Sky Imaging and Neural Network Solar Tracker for Cloudy Conditions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing solar tracking systems fail to optimize the angle of photovoltaic arrays for maximum energy harvest during partly cloudy or transitional conditions, leading to reduced power generation and increased mechanical wear due to unnecessary tracker movement.
Innovation Solution
A solar tracking system that uses a sky imager and a neural network to determine the optimal angular position of photovoltaic arrays based on current and predicted sky conditions, generating a multi-planar irradiance signal to adjust the tracker position for maximum irradiance capture without excessive movement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the tracker follows the sun's position during cloudy conditions, then the tracking system maintains operational continuity, but the solar energy capture is reduced because the optimal angle diverges from solar position
Solution Approach 1:
The patent replaces traditional mechanical sun-tracking mechanisms with a sky imaging system combined with machine learning algorithms. Instead of mechanically following calculated solar positions, the system uses cameras to capture sky images and neural networks to predict optimal angles, substituting mechanical computation with optical sensing and intelligent processing.
Solution Approach 2:
The system dynamically changes the tracking angle parameter based on sky condition analysis. Rather than maintaining a fixed relationship between solar position and tracker angle, the neural network adjusts the optimal angle parameter according to detected cloud patterns, sky brightness distribution, and atmospheric conditions, allowing the system to adapt to varying sky conditions.
2Productivity
If the tracker moves frequently to optimize position during transitional conditions, then the solar energy capture is maximized, but the mechanical wear increases leading to equipment failure
Solution Approach 1:
The sky imaging system performs preliminary analysis of cloud patterns and sky conditions to predict future optimal positions. By anticipating changes in sky conditions before they fully develop, the system can make proactive positioning adjustments rather than reactive ones, reducing the frequency of tracker movements while maintaining energy capture optimization.
Solution Approach 2:
The system implements feedback control by continuously monitoring sky images and comparing predicted optimal angles with current tracker positions. The neural network processes this feedback to determine when movement is truly necessary, adjusting the control strategy to minimize unnecessary movements that would cause mechanical wear while maintaining optimal energy capture.
3Measurement precision
If multi-angle irradiance sensors are used to determine optimal tracker position, then the measurement precision is improved, but the device complexity and cost increase
Solution Approach 1:
The patent creates an optical copy of the sky scene using cameras and image processing to represent the irradiance distribution. Instead of using physical multi-angle irradiance sensors, the system captures images of the sky and uses machine learning to infer the angular irradiance profile from these visual copies, significantly reducing hardware complexity while maintaining measurement capability.
Solution Approach 2:
The system replaces complex physical irradiance measurement mechanisms with optical imaging and computational algorithms. Rather than using multiple physical sensors at different angles, the sky camera system substitutes mechanical measurement with optical capture and digital processing, simplifying the device while achieving comparable or superior measurement precision through intelligent analysis.
Data Source
AI summary
A system and method is disclosed for solar tracking and controlling an angular position of a photovoltaic power system. The solar tracking system includes an imaging device for capturing images of the sky; a solar position data generating module; and a control system comprising a neural network. The neural network has multiple convolutional layers to generate a first output associated with the images, and a solar position data module. A first dense layer module receives the solar position data and generates a second output. A second dense layer module receives the first output and the second output and generates a concatenated data sequence. A processor is programmed to generate a multi-planar irradiance signal (MPIS) in response to the concatenated data sequence, and determine an angular position of the PV power system and adjust the angular position in response to an angle of maximum irradiance.