Prediction Method of Dissolved Gas in Transformer Oil Based on Firefly Algorithm - Random Forest

Dissolved gas in oil is an important parameter to reflect the operation state of the transformer. By analyzing the volume fraction of different fault characteristic gases, it can effectively judge the fault condition and fault type of transformer, while predicting the content of dissolved gas in transformer oil can make timely warning before further deterioration of the fault to avoid the occurrence of insulation breakdown. Therefore, a prediction model of dissolved gas concentration in transformer oil based on the firefly optimized support vector machine is proposed. In order to overcome the difficulty in parameter selection of traditional random forest model, the firefly algorithm is used to adjust the parameters in RF. The test results show that FA can effectively improve the prediction accuracy of RF, and the FA-RF model has higher prediction accuracy than existing prediction methods, which can better predict the change of gas volume fraction in oil and prevent serious faults.

I. Introduction

Transformer plays an important role in the huge power system, and their operation status has a direct impact on the security of the power grid ^[1]. There is a deterioration process for the latent fault in the transformer. The operation state of the transformer can be predicted in advance, the fault danger signal can be obtained before the latent fault deteriorates to a serious fault, and the transformer can be repaired in a timely manner, which can prevent the occurrence of major accidents and maintain the safe operation of the power grid. When faults such as discharge and overheating occur in the transformer, the insulating oil will decompose under the action of faults to produce a large number of fault-characteristic gases, which will be dissolved in the oil. The analysis of dissolved gases in the oil can effectively judge the operation state of the transformer ^{[2], [3]}. Therefore, predicting the development trend of the volume of dissolved characteristic gases in the oil has important guiding significance for predicting the development state of transformer faults.