Photovoltaic Power Forecasting via Singular Value Decomposition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The integration of large-scale photovoltaic power stations into power grids is hindered by the volatility and randomness of photovoltaic power generation due to weather and environmental factors, leading to adverse effects on power grids.

Innovation Solution

A method for photovoltaic power forecast based on numerical weather prediction, which involves reading historical and predicted light intensities and temperatures, determining key weather feature matrices, performing singular value decomposition, and using a fitting relationship to predict photovoltaic power, combining physical models with data-driven functionality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Power

If photovoltaic power stations are merged into power grids in large scale, then power supply capacity is improved, but grid stability deteriorates due to volatility and randomness

Engineering Contradiction:
Improvepower supply capacityVSAvoidgrid stability
Core Design Contradiction:
PowerVSReliability

Solution Approach 1:

The patent performs preliminary weather forecasting and photovoltaic power prediction before actual power generation occurs. By using numerical weather prediction models to forecast weather parameters and then predicting photovoltaic power output based on these forecasts, the system enables advance preparation for power grid scheduling and dispatch, allowing operators to anticipate volatility and plan accordingly, thus maintaining grid stability while accommodating large-scale photovoltaic integration

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent establishes a feedback mechanism by continuously monitoring actual photovoltaic power output and comparing it with predicted values. The system uses historical weather data and actual power generation data to refine prediction models, creating a closed-loop system that improves prediction accuracy over time and enables dynamic adjustment of grid operations to maintain stability

Inventive Principle:
Principle #23Feedback

2Measurement precision

If weather and environmental factors are considered in photovoltaic power generation, then prediction accuracy is improved, but system complexity increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the complex prediction system into distinct modules: a numerical weather prediction module that forecasts weather parameters, and a photovoltaic power prediction module that calculates power output based on weather forecasts. This segmentation allows each module to specialize in specific tasks, improving overall prediction accuracy while managing complexity through modular design where each component can be independently optimized and maintained

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transforms complex weather and environmental conditions into standardized numerical parameters that can be processed by prediction models. By converting qualitative weather conditions into quantitative parameters and using mathematical relationships to predict photovoltaic power output, the system achieves high prediction accuracy while maintaining computational tractability

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS10910840B2Method and apparatus for photovoltaic power forecast based on numerical weather prediction
Publication Date: 2021.02.02 BEIJING TSINTERGY TECH CO LTD
  • US10910840B2 patent drawing
  • US10910840B2 patent drawing
  • US10910840B2 patent drawing

AI summary

The present disclosure provides a method and an apparatus for photovoltaic power forecast based on numerical weather prediction. The method includes: determining a historical key weather feature matrix and a prediction key weather feature matrix; determining a historical weather data matrix and a prediction weather data matrix; determining a historical input matrix and a prediction input matrix; combining the historical input matrix and the prediction input matrix; performing singular value decomposition on the combined input matrix to obtain a principal component feature matrix; determining K principal component features corresponding to K historical time periods having the nearest Manhattan distances with the prediction time period; acquiring a fitting relationship according to the K principal component features and K photovoltaic powers corresponding to the K historical time periods; inputting the principal component feature corresponding to the prediction time period to the fitting relationship to obtain a photovoltaic power.