The invention discloses a multi-target optimization prediction control method based on trapezoid interval soft constraint. The multi-target optimization prediction control method comprises a step1, establishing of a prediction model; step2, calculation of prediction out; step 3, feedback of correction; step 4, building of a trapezoid interval outside a tolerance interval; step 5, calculation of an optimization variable (
img file+DDA0001238247080000011. Tif wi=70 he=47); step 6, building of a multi-target function; step 7, conversion of a multi-target problem into a single-target problem by using an
omega-constraint method; step 8, acquisition of an
optimal control increment by using a sequential quadratic
programming algorithm. When the controlled variables exceed the tolerance interval, the controlled variables can enter the tolerance interval quickly to guarantee product quality. The building of the multi-target function is used to effectively prevent the mutual
coupling among the various controlled variables, and the interval range of the constraint function is optimized by adopting an iterative
algorithm, and then
rapidity of operation of a
system is guaranteed. By adopting combination of a set value control and a trapezoid interval soft constraint, the
system can be operated at an ideal target value, and at the same time, the robustness and the freedom degree of the
system area guaranteed to the greatest extent.