摘要：AbstractSelf-learning from historical data and using the resulting information to improve the future self-decision-making focused on technical objects, results machine learning approach, is a major goal in today’s devices exploitation process, including also self-maintenance. Simulation-based risk-oriented approach into the broad machine learning machinery usually fits independent theorical distributions to historical degradation data of the device in order to forecast risk indicators and evaluate in the final stage the maintenance process standards. The study case of the paper are critical overhead cranes operating with hazardous and continuous operation conditions. The subject of the paper is a technical devices degradation self-analysis for self-maintenance strategy discussed on crane case study, also including device representative data sensitivity analysis. The proposed subject directly impacts the device risk control management and maintenance scheduling processes, and the analysis proves the robustness of the intelligent self-analysis system by introducing additional information for future decision making. The devices degradation representative data analysis impact on the preferred exploitation focused decisions have been discussed. The device features have been presented via stochastic non-linear optimization model with bounded constraint that assess a risk global-system indicator based on Monte Carlo simulations.
关键词：KeywordsReal-time algorithmsschedulingprogrammingIntelligent maintenance systemsCyber-Physical SystemsRobustness analysisSimulation of stochastic systemsModeling for control optimization
