Distributed Supervisory Predictive Control Of Main Steam Temperature For Ultra-supercritical Unit

In ultra-supercritical once through boiler,the main steam temperature control extended to the intermediate point temperature control system,which formed a large-scale complex systems composed of series of interconnected subsystems,including intermediate point temperature control system and multi-water spray desuperheating control system.However,most of the traditional control scheme is devoted to solving the control problem of a certain subsystem,and the coordination of each subsystem directly affects the stability and safety of the main steam temperature control,which is the key to enhance the control performance of the global system.In this paper,a distributed supervisory and predictive control strategy based on T-S fuzzy model is proposed.On the basis of the original control layer structure,the supervisory layer is added to transform the main steam temperature control problem into on-line optimization problem.Cooperation between systems is achieved by exchanging information between each control system and its neighbors via supervisory layer and by optimizing the local problem with the performance index considering the related systems.Ultra-supercritical units have strong non-linear characteristics,which makes distributed supervisory predictive control a complex non-convex optimization problem.However,the dynamic characteristics of large-scale ultra-supercritical units still follow a certain law,that is,the dynamic characteristics depend on the load variation.Based on the real-time data,the T-S fuzzy model of the ultra-supercritical generating set is established.Based on this model,nonlinear distributed supervised predictive control still adopts quadratic programming constrained optimization path,so as to achieve the optimization effect of the whole system.