The set-point optimization is an effective way of optimal control for complex industrial process, which can be realized by adding a supervisory optimal level without modifying the regulatory level. While model predictive control has been widely used in industrial process, incorporating of constraints is a major problem. Therefore, neuro-fuzzy networks(NFNs) with nonlinear constraint predictive control is presented, and the feasible solution is derived. Power and velocity control of gas turbine in combined cycling boiler is presented to illustrate the implementation and the performance of the proposed method. Simulation results show the superiority of the method.