This paper proposed an improved filled function method for solving non-smooth constrained global optimization problems, which often arise in engineering and scientific applications but remain challenging due to their complex landscapes and feasibility requirements. The proposed method constructs a novel filled function that integrates a distance-based term to transform local minima into local maxima, a reciprocal term to facilitate escape from suboptimal solutions, and a penalty term to handle constraints effectively. This formulation ensures that the search process navigates both non-smoothness and feasibility constraints systematically. Moreover, the method adaptively adjusts key parameters to balance exploration and exploitation, thereby improving the likelihood of locating the global optimum. Unlike existing approaches, our method does not require smoothness assumptions or restrictive conditions on the objective or constraint functions, making it applicable to a broader class of problems. Experimental results demonstrate its robustness and efficiency in comparison to state-of-the-art techniques. The proposed method offers a practical and theoretically sound framework for addressing complex optimization challenges, with potential applications in fields such as operations research, machine learning, and industrial design. Its ability to handle non-smoothness and constraints simultaneously represents a significant advancement in global optimization methodologies.