ROBUST MATHEMATICAL METHODS FOR FAILURE PREDICTION OF GAS TURBINE ENGINES

摘要

In time of technical systems operating the necessity of current technical state identification and specific operation decision making by using this information appears. Therefore the practiced diagnosis methods have to ensure sufficient depth failure detection. In other words these methods must forecast system failure in whole and detect a defect accurate within subsystems, functional unit (FU). The reason of technical system failure can have complex character in the defect or fault in whole initiation. The faults can be induced by individual functional unit defects or multiple defects. Last circumstance reduces to uncertainty in tolerance assignment for ever symptoms of different fault scenario, which are induced by multiple defects. The search problem of symptoms brackets, those conform to health technical state, refer to incorrect problem class. The problem of functional unit design parameters values estimation in operating time on basis of measuring data is also incorrect problem . In the paper the conditional correct statement problems and solution methods for diagnosis problems of complex technical systems with multiple faults in the input data uncertainly condition on basis of tolerance control concept are considered. These problems are subsystem with faults searching; searching the symptoms brackets, those conform to systems health technical state in whole; and also functional unit design parameters values estimation in operating time on basis of symptoms measuring data with known accuracy. The first problem is problem of system state classification. The statistic method for gas turbine engine technical state classification problem solution is designed. Two last problems can be solved by integration them to stochastic modification (optimization) problems [2]. The aggregate of denoted problems solution methods forms the methodology of robust systems technical state forecast. The examples of described methods application are offered for bypass turbojet engine technical state forecast.