Predicting the Performance of the Catalytic Unit Compressor at Abadan Refinery Using Artificial Intelligence

Recent advancements in digitalization and the development of digital twin technologies, driven by the expansion of artificial intelligence (AI), have positioned the refinery industry as a pioneer in this field. The emergence of the Fourth Industrial Revolution has facilitated the creation of predictive performance models and fault detection systems, with AI becoming a key component in enhancing refinery efficiency. In this study, the operational performance of the catalytic unit compressor at the Abadan refinery is analyzed. The investigation involves an assessment of the compressor's daily operational data logs, which contains the operational variables’ data on hourly based recording. The results indicate that nonlinear algorithms exhibit superior performance and higher accuracy in predicting compressor efficiency, primarily due to the heterogeneous distribution of the data. On average, these algorithms achieve an error rate of approximately ۸%. Since the predictive models are sensitive to the quantity of available data, the algorithms were initially trained using monthly data. As the process continued, the volume of data was incrementally increased, ultimately incorporating data from the past five years for comprehensive analysis.

کلیدواژه ها:

Digital Transformation ، Artificial Intelligence ، Centrifugal Compressor ، Numerical Solution ، Machine Learning