Hierarchical Load Prediction for High-Variance Power Usage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting electrical power consumption, particularly for high variance loads like HVAC systems, are inefficient and inaccurate, leading to inefficient electrical distribution and increased charges due to the inability to accurately forecast energy usage.
Innovation Solution
A hierarchical machine learning approach that classifies input data into different modes of operation, using a classifier to determine the current mode and a regression model to predict power consumption, allowing for more accurate predictions by refining models for specific operational modes without requiring expert training or precise measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single predictive model is used for all operational modes, then the model complexity is low, but the prediction accuracy deteriorates for high variance loads
Solution Approach 1:
The patent divides the operational data into multiple segments based on different modes of operation (e.g., HVAC on/off states, high/low demand periods). Each segment is then trained with a separate predictive model, allowing each model to specialize in predicting loads for its specific operational mode. This segmentation resolves the contradiction by improving prediction accuracy through mode-specific models while managing complexity through modular model structures.
Solution Approach 2:
The patent applies different predictive models with varying complexities to different operational modes. For example, simpler models may be used for stable low-variance modes while more complex models handle high-variance modes. This local quality approach allows the system to optimize prediction accuracy for each specific operational context without uniformly increasing complexity across all modes.
2Measurement precision
If expert training and precise measurements are required for physics-based models, then prediction accuracy may be improved, but the ease of operation and implementation deteriorates
Solution Approach 1:
The patent employs machine learning models that automatically learn patterns from historical load data without requiring expert training or manual parameter tuning. The system self-adjusts by training on available data and adapting to different operational modes autonomously, eliminating the need for expert intervention while maintaining high prediction accuracy for high variance loads.
Solution Approach 2:
The patent transforms the approach from physics-based models requiring precise physical measurements to data-driven models that learn from operational patterns. By changing from physical parameters (temperature, humidity, equipment specifications) to statistical parameters (historical load patterns, operational modes), the system achieves comparable or superior accuracy without requiring expert knowledge or precise physical measurements.
Data Source
AI summary
A method and system are provided for improving predictions of electrical power usage. In the method and system, load and/or environmental data is classified into data sets that correspond to different modes of operation of an electrical load. Different predictive models are also provided for each set of classified data. The predictive models may provide more efficient and/or more accurate predictions of power usage since each model is limited to a particular mode of operation.