Thermal Model Parameter Identification and Verification Using Genetic Algorithm

Abstract

In the previous chapter, two thermoelectric analogy thermal models have been developed to model thermal dynamics of oil-immersed transformers, i.e. a comprehensive thermoelectrical analogy thermal model and a simplified thermoelectrical analogy thermal model. For the problem of thermal parameter identification, a simple genetic algorithm is employed in this chapter to search global solutions of thermal model parameters using on-site measurements. Firstly, the parameter identification and verification of the comprehensive thermal model is presented. GA modelling results for the comprehensive model are compared with the historical heat run tests and the modelling results from an ANN model. For the simplified thermal model, a number of rapidly changing load scenarios are employed to verify the derived thermal parameters and finally an error analysis is given to demonstrate the practicability of the simplified thermal model.