DYNAMIC RELIABILITY MODELING OF HYDRAULIC SYSTEMS BASED ON PREDICTIVE DIAGNOSTICS ALGORITHMS
Abstract
An urgent scientific and practical problem of transitioning from time-based preventive maintenance of hydraulic drives to condition-based predictivemaintenance within the "digital twin" framework has been resolved. The authors have substantiated the limitations of classical static reliability analysismethods that operate on Boolean logic and the exponential failure distribution law, completely ignoring the chronological sequence of degradationprocesses and the accumulation of micro-damage in precision pairs and working fluid. A fundamentally new hybrid methodology for evaluating thereliability of hydraulic systems is proposed, which integrates structural-logical schemes of dynamic fault trees, stochastic modeling of wear physicsbased on the Weibull distribution, and a real-time Bayesian parameter update mechanism using telemetry subsystem data. The scientific novelty of theapproach lies in the mathematical formalization of time-dependent degradation processes and the cascade failure effect by adapting specialized logicaloperators and functional dependency triggers. This enabled the consideration of conditional failure probability, where the output event is activated onlyunder a specific sequence of physical phenomena over time. The efficiency of the developed four-level architecture of the predictive diagnosticsalgorithm was validated through testing on a virtual test rig of a 6-ton vertical load holding actuator built using Bosch Rexroth precision components.It has been experimentally proven that when the monitoring system detects an anomalous pressure drift, the dynamic model with Bayesianrecalculation of shape and scale parameters avoids the "optimistic error" of classical analysis. At 2,000 hours of operating time, the model captured a2.4-fold increase in individual failure risk (up to 18 % compared to 7.5 % in the static model). The application of the algorithm provided adaptiveadjustment of the remaining useful life forecast from the nominal 4,200 hours to the actual 1,100 hours, enabling timely unscheduled maintenance andpreventing critical damage to hydraulic units, the rectification cost of which is 5–7 times higher than preventive component replacement.