The disturbance estimate is the central idea of active disturbance rejection control (ADRC), where the disturbance is estimated via extended state observer (ESO). However, the conventional ESO requires the disturbance to have slow variation and in order to counteract such disturbance, the ESO must use high gain or discontinuous function. In this paper, we demonstrate for the first time, through a one-dimensional anti-stable wave system, the online disturbance estimation by designing an infinite-dimensional disturbance estimator. We go back to the starting point of ADRC that the exactly observable output allows identification of the disturbance which can thereupon be estimated. This realizes the estimation/cancellation strategy of ADRC in a very different way without resorting conventional ESO. We do not regard the disturbance as an "extra-state" variable in the ESO, and hence avoid the two limitations of the ESO. Since the disturbance is estimated, it can be compensated in the feedback loop. It is shown that the disturbance estimator-based feedback control can successfully stabilize the PDE system and at the same time guarantees that all subsystems involved are uniformly bounded. A numerical simulation is presented to illustrate the effectiveness of the the proposed scheme.