Performance prediction and optimization of an organic Rankine cycle (ORC) for waste heat recovery using back propagation neural network

Performance prediction and multi-objective optimization for an organic Rankine cycle (ORC) using back propagation (BP) neural network are investigated in this study. A 3 kW ORC experiment platform is used to obtain 950 sets of basic experimental data, and a BP-ORC model using back propagation neural network is established by training the experimental data. The prediction accuracy of the BP-ORC model is analyzed according to the errors of the training samples and test samples. The effects of six operation parameters on thermal efficiency and net output work are addressed. The Pareto optimal frontier for maximum net output work and maximum thermal efficiency is examined. The results demonstrate that the prediction error of BP-ORC model is very low, and the system performance can be improved by adjusting several operating parameters experimentally according to the model prediction. A tradeoff relation is appeared between net output work and thermal efficiency. The optimal thermal efficiency and net output work for Pareto-optimal solution are obtained.