The Control And Islanding Fault Detection Of Distributed PV Power Generation Systems

Due to the increasing scarcity of traditional energy sources, the urgent need of improveing environmental pollution and the green awakening of public, solar, wind, biomass and other renewable energy sources are gradually developed as alternative energy to traditional fossil sources. Especially, solar energy has been widely used because of large reserves, widely distributed, almost zero pollution and other characteristics. Photovoltaic (PV) power generation is one of the main scientific utilization ways of solar energy. Because of the great advantages of PV power generation, the PV industry has rapid development in recent years, and there have been a variety of different capacities and sizes PV power generation systems, in which the distributed PV power generation is mainly used in the medium and small rooftop PV systems. Distributed PV power generation advocates the principle of proximity to realize power generation, power conversion, feeding the grid and supplying power to local load. It can effectively improve the PV system generation capacity and reduce energy consumption caused by boost conversion and long-distance transportion.In this dissertation, the distributed grid-connected PV systems are primarily analyzed and investigated. The topologies of different PV systems are analyzed to determine their respective application scopes. Combining several key issues to be solved of distributed PV systems:distributed equalization control, the maximum power point tracking (MPPT) optimal control, islanding fault detection, etc., this dissertation makes detailed investigation and validation. The main contents of this paper can be divided into the following sections:(1). The output characteristics of PV array which constitutes distributed PV system, DC/DC converter circuit topology, the topology and modeling of PV inverter and PV power generation system grid-connected control are analyzed and investigated. The selection method of the main model parameters is given. Three main issues come out of the above parts, namely distributed equalization control problem, MPPT optimal control problem and islanding fault detection problem.(2). Network equalization control is proposed based on the PV unit and distributed PV systems model analysis. The network structure of distributed PV power generation system is analyzed by graph theory approach. On the basis of graph theory network equalization control is designed to make those geographically dispersed and output capacity varying PV power units output power according to the same output rate. Then the mixed sensitivity H∞local controller based on the state space is designed in the framework of network equalization control. In this way, PV systems have good robustness for the grid parameters disturbance. For the problem of state variables unmeasurable, distributed H∞equalization control based on observer and output feedback is proposed to achieve equalization control objectives. A5 machines14bus distributed PV power generation system is designed through MATLAB/SIMULINK to verify the proposed distributed equalization control methods.(3). PV cells are nonlinear direct current (DC) sources, the output curves have strong nonlinear characteristic. According to the input and output characteristics of PV cells, a mathematical model is established. A simulation model of PV cell is builded based on this mathematical model for the following investigation. According to the problems of the current MPPT algorithms, an adaptive variable step incremental conductance (INC) MPPT optimal algorithm is proposed to solve the contradiction of tracking speed and tracking accuracy. Shadowing effect is one of the major reasons for low efficiency of the PV array output. The multi-peak phenomenon caused by the shadowing effect is analyzed in this dissertation, and the intermittent global scanning strategy based on adaptive variable step INC method is proposed to solve this problem. Simulation results show that this method can effectively avoid local optimization problem due to shadowing effect.(4). Islanding fault is an important safety issue for distributed PV power generation systems. The causes of islanding fault and reasons of non-detection zone (NDZ) are analyzed. Based on the shortcomings of existing islanding detection methods, an adaptive phase drift (APD) islanding detection and an islanding detection based on set-membership filtering (SMF) are proposed, respectively. The former adopts phase drift to indirectly achieve frequency drift to minimize the initial disturbance for reducing the influence of output power quality. The latter is proposed based on the analysis of actual grid voltage waveforms. The quasi-proportion resonant (QPR) superimposed harmonic compensation (HC) algorithm reduces the output harmonics of inverter side, thus highlights the grid side voltage harmonics. Then the sampled voltage signal at the point of common coupling (PCC) is modeled for separating the harmonic components. SMF is designed to filter out the bounded noises and to estimate harmonics. Islanding fault is judged by this method according to energy variation of the estimated harmonics. Simulation results show the effectiveness of the two methods, in which SMF method is done comparative experiments with Kalman filtering to reflect the superiority of the new method.