High voltage SF6 circuit breakers are critical for the safety and stability of power systems. The key to research their interruption performance is focused on arc modeling and experimental data processing. Traditional black-box arc models exhibit shortcomings in parameter searching. To address it, this paper proposes an arc model parameter search method based on T100a test from a 363 kV high-voltage sulfur hexafluoride (SF6) circuit breaker, leveraging the advantages of the parameter-reduced TP KEMA model in arc modeling. An enhanced particle swarm optimization(PSO) algorithm is employed: first using variable-step search to approach optimal points, then applying the PSO algorithm for parameter fitting. The fourth-order classical Runge-Kutta method is applied to construct the difference form of the TP KEMA arc model, thereby completing the arc model construction. Simulation verification shows an average arc voltage error of 65.34 V. Among five test datasets, only one simulation interruption result differed from the experimental result, achieving an 80% accuracy rate. The enhanced PSO algorithm reduced the fitting error by 36.2%. These results demonstrate that the proposed enhanced particle swarm algorithm effectively enhances parameter search efficiency and accuracy.