The Classified Research Towards The Transient State Power Quality’s Disturbance Based On QPSO-WNN

With the development of our country’s power system,the degree of voltage and level of automation has been arising gradually,the structure and control means are so increasing complicated that user demanding more and more strict claim on the quality of electrical power.The power quality as a criteria of measuring system’s benefits and drawbacks, has been gradually drawing close attention from different areas. Thus to improve the quality of electrical power became the researching center for varies subjects. Besides, detection, localization and classification of the disturbance are the prerequisite link and important premise to improve the power quality.The thesis mainly studied the detection and classification of the transient state power of the quality’s disturbance, and put forward a classified method about the wavelet neural network and distraction of the transient state power quality which based on the optimized quantum-particle group.Firstly, the author took advantage of the multi-distinguished and multi-analytical feature of wavelet transform, and took multilayer analyzing and discomposing of the disturbing signal to many frequency band. over each disintegrate layer different threshold was installed, adopting wavelet transform and self-adaption method to conduct de-noising over the transient state power quality’s signal disturbance; Secondly, based on the wavelet transform Modulus maxima method, the author conducted detection and position of the singular point which distract signal; Lastly, the author extracted the energy of transformed wavelet’s coefficient and the energy of referenced signal, setting the difference between the two above energy as the wavelet neural network’s input characteristic vector. Then the author input these vectors into the network for testing. Though this process, the classification of several kinds of the transient state power quality’s disturbance’s can be showed.The wavelet neural network is the combination of wavelet transform and neural network which has sharing merits, and it was widely used to distinguish classifications. However, the time of network training is very long and it was easily involved in the local optimum, so the author introduced the Particle swarm optimization to conduct optimization. In order to improve the algorithm and avoid the defects of the traditional Particle swarm optimization, the thesis introduced the quantum theory. The quantum particle swarm optimization need to update the position instead of calculating velocity function of particles. It is easy to calculate for less parameters need to be set.Based on the grader of QPSO-WNN and the comparison to Wavelet Neural Network, the error of the training process is limited. And it can conduct classification of the phenomenon disturbed by the transient state power quality. By using the software of Mat lab, the author realized the detection and classification of the transient state power quality’s disturbance. Thus the author proved the exactness and validity of the method.