The linear induction machine (LIM) suffers heavily from low efficiency due to its large air-gap length, the influence of end-effects, and the partial load conditions. Several loss minimization control (LMC) techniques have been developed for LIM to adjust the flux to an optimal level to reduce the losses in the steady state. However, these methods are not applicable during dynamic process, since it requires much more complex modeling that leads to heavy computational burden due to the change of thrust in dynamic operation. In this paper, to solve the aforementioned problem, a novel LIM dynamic loss model is derived based on the full analysis of copper and core losses. A simple solution for online estimation of the optimal flux in dynamic operation and a practical dynamic LMC scheme are proposed to minimize the losses. Simulations and experiments are fully performed on a 3 kW LIM prototype, and demonstrate the effectiveness of the proposed method in loss minimization and energy saving during the dynamic process.