Shading Loss Estimation for PV Systems Using Supervised Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current models for estimating the performance of solar photovoltaic (PV) systems are inaccurate due to factors like soiling, shading, and snow cover, particularly in real-world conditions, leading to overestimation or underestimation of energy production, and lack granular loss estimates to account for short-term fluctuations.
Innovation Solution
A computer processor-based method that estimates energy losses due to shading by integrating measured and modeled data, including solar elevation and azimuth angles, beam irradiance, and system age, using supervised learning techniques to quantify underperformance and calculate shading losses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If derate factors are used to estimate energy losses, then the model is simple to implement, but the accuracy of energy production estimates deteriorates due to inability to capture short-term fluctuations
Solution Approach 1:
The patent segments energy losses into distinct categories (shading, soiling, snow cover, hardware failure) with separate estimation methods for each. Instead of using a single derate factor, the model divides total losses into component parts that can be individually analyzed and summed, allowing granular tracking of different loss mechanisms throughout the system lifetime.
Solution Approach 2:
The patent changes the parameter representation from static derate factors to dynamic, time-varying loss estimates. Each loss category is modeled as a function of relevant parameters (e.g., shading losses as a function of sun position and obstacle geometry, soiling as a function of weather conditions and system age), enabling the model to adapt to changing conditions rather than applying fixed reduction factors.
2Measurement precision
If detailed granular loss estimates are implemented, then the accuracy of performance modeling improves, but the complexity of the model increases
Solution Approach 1:
The model segments granular loss estimates into modular, independent calculation modules for each loss category. Each module processes specific inputs and produces a loss estimate that can be independently validated and adjusted, reducing overall model complexity while maintaining detailed accuracy.
Solution Approach 2:
The patent introduces intermediary parameters and intermediate calculation steps that bridge simple inputs and complex outputs. For example, intermediate metrics like 'shading fraction' or 'soiling accumulation rate' serve as mediators between basic measurements and final loss estimates, making the model more manageable and interpretable.
3Measurement precision
If complete information about site geometry and obstacles is obtained, then the accuracy of shading loss estimates improves, but the difficulty and cost of data collection increases
Solution Approach 1:
The patent performs preliminary actions by using readily available data (standard maps, typical obstacle heights, common vegetation patterns) to establish initial shading estimates before detailed site-specific measurements are made. This preliminary modeling allows for reasonable accuracy without requiring exhaustive data collection, and can be refined later if needed.
Solution Approach 2:
The model uses inexpensive, easily obtainable proxy data (satellite imagery, standard topographic maps, climate data) instead of expensive, time-consuming detailed surveys. These proxy measurements provide sufficient accuracy for most applications without requiring costly specialized equipment or extensive fieldwork.
Data Source
AI summary
The present invention provides methods and systems to estimate energy losses due to shading in PV systems from data including the measured energy and power produced over the lifetime of the system, the system size and configuration data, the weather conditions (including irradiance, ambient & panel temperature, and wind conditions) over the lifetime of the system, and derived meteorological condition information (e.g., decomposed irradiance values at any time).