Single-phase-to-ground fault caused by insulation deterioration is one of the main fault forms of medium-voltage distribution network, which seriously affects the reliability of power supply. Rapid and accurate identification of insulation deterioration types can improve the efficiency of fault inspection, and is of great significance to prevent further expansion of faults. In this paper, nine waveform features are extracted from the time domain and frequency domain of the single-phase grounding fault waveform data. MANOVA and Pearson correlation coefficients were used to analyze their effectiveness in identifying insulation deterioration types. Furthermore, based on extreme learning machine and support vector machine, the identification models of single-phase grounding fault caused by insulation deterioration are established respectively. On this basis, based on Dempster-Shafer evidence theory, the results of the two recognition models were fused and the comprehensive recognition model was established. Using the obtained field data for verification, it is proved that the integrated model has a good recognition effect.
In the operation of urban rail transit, the stray current of the subway is generated and flows into the earth, resulting in different ground potentials of each substation, which causes the DC magnetic bias of the transformer. Firstly, this paper establishes the stray current distribution model of dual-terminal power supply mode of subway, and simulates the dynamic change of stray current by adjusting the change of subway position. Then, the COMSOL software is used to build the coupling model between the subway operation line and the urban AC grid, and the ground potential distribution characteristics during the subway operation are simulated and studied. Finally, based on the neutral grounding voltage obtained by simulation and the topology of urban power grid, the variation law of DC magnetic bias single current in each substation is analyzed. The research shows that when the local railway locomotive runs in the middle of the line, the leakage of the stray current of the subway is the most serious, which will produce greater induction potential, and the influence on the substations in different locations is intensified, which makes the magnetic bias current of each substation increase.
SF6 has strong greenhouse effect and many countries have enacted laws restricting the use of the gas. There have been several feasible alternatives to SF6. As an environmental friendly gas with excellent insulation characteristics, CF3I has the hope of replacing SF6 in medium and low voltage electrical equipment. In this paper, the influence of pressure and gap distance on the insulation strength of CF3I-N2 mixed gas was studied, and the results show that the insulation strength of CF3I-N2 mixture increases with the increase of CF3I mixture ratio. In slightly inhomogeneous electric field, the breakdown voltage of CF3I-N2 increases linearly with the increase of discharge gap, and the increase rate is proportional to the pressure. The breakdown voltage of CF3I-N2 gas mixture shows a trend of saturation with the gap in the extremely non-uniform electric field.
Because a large number of distributed energy sources are connected to the grid, the system becomes a high-order LC network. When a certain disturbance frequency is close to the system resonance point, the system resonance phenomenon will occur. Therefore, it is necessary to study the resonance detection method of clustered grid-connected photovoltaic power generation. In this paper, the resonance principle of the clustered grid-connected photovoltaic system is firstly analyzed based on the resonance model of the clustered grid-connected photovoltaic system. Then an LWT-based resonance detection technique is proposed. Through this algorithm, the start-stop time and frequency range of the clustered grid-connected photovoltaic system can be obtained. Finally, the simulation results demonstrate the effectiveness of the LWT-based resonance detection method for clustered grid-connected photovoltaic systems.
The alarm management system often has the problem of "flood" alarm for a period of time. A large number of flooding alarm will cause a crisis of trust between human and machine and reduce the availability of the system. The article first investigates the problem of unbalanced alarm distribution in the actual application of the alarm system. In order to adjust the probability density of different alarms, it is necessary to introduce the concept of alarm chatter rate that characterizes the arrival density of alarms. Then, the description principles and methods of the characteristics of alarm redundancy chatter are given. An alarm chatter rate characterization algorithm combining attenuation factor and natural alarm arrival rate is proposed. Finally, the concrete realization of the compression strategy based on this feature processing is given. Comparing and analyzing in actual test, this method can effectively reduce the number of redundant alarms and improve the efficiency of human-machine information interaction expression.
