Predicting Solar Power and Energy Production Over the Course of a Full Day
Department of Mechanical Engineering-Engineering Mechanics
The majority of air and ground vehicle systems are reliant on specialized diesel fuel. This reliance increases the likelihood that operations may be operating in an energy constrained or contested environment given the state of international relations between global energy providers and consumers. Such a vulnerability has the potential to reduce operational effectiveness or efficiency if logistical supply chains were interrupted or impeded. The most effective and efficient methodology to reduce reliance on specialized diesel fuel is to hybridize our energy and power (E&P) systems, and support more diverse E&P solutions including renewable energy generation (photovoltaic (PV) arrays, wind generation, wave energy converters), nuclear, or decaying isotopes. In this paper/presentation, we present our advances in developing a set of predictive artificial intelligence and machine learning (AI/ML) algorithms that forecast E&P capabilities of a photovoltaic array indirectly and directly. These milestones are a product of two separate types of AI/ML approaches: (1) developing AI/ML based algorithms that predict ambient and panel temperature from various atmosphericbased sensor data which can then be used in combination with an irradiance profile and a MATLAB Simulink model to predict the E&P capabilities of the PV array (indirect method), and (2) developing AI/ML which predicts the resulting E&P capabilities of the PV array, using various atmospheric-based sensor data (direct method).
Proceedings of SPIE - The International Society for Optical Engineering
Predicting Solar Power and Energy Production Over the Course of a Full Day.
Proceedings of SPIE - The International Society for Optical Engineering,
Retrieved from: https://digitalcommons.mtu.edu/michigantech-p2/117