A Centralized Islanding Detection Method For Grid-connected Inverter-based New Energy Power Stations

Islanding phenomenon refers to the isolated operation mode of part of the power network that contains both loads and power sources after its disconnection from the main network. Unexpected islanding operation may result in unsynchronized integration and failure in reclosure, which will harm both the clients and power devices. Also, the islanding operation of generating devices after its disconnection from main network may seriously damage the health of clients and repairing faculties. Thus power systems must be equipped with effective islanding detection function. Existing methods for islanding detection is mainly for distributed systems with one or small number of inverters.But in large-scale centralized new energy systems such as photovoltaic power stations and wind farms, such traditional detection method based on inverter disturbance for single-inverter faces new challenges for its effectiveness in multi-inverter scenarios. Research on new methods for new energy power station islanding detection becomes critical.The thesis started with analysis and simulation to evaluate the effectiveness of traditional detection method under different scenarios. The electric parameter variation pattern of inverter is analyzed,and mechanism of reactive power disturbance’s influence on characteristic electric parameters is concluded. Based on this and taking three-phase reactive power disturbance method as study object, explanations and validations are presented for the mechanism of loss of effectiveness of the disturbance-based active method.To build the base of impedance model for centralized islanding detection method, inverter-based power supply’s impedance model is established, according to analysis on channels of external disturbance signals. On the basis of inverter-based power supply’s structure and patterns of bottom switches, analysis of inverter-supply dynamic responses to external disturbance signals is performed, revealing characteristics of 0 and Xc dynamic high-frequency state switching. High-frequency signals are used to convert the time-varying impedance into practical constant state impedance. This model generally applies to power generators that is connected to the network through full power voltage source inverters, without constraints from different power source operation modes and controlling strategies, it includes the time-varying and nonlinear characteristics of power electronic devices, which results in higher theoretical accuracy. These characteristics guarantees the general applicability and accuracy of centralized detection method based on this model.According to investigations on effectiveness and on the basis of impedance model, an islanding detection method for multi-inverter system is proposed. It applies an external centralized disturbance algorithm, avoiding mutual interferences of multi-supply disturbance signals. It employs external centralized disturbance at PCC, using amplitude-and-pulse-width-adjustable current signal with continuous frequency as the disturbance signal. Moreover, by means of intermittent Zero-Crossing-Point (ZCP) injection control proposed in the paper, the method reduces disturbances to power system caused by active detection method. Characteristic quantity distinction before and after islanding is amplified by using wideband spike pulse injection and high-frequency reactance calculation, improving the detection sensitivity. This method avoids loss of effectiveness in a multi-inverter system from inter-disturbance interference, and therefore, can be applied to both single-inverter and multi-inverter grid-connected system with different types of power supplies, and both islanding detection and favorable power quality can be guaranteed.