Building Power Load Disaggregation From Low-Sampling Meter Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for power load disaggregation in buildings using low-sampling rate data face challenges such as detecting power variations caused by internal electric circuits, similar power consumption patterns among larger appliances, and overlapping power consumption due to appliances operating at different levels.
Innovation Solution
The proposed solution involves detecting background power loads using dynamic thresholds, employing a robust event detection mechanism to identify power changes, pairing events using an iterative event pairing technique, forming event clusters with a density-based clustering technique, and classifying appliance types using a rule-based classification technique.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If low-sampling rate metering data is used for power load disaggregation, then data processing complexity is reduced, but measurement precision deteriorates due to averaging of power consumption
Solution Approach 1:
The patent segments the aggregated power consumption signal into distinct appliance-level components by identifying unique power signatures and operational patterns. This segmentation allows low-sampling rate data to be decomposed into individual appliance contributions, recovering detailed power variation information that would otherwise be lost through averaging.
Solution Approach 2:
The patent performs preliminary characterization of appliance power signatures and operational patterns before analyzing the low-sampling rate data. By pre-establishing reference profiles of how different appliances consume power, the system can accurately attribute power variations even in down-sampled data, compensating for the loss of temporal resolution.
2Ease of manufacture
If conventional algorithmic techniques are used for power load disaggregation, then implementation simplicity is improved, but reliability deteriorates due to overlapping power consumption patterns
Solution Approach 1:
The patent introduces intermediary features such as power signature profiles, operational state transitions, and temporal pattern descriptors that mediate between the raw aggregated data and the final disaggregation results. These intermediaries capture distinctive characteristics of each appliance, enabling reliable differentiation even when power consumption patterns overlap, while maintaining algorithmic implementation.
Solution Approach 2:
The patent transforms the disaggregation approach by changing key parameters from simple power magnitude analysis to multi-dimensional feature space including power signatures, operational states, temporal patterns, and transition behaviors. This parameter transformation enables distinction between appliances with similar power consumption by exploiting differences in their operational characteristics.
3Measurement precision
If high-sampling rate data is used for power load disaggregation, then measurement precision is improved, but data processing complexity increases
Solution Approach 1:
The patent extracts and removes background power loads and non-informative data components from the high-sampling rate data before processing. By separating the signal into informative appliance-specific variations and uninformative background noise, the system maintains measurement precision while reducing the complexity of subsequent processing to only the essential disaggregation tasks.
4Quantity of substance
If background power loads are not removed, then data completeness is improved, but measurement precision deteriorates due to interference with appliance event detection
Solution Approach 1:
The patent performs preliminary identification and separation of background power loads before appliance event detection. By characterizing and removing baseline consumption patterns associated with always-on devices and environmental factors, the system preserves the complete appliance event signatures while eliminating interfering background noise that would otherwise mask or distort event detection.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
This disclosure relates generally to methods and systems for determining the power load disaggregation profile of a building. Most of the conventional techniques are algorithmic centric, specific to certain scenarios and does not employ the low-sampling rate data due to the complexity involved. Present disclosure determines the power load disaggregation profile of the building using the low- sampling rate power consumption data accurately. According to the present disclosure, firstly, the background power loads are detected and removed from the low- sampled data samples. Next, a robust event detection mechanism is employed to detect the events when the change in the power consumption occurred, and such events are paired using the iterative pairing technique. Further, a set of event clusters are formed using the density-based clustering technique and lastly, each of the set of event clusters are classified with each appliance type using a rule-based classification technique.