Dynamic Bayesian Networks For Transmission Line Fault Diagnosis Under Robot Inspection

As transmission line networks expand fast because of growing electricity consumption and areas that transmission lines run through have complicated terrain and changing climate, higher requirements are raised to ensure safe operation of transmission lines. Once faults occur, serious outcomes will be caused. Therefore, it is of great significance to quickly find and identify faults with certain fault diagnosis method, which benefits timely repair and sound operation of transmission line in the long term.The fundamental issue for fault diagnosis is to obtain timely, reliable and adequate transmission line data. Most fault diagnosis methods at present use current and voltage parameters, because they are easy to obtain and support quick diagnosis. However, these methods can hardly pinpoint location and types of occurring faults. To overcome this shortcoming, researchers are focus on other ways, such as robot transmission line inspection, to obtain more data. Robot inspection can do a better job and provide abundant data, which, if properly processed and combined, will help improve fault diagnosis.To overcome problems mentioned above, this paper suggests implementing transmission line fault diagnosis with Dynamic Bayesian Network (DBN) theory and data obtained from robot transmission line inspection. It researches data obtain and process, modeling and simulation. First, it summarizes data types of data obtained through robot inspection and analyses how to process these data; Secondly, it determines factors that lead to faults and cause-effect relationship among observable factors and operation states of transmission line; Moreover, it constructs fault diagnosis model based on DBN and knowledge of transmission line operation; Finally, Netica, a probability inference tool, is applied to implement simulation and comparison of simulation outcome between diagnosis of DBN and static Bayesian Network shows the former has an edge.