Online Temperature and Irradiation Estimation for Large-Scale Solar Farms with Consideration of Computational Tractability

This paper studies state estimation for solar farms undergoing time-varying environmental disturbances. Solar farms composed of photovoltaic (PV) panels are a key distributed energy resource for smart grid systems to integrate renewable energy and reduce dependency on single-source generators. To increase viability and efficiency, the maximum power of each PV must be drawn out, which is nonlinearly dependent on irradiation, PV temperature, and applied voltage. Maximum power point tracking (MPPT) algorithms can control voltage either through model-free or model-based approaches – the former is less successful during rapid environmental changes, and the latter requires state information and computational resources necessary for farms with multiple modules. This work emphasizes mitigating shortcomings of model-based MPPT. Two contributions to the literature are provided: (1) Nonlinear observability metrics are used to determine the ability to estimate PV temperature and irradiation using only current measurements. (2) A novel multi-cluster heterogeneity estimator is developed to support computationally-efficient state observation for large-scale solar farms undergoing environmental changes. Through a case study with simulated rapid changes in irradiation, the performance of map-based MPPT is compared when using the proposed estimator versus a baseline extended Kalman filter (EKF). Results indicate that the multi-cluster heterogeneity estimator requires as little as 8% of the computing time as the baseline estimator with minimal impact on energy output.