Self-learning fuzzy predictor of exploitation system operating time

The probability that a system is capable to operate satisfactorily significantly depends on reliability and maintainability of a system. The disadvantage of classic methods of system availability determining is that the probability of realizing by system tasks with expected quality depends on history of operational states and does not take into consideration actual operational conditions that have strong influence on risk-degree of down-time occurring, while the probability of degradation failure in exploitation system is a function of operating time and actual exploitation conditions. The problem of failures prediction can be solved by applying in diagnostics methods the intelligent computational algorithms. The intelligence computational methods enable to create the diagnosis tools that allow to formulate the prognosis of operating time of a system and predict of failure occurring based on the past and actual information about system's operational state. The paper presents the fuzzy logic approach to forecast the prognoses of the operating time of the exploitation system or its equipments according to the specified exploitation conditions that characterize the system exploitation state at the current time. The fuzzy system was based on the Takagi-Sugeno-Kang type fuzzy implications with singletons specifies in conclusions of rules. The fuzzy inference system input variables are the assumed parameters according to which the current exploitation state of the considered system can be evaluated.