Photovoltaic Plant Output Forecasting via Multi-Window Trending Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Forecasting the solar power output of a photovoltaics plant is challenging due to uncertainties in environmental and weather factors, which affects the accuracy of power generation predictions for grid management.
Innovation Solution
A method that forecasts data variables affecting solar power production, computes features from prior power generation data across different durations, determines a trending model, and predicts future power output by combining multiple models with varying window sizes and exogenous variables like cloud cover and temperature.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple trending models with different window sizes are combined, then forecasting accuracy is improved, but model complexity increases
Solution Approach 1:
The patent segments the forecasting problem by dividing it into multiple trending models, each handling a specific time window size (e.g., 1-hour, 2-hour, 4-hour windows). Each model processes a portion of the historical data independently, and their results are combined to produce the final forecast. This segmentation allows the system to capture different temporal patterns without requiring a single overly complex model.
Solution Approach 2:
The patent merges multiple trending models with different window sizes into a unified forecasting system. Each model's output is weighted and combined to produce the final power output prediction. This combining approach leverages the strengths of different window sizes (short-term trends from small windows, long-term patterns from large windows) to improve overall forecasting accuracy while distributing complexity across multiple simpler models.
2Reliability
If features from multiple time durations are computed, then prediction robustness is improved, but computational requirements increase
Solution Approach 1:
The computational task is segmented into multiple feature extraction operations, each processing historical power data within a specific time duration (e.g., 1-hour, 2-hour, 4-hour windows). Instead of computing all features from the entire historical dataset at once, the system divides the computation into manageable segments corresponding to different trending models, reducing peak computational requirements while maintaining comprehensive analysis.
Solution Approach 2:
The system computes features for multiple time durations beyond what a single model would require, extracting features from shorter, medium, and longer windows. This partial computation approach for each duration (rather than full computation across all data) provides robustness through diverse temporal perspectives while keeping individual computational loads manageable through selective feature extraction.
3Measurement precision
If forecasting accounts for multiple environmental factors, then prediction accuracy is improved, but data processing complexity increases
Solution Approach 1:
The trending models are designed with multi-functionality to handle multiple environmental factors (cloud cover, temperature, humidity) within a unified forecasting framework. Each model processes various environmental inputs simultaneously rather than requiring separate processing pipelines for each factor, reducing overall data processing complexity while maintaining comprehensive environmental analysis.
Solution Approach 2:
Multiple environmental factors are merged into the feature set for each trending model. Instead of processing each environmental variable separately through independent models, the system combines them into integrated features (e.g., composite weather indices) that are fed into the trending models, reducing data processing complexity while capturing interactions between environmental factors.
Data Source
AI summary
A method of predicting an amount of power that will be generated by a solar power plant at a future time includes: forecasting a value of a data variable at the future time that is likely to affect the ability of the solar power plant to produce electricity (S301); computing a plurality of features from prior observed amounts of power generated by the power plant during different previous durations (S302); determining a trending model from the computed features and the forecasted value (S303); and predicting the amount of power that will be generated by the power plant at the future time from the determined model (S304).