Research On Fault Diagnosis System Of Substation Based On Machine Learning Algorithms

Transmission lines in the northwest region of China are generally characterized by large scale,long distance,complex structure,and harsh environment.It’s safe and stable operation is closely related to the reliability of users in the eastern load center cities.The 750kV substation is an extremely important part of the transmission line,and the equipment failures frequently occur in the substation,because of the complicated working conditions and high requirements for insulation and mechanical properties of the equipment.Once the equipment in the substation fails,the reliability of the power supply will decrease,which will directly affect the production and life of the economically developed and densely populated areas in the east,and causing huge economic losses.However,the traditional manual inspection methods have poor working conditions and high potential safety hazards.What’s more,the inspection methods cannot fully grasp the operating status of the equipment in the substation,and cannot accurately identify the faults.Therefore,it is necessary to establish a monitoring system for the equipment in the substation,and the fault can be identified based on machine learning algorithms.The establishment of the substation fault diagnosis system helps the staff to grasp the operation of the equipment in the substation in time and realize the identification of faults,which is of great significance to ensure the safe and stable operation of the 750kV power supply systemFirstly,the systematic design of the fault diagnosis system is carried on,mainly including Java technology analysis,feasibility analysis,and demand analysis.Based on the above analysis,the framework of the fault diagnosis system is constructed.In addition,the equipment in the substation and its common fault types are analyzed,and the spring operating mechanism of the 330kV circuit breaker is selected as the research objectSecondly,the states of the spring mechanism are simulated through the experiments,including the degradation of the opening spring,the degradation of the closing spring,and the oil leakage of the oil buffer,as well as the joint degradation of the opening and closing spring and the degradation of the closing spring accompanied by oil leakage of the oil buffer.During the experiments,the contact displacement is collected.Then,the above states are identified through support vector machine,random forest,and deep neural network algorithms,and the performance of different algorithms in fault identification is compared through the values of accuracy,recall rates,and F-value.The results show that the random forest algorithm has the highest recognition accuracy in the shortest time,which is more suitable for classifying the spring operating mechanismFinally,based on the design of the fault diagnosis system and the machine learning algorithm,the key functions of the substation fault diagnosis system are realized.It mainly includes user registration management,menu management,equipment management,electrical parameter display,and system alarms.Using the above system,users can grasp the operating status of the equipment in the substation in real time.The system diagnoses the fault through a machine learning algorithm,and it can send the fault information to the relevant staff in the form of SMS.