This paper proposes a new stiction compensation method that combines adaptive inverse model techniques and intelligent control theories to compensate for a process containing a sticky valve. This new method reduces the cycling caused by the nonlinear dynamic associated with stiction, while avoiding aggressive movements of the valve stem. The parameters of the compensator depend on both the type and degree of stiction, and their estimates are optimized using differential evolution. Both simulation and experimental results validate the approach. The compensation method demonstrates excellent performance in overcoming the effect of stiction and in reducing the oscillations with smooth valve stem movements.