SVM Load Identification for Mixed Electric Loads
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying miscellaneous electric loads (MELs) are limited in accuracy, robustness, and applicability, particularly due to the diversity of power supply topologies and the lack of granular load energy consumption information, which hinders energy management and savings in residential and commercial sectors.
Innovation Solution
The use of a support vector machine (SVM) based identification system, combined with a supervised self-organizing map (SSOM), to classify and identify MELs by sensing voltage and current signals and determining load feature vectors, allowing for more accurate and robust differentiation of MELs with similar features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If steady-state features are extracted from voltage and current measurements to identify electric loads, then the identification process becomes simpler and faster, but the accuracy and robustness of distinguishing similar MELs deteriorates
Solution Approach 1:
The patent transforms the identification approach by changing from using raw steady-state features directly to using a trained neural network model that processes these features. The neural network learns optimal parameter transformations during training, enabling accurate distinction between similar MELs while maintaining fast identification speed during operation.
Solution Approach 2:
The patent implements preliminary training of a neural network model using labeled training data before actual load identification. This preliminary action pre-computes the optimal feature mappings and decision boundaries, so that during runtime, only fast forward propagation is needed, achieving both high accuracy and speed.
2Measurement precision
If computational intelligent algorithms such as RBF, PSO, and ANN are used to identify MELs, then the accuracy of distinguishing similar loads improves, but the computational cost and complexity increases
Solution Approach 1:
The patent replaces complex iterative optimization algorithms (like PSO and RBF) with a feedforward neural network that uses simple forward propagation for identification. The complexity is shifted to the offline training phase, while runtime operation becomes computationally lightweight, effectively substituting complex mechanical/computational processes with a streamlined system.
3Ease of manufacture
If a predefined database of load features is used for comparison-based identification, then the system is easier to implement, but it cannot distinguish MELs with similar front-end power supply units
Solution Approach 1:
The patent uses a predefined database during the training phase to teach the neural network the characteristics of different MELs. The neural network then learns to go beyond simple database matching by identifying subtle patterns and features that distinguish similar loads, combining the ease of predefined data with enhanced discrimination capability.
Solution Approach 2:
The neural network acts as an intermediary between the predefined database and the identification process. Instead of directly comparing measured features against database entries, the neural network processes the features through learned transformations, enabling distinction of similar MELs while still utilizing the structured knowledge in the database.
Data Source
AI summary
A method identifies electric load types of a plurality of different electric loads. The method includes providing a support vector machine load feature database of a plurality of different electric load types; sensing a voltage signal and a current signal for each of the different electric loads; determining a load feature vector including at least six steady-state features with a processor from the sensed voltage signal and the sensed current signal; and identifying one of the different electric load types by relating the load feature vector including the at least six steady-state features to the support vector machine load feature database.