Research On Nonlinear Predictive Control Of Ultra-supercritical Unit Load Control And Simulation

As the boiler is moving in the direction of large capacity and high parameter,it is urgent to study the modeling and control of large units.However,due to the ultra-supercritical unit thermal process with nonlinear,strong coupling,large delay and other characteristics,the conventional linear control strategies are difficult to meet its control requirements.Based on the summary of the current situation of modeling and control strategy of super-supercritical unit load control system,this paper focuses on the improvement of the basic particle swarm optimization algorithm,the ultra-supercritical unit coordinated control system modeling and the nonlinear predictive control based on neural network and modified hybrid particle swarm optimization algorithm.Then,the nonlinear predictive control method is applied to load control,and some achievements are obtained.It is easy for basic particle swarm optimization algorithm to fall into the local optimal solution,to address this issue,a new type of optimization algorithm is proposed,which introducing genetic algorithm and simulated annealing algorithm into the basic particle swarm optimization.The modified Hybrid Particle Swarm Optimization can maintain diversity of particle and improve the global evolution accuracy.By validating the test functions and solving simple thermal process,suggesting that compared with other intelligent algorithm,modified hybrid particle swarm optimization has better searching performance and recognition ability.Through some simplifications of the ultra-supercritical unit pulverized-coal system,boiler steam-water system and turbine system,a 3×3 load control system model was established using the principle of mass balance,energy balance and momentum balance.And then using the data of a power plant unit sis system,the unknown parameters in the model are obtained.Among them,the dynamic parameters are identified by the modified hybrid particle swarm optimization algorithm.Finally,the open-loop dynamic characteristic test further validates the model's accuracy.Aiming at the non-linearity of ultra-supercritical units,this paper proposes a nonlinear predictive control method based on neural network combined with improved hybrid particle swarm optimization algorithm.With RBF neural network model used as the prediction model,and the modified hybrid particle swarm optimization algorithm used to optimize the control law,the optimal control is given.Finally,the effectiveness of the method is verified through the tests of stable operating point load disturbance and a wide range of variable conditions.