Sensorless real-time solar irradiance prediction in grid-connected PV systems using PSO-MPPT and IoT-enabled monitoring

Abstract

Accurate prediction of solar irradiance is vital for optimizing the energy output and operational efficiency of grid-connected photovoltaic (PV) systems, especially under fluctuating environmental conditions. Conventional tools such as pyranometers, though widely used, often fail to capture the actual irradiance experienced by PV modules and involve high costs and maintenance. This paper presents a simulation-based methodology for real-time solar irradiance (G) prediction, eliminating the need for external sensors by using only PV electrical parameters. The approach leverages the maximum power point current (\(\:{\text{I}}_{\text{mpp}}\)) and voltage (\(\:{\text{V}}_{\text{mpp}}\)) measured directly from a PV module to predict irradiance, utilizing a Particle Swarm Optimization (PSO)-based Maximum Power Point Tracking (MPPT) algorithm to ensure accurate tracking of power output across varying irradiance levels. The proposed system is developed in the MATLAB/Simulink environment and incorporates a complete Internet of Things (IoT)-based monitoring framework using the ThingSpeak cloud platform and Telegram app. This setup allows continuous data acquisition, real-time visualization, historical logging, and instant performance alerts. Simulations were conducted on a single 250 W monocrystalline SunPower SPR-X20-250-BLK PV module, with irradiance levels ranging from 200 to 1000 W/m² in 200 W/m² increments, while maintaining a fixed temperature of 25 °C in the first case, reflecting the standard test conditions (STC) temperature operation conditions. In the second case, three temperature values (15 °C, 45 °C, and 65 °C) were applied to account for the effect of the temperature variation on the accuracy of prediction. As well as to represent realistic PV operating conditions of 15 °C for low cell temperature, 45 °C as the nominal operating cell temperature (NOCT), and 65 °C for high cell temperature, enabling performance evaluation across a practical temperature range. Each irradiance level was applied for 7.5 s to evaluate the PSO’s tracking capability under dynamic conditions. Experimental results of the first case confirm the effectiveness of the proposed model, with predicted irradiance values of 189.67, 396.42, 597.17, 764.98, and 994.65 W/m² corresponding closely to the actual inputs. The model demonstrated high predictive accuracy, achieving a Root Mean Square Error (RMSE) of 16.63 W/m², a Mean Absolute Error (MAE) of 11.42 W/m², and an excellent coefficient of determination (R²) of 0.9965. In the second case, the predicted irradiance values at 1000 W/m² input were 1000.27 W/m² at (15 °C), 994.65 W/m² at (25 °C), 981.16 W/m² at (45 °C), and 957.40 W/m² at (65 °C). Results show slight overestimation at 15 °C and underestimation at higher temperatures. Incorporating temperature coefficient affects the prediction accuracy across all cases, confirming the model’s reliability under varying temperature conditions. Simulation results across different temperature levels (15 °C, 25 °C, 45 °C, and 65 °C) demonstrate that \(\:{\text{I}}_{\text{mpp}}\) varies proportionally with irradiance, while \(\:{\text{V}}_{\text{mpp}}\) remains relatively stable with irradiance but decreases noticeably with increasing temperature levels. This behavior confirms the suitability of these electrical parameters, \(\:{\text{I}}_{\text{mpp}}\) and \(\:{\text{V}}_{\text{mpp}}\), for reliable and accurate irradiance prediction. Integration of cloud-based IoT platforms further enhances system scalability and remote operability. This sensorless, low-complexity method offers a cost-effective and accurate solution for real-time solar irradiance monitoring, contributing to the digitization and intelligent management of PV systems.