基于灰狼优化与LSTM预测结合蒙特卡洛不确定性分析的太阳能-地热能集成多联产系统热力学分析与性能提升：以特内里费岛为例

Abstract Global warming and fossil fuel supply limitations highlight the need for sustainable energy options. Renewable-based systems provide a path to carbon neutrality but face reliability challenges due to intermittency. This study investigates Tenerife Island's potential for integrating solar and geothermal energy. A novel hybrid system is proposed, combining concentrated solar power, geothermal energy resources, with a system comprised of the following components: a supercritical CO₂ cycle, a lithium bromide-water absorption cooling system, a multi-effect desalination unit, a three-stage organic Rankine cycle and a proton exchange membrane electrolyzer. This system produces electricity, heating, cooling, freshwater, and hydrogen, achieving baseline energy and exergy efficiencies of 62 % and 17, respectively. The system's production rates are 7844 kW power, 4416 kW cooling, 6848 kW heating, 22.6 kg/h hydrogen, and 20.7 m 3 /h freshwater. Optimization using the grey wolf algorithm enhances the energy efficiency by 21 %, the exergy efficiency by 38 %, and the hydrogen production rate by 18 %. Solar energy forecasting employs direct normal irradiance data (2005–2024) with seq2seq long short-term memory predictions up to 2030. A forward uncertainty analysis using Monte Carlo simulations reveals that cooling capacity, exergy destruction rate, and net power production are most sensitive to fluctuations in direct normal irradiance, with coefficients of variation (CV) ranging from 4.4 % to 4.5 %, while energy and exergy efficiencies exhibit minimal coefficient of variation (CV < 0.1 %).