adaptive neural fuzzy system

Equipment failure, repairs and maintenance play a decisive role in the availability of the entire system. This study presents a practical solution for analyzing equipment repairs and maintenance time and predicting equipment behavior. Expert experience has been used to estimate failure time and equipment repair time; therefore, this study has focused on estimating the repair time and repair rate of equipping the main lubricating oil pump in the gas turbine power generation system with the approach of entering human experience (HE). In the next step, an analysis of the Equipment›s Annual Availability Forecast is performed over a period of 20 years, thus, the critical years of the equipment are determined in terms of downtime by evaluating and reviewing the annual availability. For this purpose, a database of human knowledge and experience has been simulated to estimate the repair times used using fuzzy logic, and the whole process of repair times has been simulated by designing a neural-fuzzy system; which is used to estimate and predict equipment repair time. Then, the annual availability, time-dependent repair rate and other availability indicators are calculated using the Monte Carlo simulation method. The target model is the main lubricating oil pump system of the gas turbine unit of Abadan refinery in Iran. According to the results, applying preventive repairs at optimal intervals of 150 to 160 days, has a significant effect on increasing the availability of equipment and leads to a reduction in additional periodic inspections. Also, the minimum and the maximum system availability is predicted to be 96% and 99%, respectively.