Research On Classification Of Power Quality Disturbances Based On Fuzzy Logic

With the rapid development of new type complex and sensitive electric loads, the requirements from user side of power grid on the power quality are increasing continuously, and the issues regarding power quality are getting more and more attentions. Among them, the classification of power quality disturbances(CPQD) is not only an essential prerequisite for evaluating and improving the power quality, but also an important topic related to analysis and control of power quality. Due to the complexity and diversity of disturbances and the difficulty of knowledge summarization and acquirement in fuzzy logic in CPQD, the parameter determination process of fuzzy classifier is complicated and too difficult to achieve the optimal performance. This thesis focuses on the disturbance data and feature extraction of CPQD based on fuzzy logic method, and the design, optimization, and classification efficiency of classifier, etc. The study mainly includes several parts as follows:First of all, to obtain power quality disturbance data with large amount of information, an extraction method of power quality disturbance data is proposed based on the improved normalized distance measure functions. By analyzing the profile of power quality disturbances, both global and local measurement factors are proposed which can suitably characterize the power quality disturbance signal, and the improved normalized distance measure functions are formulated. By using the distance measured value calculated via the improved distance measure functions, whether the signal at current period should be saved is determined. The determination method of the distance measured threshold value is also presented, and the abilities of the proposed method in identifying the disturbances from power quality disturbance data and extracting disturbance data are verified. This study can provide essential data basis for achieving high accuracy CPQD.For artificial parameter estimating problem in the design of fuzzy classifier for power quality disturbances, a classification method of power quality disturbances is proposed based the T-S type fuzzy inferences. The multi-resolution S-transform module is adopted with this method to extract power quality disturbance features which represent the essential characteristic of disturbance signal, and provide input values for the fuzzy classifier. Afterwards, power quality disturbance classifier is designed by bringing in T-S fuzzy reasoning mechanism. The classifier parameters such as the center and width values of membership functions are confirmed by using equalized universe method based on the fuzzy cluster. And the mapping relations between fuzzy inference conclusion and classification conclusion are established. Base on this, the fuzzy classification rules are set up, and the fuzzy rule base is generated. The feasibility of the proposed classification method regarding single and complex power quality disturbances is verified, and the efficient classification of power quality disturbances is fulfilled.In order to improve the accuracy rate of fuzzy-CPQD, a T-S fuzzy classification optimal method of power quality disturbances is proposed based on the improved bacterial foraging optimization(BFO). By adding mutation operators to the chemotactic operation of BFO for regulating the current location of bacterial, searching scope is expanded on the basis of keeping the roughly research direction unchanged. Furthermore, appropriate parameters of modified BFO are selected through parameter comparison, and the global searching ability and optimal performance of the proposed method are verified. The optimization of reasoning accuracy and classification performance in the T-S fuzzy classifier of power quality disturbances are fulfilled.Finally, with accuracy prerequisite satisfied, a classification method of power quality disturbances based on pattern linguistic values(LVs) is proposed to improve the fuzzy classification efficiency of power quality disturbances. For the differences between the simplified fuzzy rules requirement in fuzzy-CPQD and the conventional fuzzy reasoning mechanism, this method takes the classification of power quality disturbances as the LVs of the membership functions, presents the design method of the fuzzy classifier based on the pattern LVs, determines the membership functions, and generates the fuzzy rules. The integration among classification purpose, the LVs selection, and the membership function design is achieved which provides an important foundation for the simplification of the fuzzy reasoning process and fuzzy rule base of the classifier.