Parameter Importance Ranking of a Heat Pump–Organic Rankine Cycle Pumped Thermal Energy Storage System: A Weighted Neighborhood Rough Set Feature Selection

Industrial waste heat and renewable thermal sources below 350 °C can be stored and dispatched through heat pump–organic Rankine cycle (HP-ORC) pumped thermal energy storage (PTES), yet its performance is governed by a series of coupled thermodynamic and operational variables whose relative importance to competing objectives remains poorly quantified. This study, for the first time, applies a weighted neighborhood rough set (WNRS) algorithm to rank the significance of fifteen continuous decision variables of a latent-storage HP-ORC PTES system with respect to electrical-power-to-power ratio, exergy efficiency, levelized cost of storage (LCOS), and life-cycle carbon intensity (CI). The WNRS-reduced feature subsets, with neighborhood radius calibrated by KNN and SVM cross-validation, are then optimized through a two-stage particle swarm optimization (PSO)–technique for order of preference by similarity to ideal solution (TOPSIS) framework. Results show that the four objectives are governed by distinct parameter subsets: electrical-power-to-power ratio depends on a compact set of cycle temperatures, LCOS on heat-source-related operating variables, and exergy efficiency on a broader set including heat-transfer matching parameters, while CI on the mass flow rates. The HP evaporating temperature emerges as the dominant coupling variable. For the composite TOPSIS objective, the eight-variable reduct selected for the composite objective recovers about 69% of the 11-feature improvement while markedly reducing dimensionality. An independent benchmark confirms that the WNRS reduction is physically interpretable rather than purely statistical.