Electrode wire feeding mechanisms (EWFMs) are vital for maintaining the precision and stability of arc welding processes, where consistent control of wire feed directly influences weld quality. Active disturbance rejection controllers (ADRC) are widely utilized for their ability to handle disturbances and uncertainties in control systems, making them essential for improving industrial automation. This research introduces an approach to enhance the performance of EWFMs by integrating a self-tuning ADRC with a radial basis function neural network (RBFNN). The RBFNN is used to precisely tune the ADRC controller’s parameters, improving its tracking and disturbance rejection capabilities. Additionally, a fuzzy algorithm-based method is employed to tune the parameters of the ADRC controller for comparison. Real-time validation on the dSPACE platform confirms that the RBFNN-based tuning outperforms the fuzzy-tuned ADRC (F-ADRC). Results demonstrate that the proposed method effectively rejects disturbances, minimizes overshoot, and ensures a faster and more precise response, ultimately enhancing the stability and quality of the arc welding process.