A fault diagnosis method for power transformer using Bayesian data analysis

This paper presents a fault diagnosis method for power transformer. Fault diagnosis plays an importance role in the efforts for transformer diagnosis to shift form “preventive maintenance” to “condition based maintenance” (CBM), and consequently to reduce the maintenance cost. Ever since its birth, numerous techniques have been researched in this field, each method however, has its own advantages and disadvantages. Fault diagnosis is a challenging problem because there are numerous fault situations that can possibly happen to a electrical transformer. Temperature is one of the most importance parameters for the diagnosis of electrical transformers fault situation. Based on the appearance information of temperature, a fault monitoring system can make diagnosis intelligently, therefore it can provide a rapid scientific treatment options for the field staff. For the problem of transformer fault diagnosis based on temperature information, we use Bayesian probability density theory method to explain the equipment fault diagnosis results. The maximum membership probability principle will be adapted to judge whether the equipment is malfunctioning. The results shows that when the working temperature is between 64~72 °C, the diagnosis results without compensation is working on normal state, while the diagnosis results with compensation is general fault state. Therefore, the diagnosis results without compensation have large error with the actual situation. The proposed methodology in this paper was testified having important significance in improving the reliability and the information level of transformer operation.