Load Decomposition at Smart Meters Level Using Eigenloads Approach

Abstract

The deployment of the advanced metering infrastructure (AMI) in distribution systems provides an excellent opportunity for load monitoring applications. Load decomposition can be done at the smart meters level, providing a better understanding of the load behavior at near-real-time. In this paper, loads' current and voltage waveforms are processed offline to form a comprehensive library. This library consists of a set of measurements projected onto the eigenloads space. Eigenloads are basically the eigenvectors describing the load signatures space. Similar to human faces, every load has a distinct signature. Each load measurement is transformed into a photo and an efficient face recognition algorithm is applied to the set of photos. A list of all the online devices is always stored and can be accessed at any time. The proposed method can be implemented at the smart meters level. The distributed computation that can be achieved by performing simple calculations at each smart meter, without the need for sending intensive data to a central processor, is beneficial. From a system operator perspective, load composition in near-real-time provides the loads' voltage dependence that are needed, for example, in volt-VAR optimization (VVO) in distribution systems. Further applications of load composition data are also discussed.

Electric Load Data Compression and Classification Based on Deep Stacked Auto-Encoders

“…In terms of CR and RMSE, the compression performances of SAE (with fine-tuning) and several other methods are compared on case 1, including PCA [24], DWT [25,26] and k-SVD [7,21]. Figure 10 displays the results.…”

“…Users should be classified into more specific and meaningful types. In non-intrusive load identification, the readings are investigated to match the household appliances [24,31], e.g., refrigerator and air conditioner.…”

Non-Intrusive Load Monitoring System Based on Convolution Neural Network and Adaptive Linear Programming Boosting

“…Several NILM systems have been studied, which could roughly be divided into two categories. One is the method applying mathematical optimization [8,10,11,13,29], and the other one is based on pattern recognition [14][15][16][17][18][19][20][21][22][23][24][25]30].…”

“…In 2015, Ahmadi et al [29] first merged the loads' current and voltage waveforms into a comprehensive library. Then they found that each load of the appliance is similar to the Face recognition [33].…”

“…The NILM systems [24][25][26]29,35,36] based on unsupervised learning could effectively weaken the impact of the tag data of the appliances, which means that the manual intervention could be reduced and the utility could be enhanced. These methods generate the appliances' type through the steady-state characteristics of the monitoring data to screen out the status of the appliances.…”