| With the progress and development of society,it is difficult for traditional AC power grids to meet the access requirements of renewable energy and users’ demand for higher power quality.The low-voltage system using DC power supply has become a research focus of scholars due to its greater advantages in reducing transmission loss,being compatible with distributed energy sources,and improving power supply quality.Low-voltage DC protection technology is a key technology for large-scale application of low-voltage DC systems,and its development has been restricted by the backwardness of DC system protection equipment.The most important of these is the lack of economical and reliable DC circuit breakers and mature DC arc fault detection and protection devices.For a typical low-voltage DC system,the most common and most dangerous fault is a short-circuit fault.When a short circuit fault occurs in the system,if the fault cannot be cut off in time,the system current will quickly rise to an unsustainable level of the power grid,causing system damage.Therefore,the DC circuit breaker,which is a key protection device in the system,has received widespread attention.In addition to short-circuit faults,in lowvoltage DC systems,common faults such as wire breakage and poor contact between the wire and the switch can also cause DC arc faults.For the DC arc fault,if the arc cannot be detected and cut off in time,it will cause fire and other disasters in severe cases.Therefore,in this paper,a low-voltage DC system is taken as the research object.A new hybrid DC circuit breaker based on a fully-controlled device is designed for short-circuit faults.An arc fault detection and protection method based on convolutional neural networks is proposed for DC arc faults.Specific research work and conclusions are as follows(1)Aiming at the current situation of low-voltage DC system lacking effective circuitbreaking equipment,a new hybrid DC circuit breaker based on zero-current interruption is proposed.The topological structure of the new DC circuit breaker is designed and its working principle is introduced.In Matlab / Simulink,a hybrid DC circuit breaker simulation model is established for a low-voltage DC system with a voltage level of 400 V and a short circuit simulation of 400 V / 40 A is carried out.The simulation results verify its feasibility.Then,the current transfer characteristics of the circuit breaker are simulated and analyzed,and a series of conclusions affecting its breaking fault current were obtained,which provids the basis and guidance for the subsequent construction and testing of the DC circuit breaker.Finally,a50V/1A small reference test prototype is built in the laboratory,,and relevant tests are performed for the current transfer process under short-circuit conditions.The results obtained are basically consistent with the simulation,which confirms the effectiveness and feasibility of the proposed hybrid DC circuit breaker.(2)Aiming at the problem of arc fault of low-voltage DC system,a test platform for series arc fault of low-voltage DC system is built according to the American UL1699B-2011 standard.The test data is collected using resistors,inverters and DC-DC switching power supplies as typical loads.The time-frequency and frequency-domain characteristics of the DC fault arc current and voltage of the three types of loads are analyzed.Finally,the wavelet transform method is used to extract the time-frequency domain features of the fault arc as the feature quantity of fault detection.(3)Based on the feature quantity obtained by wavelet transform,the concept of transfer learning is introduced,and a DC fault arc detection method based on transfer learning convolutional neural network is proposed.Based on the experimental data,the proposed algorithm is verified in Matlab.The results show that the detection method can detect the series fault arc of the low-voltage DC system with an accuracy of more than 97%. |
PDF Full Text Request