The Research Of Hybrid Active Power Filter

Nowadays, the power quality problems have become a serious concern in industry, electrical distribution systems and people’s lives. This is especially so with the wide use of the non-linear loads in the grid which cause the supply current and voltage to experience distortion thereby causing harmonic pollution and energy losses. In order to solve harmonic problems, the Passive Power Filter (PPF) is often used. Although, it has simple structure, least expense and easy use, the PPF inherits several disadvantages due to resonance, instability, mistuning, difficulty to improve total harmonic distortion and no online compensation capability, etc. Thus, the Active Power Filter (APF) appears to be the solution for eliminating harmonic current and providing reactive power compensation. It is often parallel-connected with a nonlinear load and has capability of online compensation and tracking harmonic current of a nonlinear load. Nevertheless, it is limited by high cost, low-power capacity and is difficult to use in high-voltage grids. Another solution to effectively handle the harmonic problem is to adopt a hybrid active power filter (HAPF). The HAPF is the combination of active power filter and passive power filters. The aim of HAPF design is to reduce APF capacity. Besides, the HAPF inherits the advantages of both the passive filter and the active power filter. The IHAPF is a novel HAPF with injection circuit; it has great promise in reducing harmonics with a relatively low capacity APF. The working mechanism, mathematical model,parameter design, control strategy, and the application of IHAPF are described in detail in this thesis.The emphasis and achievement of the thesis mainly are manifested in the following aspects:1) Overview about the interesting problems of the electric power quality, IEEE harmonic standards, effect of harmonic distortion on electric power quality and harmonic elimination methods.2) Based on the disadvantages of the p-q and ip-iq harmonic current detection methods are usually used IIR filters because the IIR filters are infinite and used for applications where linear characteristics are not of concern, better for lower-order tapping, recursive and used as an alternate, whereas FIR filters have become too long and cause problems in various applications. To improve the disadvantages of the p-q and ip-iq method, a novel harmonic current detection method using neural network-fuzzy logic is proposed. Generally, the nonlinear load current can be expressed as the sum of the fundamental and harmonics components. In this proposed method, the selected neural network structure is Adaline structure. The weight factors of the Adaline network respectively with magnitudes of the component currents. Here, the Adaline network is as the same a harmonic identification model. The error value between the output of neural network and input of neural network is used to adjust the weights of the neural network. Training algorithm using the Least Mean Square (LMS) algorithm or Widrow-Hoff learning algorithm is based on an approximate steepest descent procedure. The Widrow-Hoff learning algorithm had the insight that they could estimate the mean square error by using the squared error with respect to the weights. Moreover, a special point of this method is the convergence coefficient a of the neural network is adjusted by the fuzzy logic controller. Contrarily, a smaller a can decreased the stability error, but the convergence rate of system will be very slow. To ensure stable learning, the convergence coefficient must be less than the reciprocal of the largest eigenvalue of the correlation matrix of the input vectors. Compared to the conventional p-q and ip-iq harmonic current detection methods, the simulation results show that the proposed harmonic current detection method has the advantages of shorter dynamic response time and harmonic current detection more accurate.3) Control analysis of the typical Hybrid Active Power Filter topologies is proposed. From the disadvantages/advantages analysis of the Hybrid Active Power Filter topologies, the Hybrid Active Power Filter with Injection Circuit (IHAPF) topology is selected. The control strategies and parameters design for IHAPF topology are introduced. Based on the chosen parameters and the control strategies analysis results of IHAPF, the control strategy based on load harmonic current is the best control strategy of IHAPF.4) Because, the structure of IHAPF combines many elements connected together. As the result, there will be a phase delay in the output current. Therefore, this thesis proposes a novel online control method with time delay compensation for IHAPF. It includes two closed-control loops:the lower closed-control loop contains a conventional PI controller and IHAPF model, whereas the upper closed-control loop is composed of predictive neural network, generalized predictive control (GPC) and IHAPF model. The purpose of the upper closed-control loop is to find a set of optimal PI parameters for the lower closed-control loop. The goal of the predictive neural network model is to determinate a structure that emulates the nonlinear process of IHAPF. A7t-Smith predictive compensator is established to generalized current time delay compensate and makes a180degree delay between the output current of IHAPF and load harmonic current. The proposed control method is analyzed for stability by Lyapunov theorem. Compared to the conventional PI control method, the simulation and experimental results show that the proposed control method has the advantages of shorter dynamic response adjust time, effectiveness in reducing harmonics and able to online control very well.5) A novel control method for IHAPF using Hysteresis-adaptive fuzzy neural network is proposed. This method has the advantages of both Hysteresis and adaptive fuzzy neural network control. It includes two closed control loops. The lower control loop contains a conventional Hysteresis controller and IHAPF model, while the upper control loop is composed of a fuzzy-neural controller, identify and predictive model and IHAPF model. The purpose of the identification and predictive model is to find a set of optimal parameters for the fuzzy-neural controller. Therefore, the fuzzy-neural controller parameters are adjusted according to the cost function minimum criterion. For this reason, the proposed control method has a capability online control following variable of the harmonic currents. There are two control modes in the proposed control scheme are Hysteresis control and adaptive fuzzy-neural control modes. The selecting between two modes is decided by multi-modes switch K. The proposed control method is analyzed for stability by Lyapunov theorem. Compared to the conventional Hysteresis control method; the proposed control method shows the advantages of shorter dynamic response adjust time and able to online control very well. Simulations verified the effectiveness of this control method.6) A novel control method for IHAPF using Hybrid fuzzy is proposed. It consists of two closed-control loops. The upper closed-control loop consists of a single fuzzy logic controller and IHAPF model, while the lower closed-control loop is composed of an Adaptive Network based Fuzzy Inference System (ANFIS) controller, a Neural Generalized Predictive (NGP) regulator and IHAPF model. The purpose of the lower closed-control loop is to improve performance for the fuzzy logic controller. A single fuzzy logic controller for IHAPF is presented at first. At the same time, the output signals of ANFIS controller are added to the output of single fuzzy logic controller. Therefore, the output of the proposed controller will be changed and the purpose is to reduce the response time and minimize error in steady-state. This new control method is very suitable for nonlinear controls. Compared to the single fuzzy logic control method, the simulation results show that the proposed control method has the advantages of shorter dynamic response adjust time, able to online control and effective in reducing harmonics.Besides, the online control method with time delay compensation for IHAPF in this thesis is also successfully implemented in the IHAPF prototype in the laboratory, and the IHAPF topology is successfully applied with a100-kVA APF system for compensating a large power rated industrial rectifier in a copper mill in Northern China.