islanding detection

This paper proposes an index based on zero-sequence components that serves as a single indicator for islanding detection. The proposed index demonstrates satisfactory performance, achieving an accuracy of 97.84% using a simple threshold adjustment model. Additionally, this index can be integrated into artificial intelligence-based methods to enhance accuracy further. The results indicate that utilizing the proposed index as a single indicator within a one-dimensional convolutional neural network (1D-CNN) model yields a competitive accuracy of 99.78%. This outcome is noteworthy when compared to more advanced artificial intelligence methods, such as long short-term memory (LSTM) neural networks, which rely on a larger set of features. The data collection for testing encompasses various islanding and non-islanding conditions, including islanding under different loading conditions and power factors, as well as varying quality factor values. Non-islanding scenarios include the activation and deactivation of large loads and capacitor banks, along with the application of various short-circuit faults at different locations and resistances. The results from all tests demonstrate the superiority of the proposed method and index.

In Recent advances in renewable energy technologies as well as the restructuring of the electricity market have increased the use of renewable energy resources. The use of these energy sources has many benefits such as increasing reliability, improving power quality, and peak shaving. However, one of the main disadvantages of these energy sources is the islanding detection. The island mode occurs when the main grid is interrupted for any reason, and the part of the grid that contains distributed generation supplies local loads independently from the main grid. Unintentional islanding causes problems such as protective interference and damage to consumer equipment. As a result, the fast detection and disruption of the distributed generation is essential to prevent equipment damaging. This paper presents a new hybrid islanding detection method for distributed generation resources to activate the real power shift method in case the rate of frequency change, in the first stage, exceeds the threshold. Then, if by changing the real part of the DG output current, the voltage amplitude changed, the islanding event would be detected.  The proposed method is implemented and simulated in the Matlab/Simulink environment.  It was also investigated under various conditions such as load switching, motor starting, and multi-DG mode. The simulation results showed the performance of the proposed method in islanding detection in these conditions.

According to wide variety of advantages of distributed generation units, especially those using clean and renewable energies, their installation are dramatically increasing in distribution networks. Among renewable energy generators, photovoltaic generation units have been paid more attention compared to other technologies because of clean energy generation, simple operation and maintenance, and availability in most areas. Despite the mentioned advantages, connecting the photovoltaic generation units to the distribution network will cause problems in the field of network protection. These include the possibility of overlapping relays and their non-coordination, bidirectional feeding network, and the occurrence of an islanding state. On the other hand, by increasing the penetration of these generators, the activation of advanced grid support functionalities such as active and reactive power regulations, low voltage ride through capability (LVRT) and seamless transition to islanded operation mode are included in current grid codes to compensate for their unpredictable energy production. Since these ancillary services can disrupt the operation of islanding detection methods, it is necessary to provide islanding detection method compatible with mentioned ancillary services. In this paper, islanding detection methods compatible with the advanced grid support functionalities such as power regulations, low voltage ride through, and seamless transition to islanded mode are reviewed.

For correct performanceof protection and control systems in microgrids, the islanding conditions should be identified as soon as possible.In this paper, an optimal method is proposed to determine the input parameters of classification method by using local information. The goal of optimization is to minimizie the time and maximize the percision of microgrid connection state detection. Also, in classification method, three states, i.e. islanding, reconnecting to the utility and other events are considered. The objective function of this problem is defined as summation of weighted time factor and detection error. The inputs of the islanding detection problem can be a large number of network parameters such as frequency, voltage, current, power, their sequence components or the rate of change. Consequently, the variables of the classification problem are the number and type of input parameters of the classification problem.The genetic algorithm is used to solve this optimal problem, and support vector machine is used for the classification. In order to assess the validity of the proposed method, different situations are considered for the operation of the grid, and in each case, different events are modeled. The results are compared with the methods of other papers, and the advantage of the suggested method is shown.

In this paper، an intelligent islanding detection method is presented for inverter based distributed generation (DG) using probabilistic neural network (PNN) and wavelet transform. The presented method is based on the change of DG reactive power reference (Qref) in inverter control interface to create a small forced transient in frequency and its derivative. Changing the Qref causes a forced transient in system frequency and also in its derivative in islanding conditions. The main idea is to use the created transient in frequency derivative in islanding conditions. The PNN is trained by features extracted from the frequency derivative data through the discrete wavelet transformation (DWT) in islanding and non-islanding conditions. The proposed method is evaluated in islanding and non-islanding conditions، using PSCAD/EMTDC and MATLAB software. Simulation results show that the proposed method has a proper operation in the islanding and non-islanding conditions.