Experimental Study On Turbulence Control Based On Artificial Intelligence

With the rapid development of China’s economy,energy demand and environmental pollution are becoming more and more important.Efficient and clean use of energy has become an inevitable choice,the rapid mixing of fuel and air directly affects the combustion efficiency,pollutant emission and other indicators.On the other hand,under the combined pressure of huge fuel consumption and serious environmental problems,it is particularly urgent to develop more effective drag reduction technology for automobile.The simplified Ahmed body,which has been widely studied,provides a reference for studying the flow around real vehicles with high three-dimensional and complexity.The research of drag reduction technology can break the international monopoly,and also help to improve China’s technological innovation ability and comprehensive national strength.These applications are closely related to turbulence control.As a result,studying turbulence control and understanding the physics behind the turbulence control is crucial for key technological breakthroughs in strategic emerging industries such as energy conservation,environmental protection and high-end equipment.In addition,it has a strong theoretical guiding significance for the related turbulence control problems.Turbulence control is an interdisciplinary discipline arising in recent years,including fluid mechanics and control theory.According to classical turbulence control,it can be roughly divided into passive control and active control.But the traditional control method has some disadvantages.This also leads to a new research direction in the field of turbulence control,which is to break through the current limitations of turbulence control based on artificial intelligence.At present,it is urgent to develop a new type of turbulence control system based on artificial intelligence related algorithms for the controlled objects in order to realize automatic adjustment of parameters and improve the control effect.In this paper,classical turbulent jet flow and three-dimensional blunt body flow are selected as the research objects of turbulence control.Similarly,jet mixing enhancement and Ahmed model drag reduction are also hot topics of turbulence control in recent years.In summary,the research purpose of this paper is to develop control systems based on intelligent algorithms to enhance jet decay rate and reduce drag of the Ahmed model,respectively.In this paper,a multi-input-single output closed-loop turbulence control system is designed for a turbulent round jet mixing control.The system consists of four parts:jet facility,closed loop controller based on extended Kalman filter(EKF)extreme-seeking algorithm,hot wire probe and minijet actuator.The system also adopts the velocity attenuation rate 5D(D is jet exit diameter)downstream of the central axis jet outlet as the feedback signal,and the control signals are the excitation frequency,duty cycle and mass flow rate of the minijet.The experiments were carried out in the Reynolds number range of 5700 to 13300.When the Reynolds number is 8000,the system will be used for rapid optimization of the controlled jet.The optimal control parameters under the Reynolds number are obtained,that is,excitation frequency,duty cycle and flow rate.Then,the adaptive of the system is verified by changing the initial parameters.Finally,the robustness of the system is verified in the range of 5700 to 13300.Based on the research of single minijet,the control parameters of microjet are extended,and six independently regulated minijets are used to control the mixing of a turbulent round jet.In this way,the controllable parameters of microjet are frequency,duty cycle,mass flow rate,phase and the number of minijets working at the same time.Traditional method,such as extremum-seeking algorithm,is difficult to solve such a complex parameter optimization problem.Therefore,an artificial intelligence turbulent jet control system based on linear genetic programming is developed in this paper.The system consists of jet facility,sensor system,execution system and thinking/innovation system.In the process of optimization,the optimal jet decay rate of the system is explored.Then,all the individuals obtained in the optimization process was analyzed by the proximity map.Finally,the flow physics of the optimal control mode is measured and analyzed in detail.The deep understanding of turbulent jet mixing control can further provide guidance for another classical wake control.Subsequently,drag reduction control was carried out on the wake of the square-back Ahmed model.Therefore,an EGM algorithm based on Latin hypercube sampling(LHS)and simplex optimization is used to construct a drag reduction control system for turbulent wake flow.Four kinds of excitation were applied on the square-back Ahmed model,including steady blowing along the vertical rear surface,two sides,and the lower edge,which is referenced as A1,A2,A3,and A4,respectively.Control parameters include excitation frequency fe,duty cycle α and mass flow coefficient Cm.Then,the optimization process of the system and the maximum drag reduction are explored.In order to understand the fluid physics behind the control,extensive flow field measurements were made with and without control.Furthermore,the machine learning response prediction model is calculated based on a large number of experimental data obtained from the MLC experiment.Finally,the sensitivity of each control parameter is analyzed based on the response prediction model.To sum up,a variety of turbulence control systems were designed in this paper by extremum search method,machine learning method based on linear genetic programming and EGM,respectively,to achieve the purpose of enhancing jet mixing and improving drag reduction and energy saving of square-back Ahmed model.Multi-parameter nonlinear optimization can be realized by using artificial intelligence control,and the best control effect is achieved among the three control methods.The results show that this paper has made some contributions to the improvement of turbulence control system.At the same time,the turbulence control system based on intelligent algorithm shows a strong application potential,and has a good reference value for the study of other turbulence control problems.