An Intelligent Autonomous System for Condition-Based Maintenance- Case Study: Control Valves

Maintenance process plays a vital role in oil and gas refineries. Process owner expectations and new technology acquisition have been changing the mindset of domain experts to the new maintenance approaches from reactive to condition-based maintenance models, so maintenance is considered as a part of the business process which provides added value for the company. With the advent of Neuro-fuzzy and intelligent systems and potential capability of them, we have presented a new model by using neuro-fuzzy model and MAPE control loop using real-world data to improve the reliability and also decreasing the cost of maintenance for a control valve in a gas refinery. In this research, we employed ANFIS model for the reasoning process which has six inputs (Inlet/outlet Pressures, temperature, flow rate, controller output, and valve rod displacement) and one output that is the type of failure of the control valve. We considered the most failures based on domain expert knowledge. We also made use of the MAPE as a control loop to unceasingly monitor, analyze, plan and finally execute the process of prediction of failures. Due to undertaken improvement, there was a considerable change in reliability and financial indices. Moreover, we compared the proposed approach with two different methods. The result of the comparison shows that our proposed model is comprehensively more acceptable as compared to the other approaches by accuracy.