The dynamic economic dispatch problem of the power system has the characteristics of high dimensionality, strong non-linearities, and multiple constraints. In this study, a global and locally enhanced water cycle algorithm is proposed to solve the problem. Firstly, distance control parameters and adaptive t-distribution non-linear adjustment strategies are used to balance the global exploration and local exploitation capabilities of the algorithm. Secondly, a quadratic interpolation operator is introduced in the later stages of the algorithm's search to further improve the speed and accuracy of the algorithm's search. Finally, a constraint repair technique is used to deal with the constraints of dynamic economic dispatch. To verify the effectiveness of the improved water cycle algorithm, a 10-generator power system including a wind farm is used as a test case for simulation experiments and compared with current advanced optimization algorithms. The results show that the improved water cycle algorithm is effective and superior in solving the dynamic economic dispatch problem of power systems.