Russ.). DOI: 10.24000/0409-2961-2021-6-28-33
2. Downey A.B. Think Bayes: Bayesian Statistics Made Simple. Needham: Green Tea Press, 2012. 210 p.
3. Zhukovskiy E.V., Kalinin M.O., Marshev I.I. Detection of malicious executable files based on machine learning algorithms. Problemy informatsionnoy bezopasnosti. Kompyuternye sistemy = Information Security Problems. Computer Systems. 2019. № 1. pp. 89–99. (In Russ.).
4. Zavyalov Ya.O. Investment strategies based on anomalies. Put nauki = The Way of Science. 2015. № 10 (20). pp. 56–60. (In Russ.).
5. Lykova O.V. Automatic Detection of the Main Markers of Insincerity in Texts. Vestnik Moskovskogo gosudarstvennogo lingvisticheskogo universiteta. Gumanitarnye nauki = Vestnik of Moscow State Linguistic University = Humanitarian Science. 2019. Iss. 7 (823). pp. 146–154. (In Russ.).
6. Kharya S., Agrawal S., Soni S. Naive Bayes classifiers: A probabilistic detection model for breast cancer. International Journal of Computer Applications. 2014. Vol. 92. Iss. 10. pp 26–31. DOI: 10.5120/16045-5206
7. Mandal S.K. Performance analysis of data mining algorithms for breast cancer cell detection using Naive Bayes, logistic regression and decision Tree. International Journal of Engineering and Computer Science. 2017. Vol. 6. Iss. 2. pp. 20388–20391. DOI: 10.18535/ijecs/v6i2.40
8. Jing Y., Pavlovic V., Rehg J.M. Boosted Bayesian network classifiers. Machine Learning. 2008. Vol. 73. Iss. 2. pp. 155–184. DOI: 10.1007/s10994-008-5065-7
9. Bassett R., Deride J. Maximum a posteriori estimators as a limit of Bayes estimators. Mathematical Programming. 2019. Vol. 174. Iss. 1–2. pp. 129–144. DOI: 10.1007/s10107-018-1241-0
10. Lyubushin N.P., Brikach G.E. Harrington's generalized desirability function in the multiparameter economic problems. Ekonomicheskiy analiz: teoriya i praktika = Economic analysis: theory and practice. 2014. № 18 (370). pp. 2–10. (In Russ.).
11. Poroshin A.A., Udavtsova E.Yu., Bobrinev E.V., Kondashov A.A., Kharin V.V. Assessment of Fire Hazard Level of Industrial Objects based on the Statistic Methods. Bezopasnost Truda v Promyshlennosti = Occupational Safety in Industry. 2020. № 3. pp. 12–17. (In Russ.). DOI: 10.24000/0409-2961-2020-3-12-17
12. Poroshin A.A., Shishkov M.V., Mashtakov V.A., Putin V.S., Bobrinev E.V. Dependence of the Traumatism of Firemen on Complexity of the Fire. Pozharnaya bezopasnost = Fire safety. 2013. № 2. pp. 92–94. (In Russ.).
DOI: 10.24000/0409-2961-2021-9-26-30
Year: 2021
Issue num: September
Keywords : occupational safetу management system Harrington function Bayesian method apriori probability likelihood function maximization of a posteriori density dynamic model
Authors:
  • Poroshin A.A.
    Poroshin A.A.
    Dr. Sci. (Eng.), Chief Research Associate FGBU VNIIPO EMERCOM of Russia, Balashikha, Russia
  • Bobrinev E.V.
    Bobrinev E.V.
    Cand. Sci. (Biol.), Lead Researcher FGBU VNIIPO EMERCOM of Russia, Balashikha, Russia
  • Udavtsova E.Yu.
    Udavtsova E.Yu.
    Cand. Sci. (Eng.), Lead Research Assistant FGBU VNIIPO EMERCOM of Russia, Balashikha, Russia
  • Kondashov A.A.
    Kondashov A.A.
    Cand. Sci. (Phys.-Math.), Lead Researcher, otdel_1_3@mail.ru FGBU VNIIPO EMERCOM of Russia, Balashikha, Russia