Abstract: To improve the economic performance and control performance of multi variable control system in the range of safe and controllable, alarm threshold management and process control unifled into a single framework under the effects of uncertainty dynamic, the idea that the stochastic optimization algorithm helps to identify the optimal working point was proposed. To solve the problem of the different prioritization and infeasibility of the multivariable alarm threhold control, a mixed logic method was introduced to describe the priority of the alarm system to handle the infeasibility and prioritization issues in the predictive control. In order to determine the feasibility by adjusting soft constraints, multivariate alarm system coordination problem was converted to the logical constraints problem. Therefore, the problem of adjustment of alarm priority and the coordination of the objectives were solved systemically. At the same time, the economic performance of the system was improved. A simulation example was applied to illustrate the validity of the algorithm.