A power-quality event data compression algorithm based on advanced Support Vector Machine

An algorithm using SVM (support vector machine) regression for power-quality (PQ) event data compression is presented, the advanced SVM regression can learn dependency from an array of wavelet coefficients which is 2-D representation of PQ event data decomposed by 2-D discrete-time wavelet transform (2-D DWT), using fewer Support Vectors (SV) to represent the original data, thus, the data could be compressed based on this feature. Experiment results show the good performance of proposed algorithm in PQ event data compression, comparing with traditional SVM compression data under the same conditions.