Parameter Estimation of Black Box Arc Model based on Heuristic Optimization Algorithms

The black box arc model is suitable to describe the arcing process of hybrid circuit breaker, since it provides an effective way to integrate the arc model into an electrical circuit. However, it is more difficult to determine the parameters of the arc model according to the performance of circuit breaker than to solve the analytical equations of arc model. A library of arc models including Mayr, Cassie, Kema and other models has been built and verified based on MATLAB Simulink/Simscape. Heuristic optimization methods, such as Genetic Algorithm, Simulated Annealing and Particle Swarm Optimization, have been used to develop the algorithms to estimate the parameters of arc models. The results show the inverse modelling technique is effective to find the proper parameters which describe the changing of conductance of arc plasma. It is also found that the GA with well-selected parameters has the advantage over other methods.