To the analysis of methods and mechanisms of predictive modeling of onboard equipment reliability when solving problems of aircraft maintenance workload planning

The article deals with a method for aircraft maintenance planning based on advanced mathematical modeling techniques. In the course of the research, a mathematical model for forecasting the failure rate of onboard equipment is developed and tested, designed to solve the problems of optimizing decision-making processes for maintenance on the basis of reliability assessment of aviation equipment. The application of Poisson distribution regression in combination with polynomial features allows to reveal the regularities of equipment failures, which depend on operating conditions and maintenance history. For the study, a synthesized dataset was created to simulate different operational scenarios and equipment degradation process. At the first stage, the data were freed from outliers and errors, then normalized to unify the scale of different variables. Next, the data were categorized according to the operating conditions, after which Poisson distribution regression was applied to predict failures. Finally, an efficient maintenance plan that takes into account the predicted failures has been developed using an optimization algorithm. Validation of the model’s predictive capabilities and optimization of the maintenance strategy are performed by comparing with archived data on previously performed work. The analysis of the results revealed the peculiarities of the model operation, namely, the application of least squares regression with single coding demonstrates perfect forecasts, which may indicate the need for model transformation and requires additional verification. At the same time, alternative versions of the methodology revealed more realistic error and correlation limits, which also confirms the reliability of the predictive models. The results of the study show that a combined approach using Poisson distribution regression and polynomial signs can significantly improve the accuracy of forecasts. This method, in particular, has demonstrated its effectiveness in modeling onboard equipment failures, which allows to optimize maintenance processes in order to reduce repair costs. The obtained conclusions confirm the possibility of introducing more accurate proactive methods of maintenance planning, which allows to improve aircraft reliability and reduce the inefficiency of their downtime on the ground.