Deep reinforcement learning‑based smart control of solar‑driven power cycle with thermal energy storage: A Los Angeles case study

• DDPG controller stabilized turbine pressure within nearly 4 % and superheat within nearly 0.2 K tolerance range. • Annual mean efficiency improved by approximately 6 percentage points across variable seasonal and daily solar conditions. • Disciplined TES cycles eliminated overcharge and depletion, sustaining reliable storage performance year-round. • GA Pareto front clustered around 2.55 MPa and 10.1–10.7 K superheat for optimal operation. Solar–driven organic Rankine cycles (ORCs) require precise coordination of working fluid superheat, turbine inlet pressure, and net efficiency. Conventional single–loop control cannot effectively manage these tightly coupled objectives. This study introduces a deep deterministic policy gradient (DDPG) supervisory controller, trained using an 8760–hour global horizontal irradiance (GHI) dataset in a MATLAB–CoolProp environment. The reward function penalizes deviations in superheat, pressure, and efficiency. Over a full–year simulation, compared to a fixed–flow baseline, the DRL controller significantly dampened seasonal and transient variability: turbine inlet pressure stayed within around 4 % of 2.5 MPa, superheat within nearly 0.2 K of a + 10 K target, and efficiency between 20 and 30 percentage, improving the annual mean by 6 percentage points. Thermal energy storage exhibited stable, daily state–of–charge cycles, avoiding overcharge and depletion. A genetic algorithm (GA), applied exclusively to DRL–controlled data, mapped the pressure, temperature and efficiency trade space. The Pareto front highlighted non–dominated optima near 2.55 MPa and 10.1–10.7 K, while dominated clusters corresponded to off–design pressure regimes. The integrated DRL-GA framework thus enables consistently optimal ORC operation under variable solar input, enhancing efficiency, stability, and component lifespan, and offering a deployable pathway for advanced renewable energy systems.