Adaptive Combustion Optimization Control Based On Dimension Reduction With Applications To Tangentially Fired Boiler

Combustion optimization technology is an important means for coal-fired boilers in service to achieve energy saving and emission reduction,but this technology still has shortcomings when applied to the combustion system of tangentially fired boilers.Therefore,based on the characteristics of the tangentially fired boiler,this paper develops the support vector regression model of the combustion system with the three characteristics of "dedimensional,adaptive and dynamic",and further combines the sequence quadratic programming algorithm to solve the optimal combination of control variables,and finally develops the above algorithm into the combustion optimization control system software.The combustion optimization control system can dynamically optimize the combustion conditions by adjusting the operation mode under the premise of ensuring the safety and stability of the boiler,so as to achieve the goal of improving boiler efficiency or suppressing NOx generation.The focus of this paper is on modeling the combustion system of the tangentially fired boiler.Firstly,on the basis of the traditional support vector regression,a quick solution method based on iterative quadratic program is proposed.Then,when it is applied to the modeling of the tangentially fired boiler combustion system,taking into account the object’s characteristics of “many control variables,time-varying,and large delay”,the control variable dimensionality reduction + kernel matrix dimensionality reduction,the method of the iterative quadratic program adaptive update,delay time analysis is adopted,and a dimensionality reduction,adaptive,dynamic combustion system model is established.Next,this paper proposes a dynamic model multi-step prediction accuracy evaluation method combined with the dynamic time wraping,which is used for online performance monitoring and adaptive update of the model.Finally,the model hyperparameters are tuned,and the performance of the model is compared and tested.The results show that the algorithm proposed in this paper has significant advantages compared with the current optimal adaptive incremental update algorithm,and the proposed improvement measures in three aspects have achieved the expected results.The combustion system model has strong dynamic prediction ability and high computational efficiency.Based on the above model of tangentially fired boilers combustion system,the economic predictive control framework is used to construct the optimization objective function,and multiple constraints are set to ensure the safe operation of the system.The rolling optimization is combined with the sequence quadratic programming algorithm to solve the optimal control combination that meets the optimization target and constraints.On this basis,a complete combustion optimization control software including algorithm modules,integrated data communication and interface interaction functions is developed.According to the requirements of the project,the software is currently put into operation on the 600 MW subcritical coal-fired boiler in a power plant in an open-loop manner to provide operation guidance for the operators.And,through further transformation,the system can be upgraded to a closed-loop combustion optimization control system.