Research On Intelligent Recognition System Of Converter Station Arrester Dashboard Based On Image Collection

With the extensive use of traditional fossil energy,the problem of environmental pollution is becoming increasingly serious,the development of clean new energy imperative,distributed energy(distributed energy resources,DER)as the grid effective supplementary way,because of its clean installation,operation and maintenance convenience advantages have gradually been accepted and achieved rapid development.When the DER is connected to the distribution network,the radiation type network into a power supply and users across many interconnected network,a fundamental change in the distribution network of relay protection system for some of the affected to varying degrees,converter arrester for overvoltage as an important equipment in power system,its behavior is also affected by the distributed power therefore,in order to ensure the safe operation of the power grid in this model,it is important to monitor the action times and leakage current of arrester.In this thesis,the new type of lightning arrester State Patrol technology is studied.Firstly,study the correction algorithm of the instrument distortion of lightning arrester.Considering the particularity of the installation location of the lightning arrester and the limitation of the camera's orientation,the image of the collected instrument will be distorted,and the image rectification technology based on perspective transformation is studied.Then,a recognition algorithm for the number of lightning arresters.First,the vertical projection method is used to separate the individual characters,and then the improved BP neural network is used to recognize the number of actions.When BP neural network weights are corrected,momentum items are added,and adaptive adjustment learning rate is used,which can accelerate convergence speed,improve recognition speed and accuracy.Finally,the identification algorithm of leakage current of lightning arrester is studied.First,the pointer is extracted by subtraction,and the leakage current is quickly and accurately identified by the improved Hough transform.The experimental results show that the improved Hough transform is better than the traditional Hough transform in the recognition time.The simulation results show that the proposed method can effectively overcome the limitations of the traditional automatic recognition algorithm and improve the accuracy of reading recognition.The improved BP network recognition character is not only influenced by the size of the font,strong anti-interference,but also fast in recognition.Identification of improved Hough transform dual threshold right out of the station for lightning arrester leakage current based on the accuracy at the same time,greatly reduces the time for reading,to achieve the real-time requirements of the system,has significant social benefits and potential economic benefits.