Research On Distributed Optimal Control Of Direct Air-Cooled Unit Based On Data-driven

Direct air-cooled units are widely used in thermal power generation units in the "three northern" regions of China,where coal is rich and water is scarce.Driving forced convection of air by axial flow fans and exchanging heat with turbine exhaust are significant in water saving effects.Currently,direct air-cooled system is exposed to outside,and the aerodynamic characteristics of the air-cooled unit array are constantly changing,resulting in a reduction in the working efficiency of the air-cooled unit,base on the coupling effects of natural factors and fans cluster effect.Meanwhile,the high power consumption of fans clusters leads to a decrease in the net power generation rate of thermal power units,ultimately affecting the economic operation of thermal power units.Therefore,studying the aerodynamic characteristics and operation optimization of direct air-cooled units can improve the safety and economic performance of the units.Taking 1×8 air-cooled array of a certain 660MW unit as research object,based on computational fluid dynamics driven by numerical and model,a transient mathematical model of air-cooled arrays under natural wind transient disturbances and random fans speed is solved,and operating data under different operating conditions are calculated.The unit operation optimization is further studied.The main research contents include:Firstly,a numerical simulation model of the direct air-cooled system is established,and the corresponding control equations,turbulence equations,and discrete solution methods are introduced.Finally,the mass flow rates of each air-cooled units under typical operating conditions are simulated and calculated,and compared with the actual measured data of the direct air-cooled units to test the accuracy of the mathematical model of aircooled unit array,providing a model basis for subsequent research.Secondly,based on the numerical model of the air-cooled island,the flow field distribution of the air-cooled array under natural wind step and fans speed step conditions is analyzed,to explore the impact of fan speeds and natural wind on the fan inlet flow.The research shows that the increase in the rotational speed of the upwind fan worsens the cooling performance of the air-cooled island,while the increase in the rotational speed of the downstream axial flow fan is beneficial to the cooling performance of the air-cooled island;Natural wind can worsen the cooling performance of air-cooled islands,and different arrangement positions of air-cooled units lead to inherent interference resistance differences.Based on the above analysis of dynamic characteristics of fans cluster,a distribution adjustment scheme for fans cluster is proposed using fuzzy C-means to analyze the characteristics of dynamic characteristics data.Furthermore,an adaptive generalized predictive control strategy based on multivariable Hammerstein model is designed for summer and winter conditions,where the natural wind direction is relatively fixed and the wind speed is relatively slow.Firstly,the extended random gradient algorithm based on hierarchical identification principle is used to identify the distributed Hammerstein system model;Secondly,based on the predictive control principle,considering both the nonlinear characteristics and constraints of the system,.the GPC strategy is directly implemented for the entire nonlinear system.When considering model mismatch,a parameter self-tuning strategy is formed by combining the least square algorithm with forgetting factor to identify the system model online and in real time.Adaptive GPC control strategy has the advantages of small output deviation and stable output.Finally,a nonparametric dynamic response model for fans cluster is constructed using Long term and short term memory networks for transient and random natural wind conditions in spring and autumn.By improving the non dominated sorting genetic algorithm with elite strategy a multi-objective distributed optimal control strategy is proposed that comprehensively considers performance indicators such as air-cooled units inlet air flow,fan energy consumption,and mechanical and electrical losses,and calculates the optimal rotational speed command for each region of fans through fuzzy membership function method.The optimal distributed control strategy is determined through simulation research for two conditions of no natural wind and time-varying continuous natural wind.Compared with the centralized speed regulation strategy of air-cooled array,the multiobjective distribution optimal control significantly improves the operating economy of the cold-end of the unit,reduces the mechanical and electrical losses in the regulation process,and significantly improves the control quality of the air-cooled unit.The research results can provide theoretical basis and engineering reference for the optimal energy-saving and safe operation of direct air cooling units.