<p>Arc faults are inclined to trigger fires in residential electrical systems, and the timely and precise detection of arc faults is highly significant for ensuring the safety of residential electrical systems. Arc models are beneficial for conducting research on arc fault detection methods in simulation circumstances, which can effectively lower research costs and enhance development efficiency. Nevertheless, the present arc models have challenges in accurately depicting the characteristics of arc faults. Hence, this paper innovatively puts forward an enhanced arc model and parameter identification approach. Firstly, taking into account the influence of current variations on the response speed of the arc during its development process, an improved Schwarz model based on a dataset is proposed. Subsequently, this paper enhances the growth optimization algorithm (GO) by incorporating the dynamic fitness-distance balance (dFDB) selection method, thereby improving the algorithm’s optimization performance. Ultimately, the advanced growth optimization algorithm (dFDB-GO) is utilized to identify the parameters of the improved Schwarz model using experimental data. Additionally, this paper conducts a comparative analysis of multiple arc models and optimization algorithms to validate the effectiveness and advancement of the proposed model and algorithm.</p>