Abstract: To deal with the difficulty of establishing mathematical models for non-linear system, a practical data-based second-order linear extended state observer (LESO) filtering method is proposed. By ESO estimating the uncertain disturbance in real time, a simple linear ESO filter with less tunable parameters was designed in the absence of dynamic model. Then the parameters tuning way of the filter and the method of discrete-time realization were given respectively. Last, the designed LESO filter was applied to the tracking of square signal and sine signal, to filtering the signal with uniformly distributed white noise and the width data of hot rolling strip in real time. Simulation results show that the proposed filter is effective and practical.