Low Sampling Rate Electrical Load Disaggregation via Graph Signal Propagation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods face challenges in accurately disaggregating electrical loads at low sampling rates due to limited data granularity, resulting in poor consumption estimates and inability to determine individual load signals effectively.
Innovation Solution
A system and method that utilize a combination of sparse dictionary representation and graph signal smoothness-based label propagation for energy load disaggregation, constructing a Laplacian graph and optimizing dictionary coefficients to estimate load consumption, iteratively refining estimates from aggregate power data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If low sampling rate aggregate power measurements are used, then energy consumption data can be obtained at regular intervals from smart meters, but load disaggregation becomes tedious due to unavailability of events and signatures of constituent loads
Solution Approach 1:
The system performs preliminary training by collecting high-resolution power measurements during a training period to learn appliance signatures and characteristics. This pre-acquired knowledge is stored and later applied during the test phase with low-resolution data, enabling accurate disaggregation without requiring high sampling rates during actual operation
Solution Approach 2:
The patent introduces an intermediary approach by using learned appliance signatures from training data as a bridge between the aggregate low-resolution measurements and individual load identification. These signatures act as reference patterns that enable the system to decompose low-resolution aggregate data into individual appliance consumption patterns
2Measurement precision
If conventional Graph Signal Processing methods are used for low sampled power data, then load identification can be performed using graph smoothness assumptions, but consumption estimates are poor as they are merely the average consumption of the load
Solution Approach 1:
The system performs preliminary training by collecting high-resolution power measurements during a training period to learn appliance signatures and characteristics. This pre-acquired knowledge is stored and later applied during the test phase with low-resolution data, enabling accurate disaggregation without requiring high sampling rates during actual operation
Solution Approach 2:
The patent changes the parameter representation by learning appliance signatures in the time domain from high-resolution training data, then applying these signatures to interpret low-resolution measurements. This parameter transformation allows the system to recover detailed consumption patterns from coarse measurements by matching against learned appliance behavioral patterns
3Ease of operation
If matrix factorization with graph shift quadratic form constraint is used, then load disaggregation can be performed on low sampled data, but individual load signals are estimated as piece-wise constant values instead of real values
Solution Approach 1:
The patent replaces the matrix factorization mechanical approach with a signature-matching methodology. Instead of factorizing the measurement matrix and applying graph constraints, the system substitutes this with a direct matching approach where learned appliance signatures are correlated with the aggregate measurements to identify individual load contributions, yielding more physically meaningful real-valued estimates
Data Source
Figure 1
Figure 2
Figure 3
AI summary
This disclosure relates generally to method and system for low sampling rate electrical load disaggregation. At low sampling rates, disaggregation of energy load is challenging due to unavailability of events and signatures of the constituent loads. The disclosed energy disaggregation technique receives aggregated load data from a utility meter and sequentially obtains training data for determining disaggregated energy load at low sampling rate. Dictionaries are used to characterize the different loads in terms of power values and time of operation. The obtained dictionary coefficients are treated as graph signals and graph smoothness is used for propagating the coefficients from the training phase to the test phase by formulating an optimization model. The derivation of the optimization model identifies the load of interest and estimate their power consumption based on optimization model constraints. This method achieves accuracy greater than 70% for the loads of interest at low sampling rates.