High Power Series Resonant Injection Type Hybrid Active Filter

This dissertation is focused on SRITHAF — series resonant injection type hybrid active filter, a novel device for substation harmonic elimination. The study mainly includes the new topology, the design optimization of output filter and snubber, the control method and the implementation of SRITHAF.Through the analysis and comparison of the existing active filters, a novel SRITHAF is proposed in order to meet the need of substation harmonic elimination. The principle is that active filter of SRITHAF is equivalent to a harmonic impedance series connected with the main line. The harmonic impedance can hardly hinder fundamental current but extremely hinder harmonic current flowing into the grid. Besides, the passive filters built in SRITHAF also have a function of reactive power compensation. The fundamental component of voltage across the inverter approximates zero because the inverter is parallel with the fundamental series resonant circuit. As a result, the rating of the inverter is reduced. In addition, the eliminating harmonic performance and the resonance damping capability of SRITHAF are studied under the situation of variety of power source impedance and mistuning of passive filters.The existing design methods of output filter mainly depend on engineering experience with the low efficiency inverter and harmonic compensation capability of the system. An optimization design method for output filter is proposed. In the frame of high power active filters, the proposed design method analyzes the influence of output filter on the capability of eliminating harmonic. Thoroughly considering the requests of the output performance, the eliminating on-off harmonic performance and power loss, the proposed method builds the math model of optimization and uses the genetic algorithm to optimize. Output filter designed by the proposed method can not only effectively eliminate on-off harmonic but also enhance the efficiency of the inverter. Besides its power loss is low.Based on the analysis of the operation principle of snubber, the theory formulas of the turn-off voltage of IGBT are concluded and amulti-objective optimization design method of snubber is proposed. Considering the limit of diode rating, it takes the lowest turn-off voltage and the minimum cost as targets. Then it uses fuzzy optimization algorithm to search the best parameters for snubber in order to obtain the comprehensive best performance. Snubber designed by the proposed method does not need the application of parallel diodes and are more fit for engineering application.From three aspects of dynamic performance, steady-state performance and on-off harmonic, some common control methods are analyzed and their characteristics are concluded. A recursive integral PI control method based on an improved single neuron parameter self-tuning algorithm is proposed. According to the periodic character of the reference signal, the recursive PI algorithm is proposed to archive zero steady state error. The tuning strategy considers the controller parameters as the product of the proportional coefficient and weight coefficients of the improved single neuron. In big error state the fuzzy inference on-line tunes the proportional coefficient to enhance the dynamic performance. In small error state the self-study algorithm on-line tunes the weight coefficients to improve the robustness and adaptive ability. Moreover the least-square style algorithm is adopted to reduce system model. Based on the reduced model the tuning scope of controller parameters is obtained in order to ensure the system stability.An eliminating harmonic device for substation based on SRITHAF is developed. The system includes two functions that are harmonic eliminating and power system monitoring. The current control is archived by a DSP system and the monitoring function is archived by an industry application computer. It can not only enhance power quality but also be useful for the management of power system.