Double-layer structure model prediction control, composed of steady-state target calculation and dynamic optimization, is restricted in the field of high real-time control and poor computational controller. A new method of steadystate target calculation, called off-line optimization and online table lookup, is proposed. Also the calculation method is given in detail. First, the steady-state optimization database is estimated through enumeration and off-line optimization. Then the steady-state optimization is addressed based on online lookup table. For the case that online table lookup is not feasible, the steady-state optimization is calculated approximately by seeking the multiple points which are nearest to the lookup table point in the lookup table, and selecting the distances as the weighting coefficient. Finally, The capabilities of the method in this paper are investigated through computational complexity analysis and computer simulation.