Machine Learning Predictor for PV Energy Flow Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Accurate predictions of instantaneous electric import values for photovoltaic solar systems are challenging due to energy consumption data having a lower resolution than the metering resolution used by electric utilities, leading to potential inaccuracies in energy and financial predictions under certain utility rate structures.
Innovation Solution
A computer system programmed to use machine-learning predictors trained with high-resolution electric energy consumption and photovoltaic production data to determine predicted import time series, minimizing self-consumption error by downscaling data to match utility metering resolution and simulating energy flows to provide accurate import value predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If energy consumption data is recorded at lower resolution to reduce data storage and processing requirements, then data storage and processing complexity is reduced, but prediction accuracy of electric import values deteriorates
Solution Approach 1:
The system performs preliminary training of machine learning models using high-resolution data before actual prediction operations. This pre-training phase stores learned patterns and relationships in model parameters, enabling accurate predictions from low-resolution input data without requiring high-resolution data during operational predictions
Solution Approach 2:
The machine learning model acts as an intermediary between low-resolution consumption data and the prediction of electric import values. The model bridges the resolution gap by applying learned relationships from high-resolution training data to interpret and predict outcomes from lower-resolution operational data
2Measurement precision
If high-resolution training data is used to train machine learning predictors, then prediction accuracy is improved, but data processing requirements and computational resources increase
Solution Approach 1:
The system performs comprehensive model training using high-resolution data in advance, before deployment. This preliminary action transfers the computational burden to the training phase, allowing the operational phase to use lightweight models that require minimal computational resources for predictions
Solution Approach 2:
The system creates a simplified computational copy of the high-resolution data relationships through the trained machine learning model. Instead of processing actual high-resolution data during operations, the system uses the model's learned representations that capture essential patterns with minimal computational overhead
Data Source
AI summary
Methods, systems, and computer storage media are disclosed for determining electric energy flow predictions for electric systems including photovoltaic solar systems. In some examples, a method is performed by a computer system and includes supplying a consumption time series and a predicted production time series for an electric system to a machine-learning predictor trained during a prior training phase using electric energy consumption training data and photovoltaic production training data. The consumption time series has a first data resolution, and the electric energy consumption training data and the photovoltaic production training data have a second data resolution greater than the first data resolution. The method includes determining, using an output of the machine-learning predictor, a predicted import time series of electric import values each specifying an amount of electric energy predicted to be imported by the electric system with a prospective photovoltaic solar system installed.