A Low-Complexity CNN-LSTM Framework for Solar Power Generation Forecasting Using Weather Parameters

Abstract

Accurate solar power forecasting is very important for operating photovoltaic (PV) systems and using energy resources efficiently. This paper proposes a lowcomplexity CNN+LSTM model to predict both solar power and weather conditions by using real-time data collected from IoTbased sensors. To monitor performance and anticipate future power output, the system gathers solar irradiance, temperature, humidity, voltage, and current data from the PV system. The convolutional neural network (CNN) extracts spatial features from the input data, while the long short-term memory (LSTM) network captures temporal dependencies to improve forecasting accuracy. The data collected will be processed in a cloud-based monitoring platform that provides real-time visualization of conditions and allows for intelligent management of energy resources. Based on experimental results, this model produced very high prediction accuracy and had much less computational complexity than traditional forecasting methods. Additionally, the proposed system provides the ability to detect faults, optimize energy use, and help to make proactive decisions related to smart solar energy projects. Ultimately, this proposed framework is an efficient and scalable solution for intelligent solar energy management and weather forecasting within today's modern renewable energy system.