การประยุกต์ฟัซซีลอจิกสำหรับประเมินสภาพหม้อแปลงไฟฟ้าระบบจำหน่าย 33 kV ของการไฟฟ้าส่วนภูมิภาค

Abstract

The objective of this research was to condition assess 33 kV distribution transformers by Health Index using fuzzy logic analytical decision-making methods. The data for the assessment of transformer condition consisted of 5 factors: oil quality; aging; visual inspection; winding insulation; and load from the results of maintenance each year. The 180 samples are divided into 150 samples for training and 30 samples for testing, which are divided into 4 groups: group 1 is a transformer in good condition; group 2 is a transformer in acceptable condition; group 3 is a transformer in poor condition; and group 4 is a transformer in very poor condition. The design uses the triangular, the square, and the trapezoid membership function for the each factors and the transformer condition level. The weighting of the 5 factors was determined using an artificial neural network (ANN) method based on the fuzzy logic decision output values for each of 150 samples (Train) to calculate the health index and the transformer condition level. The results of testing transformer condition assessment system from the data of 30 test transformers were sent into the designed assessment system. Then compare the condition level values assessed by the system with the actual transformer conditions. It was found that out of 30 samples, the condition could be assessed according to the actual transformer condition of 25 samples, and there were 5 samples of errors, representing an accuracy of 83.33%. The maintenance planning in sequence before-after can be considered at the condition group level from very poor, poor, acceptable, and good, respectively.