Abstract: Impact of friction on machining accuracy and dynamic performance in CNC machine tool feed servo systems is addressed through friction compensation and parameter optimization. Firstly, an electromechanical coupling model of the linear motor servo system was established to construct the fundamental framework of its control system. Secondly, to accurately estimate dynamic characteristics and position-dependent friction effects, the Lundt-Grenoble (LuGre) friction model was optimized, and an improved particle swarm optimization algorithm was employed for parameter identification. Finally, a hybrid control method combining LuGre feedforward compensation with pseudo-derivative feedback with feedforward (PDFF) was adopted to compensate for system nonlinear friction, followed by performance evaluation of the feed servo system and controller parameter optimization. Experimental results demonstrate that the actual position output overshoot is reduced by 10.7% compared with pre-optimization values, the time lag shorten by 0.05 s, and tracking error reduces by 0.06 mm.