AGING ASSESSMENT OF POWER TRANSFORMER INSULATION OIL USING IMPROVED COLLIDING BODIES OPTIMIZATION-BASED HYBRID CLASSIFIER

The degradations present in the power system networks have lead to the competition in the distinct beneficial operations for optimizing their cost of operations and enhancing the stability of their electrical communications. The comprehensive and specific assessment of state present in the electrical tools supports in handling the group of the appropriate safeguarding plan. Therefore, to achieve a proficient and comprehensive transformer condition assessment, diagnostic and insulation condition monitoring methods are required for the economic and reliable transformers. The major idea is to propose a novel forecasting model for the aging assessment that is associated with the power transformer insulation oil. The data is collected by means of 20 power transformers of 16 to 20 MVA that is functioned at several substations in Punjab, India. It involves several parameters that are related to the transformer like flash point, interfacial tension, tan delta, resistivity, moisture, and Breakdown Voltage (BDV). These collected data are applied as input to forecast the insulation oil's age. The developed model makes use of a hybrid classifier model by integrating the Deep Belief Network (DBN) and Neural Network (NN). A major enhancement is made by replacing the training algorithm of both the DBN and NN with the adaptive form of Colliding Bodies Optimization (CBO) algorithm called Self Adaptive-based Colliding Bodies Optimization (SA-CBO) for reducing the error difference among the actual outcome and the forecasted outcomes. In the final step, the relative analysis of distinct forecasting models with various error measures confirms the effectiveness of the developed model.