Incipient Fault Determination in Oil-Insulated Power Equipment via Neural Network-Social Group Optimization

Oil-insulated power equipment such as power transformers are one of the most imperative facilities in power systems. However, they are constantly subjected to electrical and thermal stresses, which accelerates their ageing process and heightens the risk of malfunction during operation due to incipient faults. Therefore, determination of incipient faults in power equipment is of utmost priority, where faults must be detected and diagnosed accurately in the early stages. In this work, determination of the incipient faults within oil-insulated power equipment based on dissolved gas analysis (DGA) data is proposed using artificial neural network (ANN)-social group optimization (SGO) technique. The method was compared with combination with other algorithms, which include particle swarm optimization (PSO) and artificial bee colony (ABC) algorithms. The results in this work demonstrate an improvement in the classification accuracy for the optimized ANN compared to the non-optimized ANN. Comparison among the optimized methods shows that ANN-SGO yields higher classification accuracy compared to ANN-PSO and ANN-ABC. The results obtained indicate that the proposed technique could benefit the power industries in determination of the fault type within oil-insulated power equipment automatically.