Development of integrative data intelligence models for thermo-economic performances prediction of hybrid organic rankine plants

Abstract

Computer aid models such as machine learning (ML) are massively observed to be successfully applied in different engineering-related domains. The current research was designed to predict the thermo-economic performances of hybrid organic Rankine plants. The XGBoost optimization algorithm was used to select the influencing parameters for the plant's highest first/second law (energy and exergy) efficiency (ηI, ηII) and lowest levelized energy cost (LEC). Random Forest (RF), Kernel Ridge Regression (KRR), and Artificial Neural Network (ANN) were developed for prediction in the second stage of modeling. The three indicators were predicted based on five combinations: design variables, temperature variables, power variables, heat exchanger variables, and all variables together. In the training and testing phases, XGBoost-RF consistently outperformed other models in predicting ηI, ηII, and LEC across various input combinations. The 5th combination of all variables always yielded the highest predictive accuracy, emphasizing the importance of comprehensive input data. The study demonstrates that computational intelligence tools are appropriate for measuring and evaluating system efficiency and offer stakeholders and decision-makers an approachable modeling method.