This paper proposes dynamic constrained version of NSGA-III to handle constraints for constrained optimization problems (COPs). The methodology first constructs a dynamic constrained multi-objective optimization problem (DCMOP) equivalent to the COP by converting the constraints into some violation objective functions and gradually shrinking the initially broadened boundary to the original one. Then a dynamic constrained version of the state-of-the-art NSGA-III is implemented to solve the DCMOP. Differential evolution (DE) is used as the evolutionary algorithm to generate offspring. Experimental results show that it is competitive to peer algorithm referred in this paper, and has better performance on global search.