Power Quality Data Compression Based on Iterative PCA Algorithm in Smart Distribution Systems

Abstract

To reduce the stress of data transmission and storage for power quality (PQ) in smart distribution systems and help PQ analysis, a multichannel data compression based on iterative PCA (principal component analysis) algorithm is introduced. The proposed method uses PCA to reduce the redundancy of data to achieve the purpose of compressing data. In order to improve the calculating speed, an iterative method is proposed to compute the principal components of the covariance matrix. The correctness and feasibility of the proposed method are verified by field PQ data tests. Compared with discrete wavelet transform (DWT) method, the proposed method has good performance on compression ratio and reconstruction accuracy.

Algoritmo para a compressão de sinais da rede elétrica baseado na combinação da técnica Compressive Sensing com uma abordagem dissociativa

“…Em (Abuadbba et al, 2018), aplicam-se a transformada Burrow -Wheeler (BWT) e a aproximação gaussiana para essa finalidade. Em (Zhang et al, 2017), emprega-se um algoritmo iterativo baseado na análise de componentes principais (PCA), como um método de compressão de dados de multicanais de qualidade de potência. Os métodos anteriormente propostos tratam o sinal após sua aquisição.…”

Estimation of an Extent of Sinusoidal Voltage Waveform Distortion Using Parametric and Nonparametric Multiple-Hypothesis Sequential Testing in Devices for Automatic Control of Power Quality Indices

“…Statistical analysis [39]; • Logistic regression [40]; • Principal component analysis [41]; • K-nearest neighbors method [42]; • Wald's sequential analysis [43];…”

“…These methods should ensure that a decision on PQI deviation from standard values is made in real time. In this context, the implementation of methods for PQI control based on machine learning [33][34][35][36][37][38][40][41][42] is difficult due to the complex process of simulation modeling of the power supply system of an industrial consumer. Statistical methods [39,43] show great promise.…”