The significant meaning of power quality comprehensive evaluation (PQCE) is disserted in the first part. Two new methods of PQCE based intelligence, which including artificial neural network (ANN) and genetic projection pursuit method (GPPM), were proposed in the paper. ANN based PQCE made the training samples according to the random-distribution theory on the basis of grade of each power quality index. The radial basis function (RBF) neural network-based model for PQCE has been made with the trained network. GPPM based PQCE constructed the projection function with the power quality indexes, and made the optimization by genetic algorithm. The GPPM-based model for PQCE has been made finally. Then the regional grid's harmonic had been evaluated by the GPPM and eigenvalue weighted methods. The paper compared the advantages and diadvantages of probability and vector algebra, fuzzy, matter-element methods and the methods which were proposed in this paper by the same pratical examples. The comparison results improved that the methods which were proposed in this paper are impersonal and reasonable. It did a significative exploration of intelligentized PQCE.
|