| With the rapid development of the economy and accelerating urbanization construction, the power cables are used increasing gradually whose layout is also more complex. So the faults frequency of the cable is gradually increasing, and so dose troubleshooting. The spark fault which due to the cable bolted loose is the most prone failure in them, it will be developed to the arc fault and causing serious economic losses, even take a threat to personal safety if the spark isn't controlled in time. Therefore, it is greatly significant for us to detect and take measures rapidly and accurately for the spark fault.In this paper, the research of spark fault due to the cable bolted loose has been launched. This paper studies the characteristics of spark fault, and because the spark is the is the previous stage of the arc so we can see the spark as a momentary small arc, therefore, this paper alse introducs the characteristics of arc. At the same time, in order to study the detection method of spark fault, this paper developed a cable connector loose spark generating device, we carried out a large number of experiments at different loop current conditions and different power factor under the different degrees loose conditions respectively. What's more, collecting a large amount of experimental datas using data acquisition card and create a relevant database which made preparations for the experimental analysis next step.In order to find the characteristics of spark fault and normal state under different experimental conditions and the difference between them, this paper studied contact current based on wavelet energy entropy combining probabilistic neural network to detect Loose spark fault. The results show that savelet packet energy entropy algorithm can effectively extract loose electrical connection fault feature and make them as the probabilistic neural network (PNN) input, which can effectively identify whether there are spark faults due to cable loose or not in the distribution lines. |
PDF Full Text Request