This paper proposes a nonlinearly activated neural dynamics to solve time-varying nonlinear equations in real time. Different from most existing neural dynamics for solving time-varying nonlinear equations, the proposed neural dynamics can converge in finite time. In addition, the upper bound of convergence time is estimated analytically in theory. Simulations are performed to evaluate the performance of the proposed neural dynamics, which substantiate the effectiveness and superiority of the finite-time convergent neural dynamics for solving time-varying nonlinear equations in real time, as compared with the conventional gradient-based neural dynamics and the recently proposed Zhang neural dynamics.