【作者】 张晓佳；

【导师】 周建华； 郭建伟；

【作者基本信息】 东南大学 ， 电气工程， 2006， 硕士

【摘要】 随着目前变电站无人值班和状态检修工作的开展,远方实时监测电气设备状况非常重要,安装容性设备在线监测装置应该是一个很好的解决手段。但是由于现场的工作环境和装置本身的一些原因,其最能反映设备绝缘状态的介质损耗因素的数据波动较大,目前所投入运行的装置都存在此问题。因此通过何种方法解决,使介质损耗因素值能不受或少受外部原因的干扰,从而较准确的实时反映设备绝缘状态就显得尤为重要。在分析研究国内外已有的优化容性设备在线监测的硬件和软件的基础上,提出了实用及改进方法。首先,从应用较广的容性设备在线监测的机理角度,分析了谐波分析法的算法原理,总结了其引起数据误差的原因,提出了可能的解决方法。然后,给出了人工神经网络的一般描述,以及在Matlab中的实现方法,提出了人工神经网络在容性设备在线监测中应用的具体实现,即利用人工神经网络来建立模型预测容性设备介质损耗因素,并利用Matlab工具进行仿真。最后,结合Matlab工具和实际数据对所提出的诊断分析模型进行了分析验证。更多还原

【Abstract】 It becomes very important to monitor the status of power equipments on-line and remotely due to the development of unmanned substation and works of condition based maintenance. Installing on-line monitoring devices for capacity equipments will be a good solution. However, because of the substation’s environment and the device’s internal states, the data of the dielectric lose factor which most reflects the equipment insulation has big fluctuation. Currently this problem exists in all running devices. Therefore, it’s very important to find out how to avoid or reduce the external interfering to the data of dielectric lose factor, then help us to get the accurate status of equipment insulation.To bring up some useful and improved methods on the base of making researches on improving the on-line monitoring data of capacity equipments both on hardware and software inside and outside the country.First, focus on the theories which are popular used by the on-line monitoring devices for capacity equipments, analyze the harmonic analysis theory, conclude the reason causing error and give some possible solutions.Then, give the description of the Artificial Neural Networks(ANN) and the methods to achieve with Matlab tools, put forward the material realization with applications of ANN in on-line monitoring for capacity equipments -create model which could forecast the dielectric lose factor of capacity equipments by using ANN. Meantime, using Matlab tools to make simulation.Finally, analyze and validate the diagnostic and analytic model by existing historic data and Matlab tools.更多还原