Jet Mixing Parametric Scaling And Machine Learning Control Based On Single Pulsed Minijet

The energy shortage is a major problem in today's world.How to use energy more effectively is a way to solve this problem.Jet control technology can improve the combus tion efficiency of the fuel by changing the blending of fuel and air,so that the fuel can be better utilized.The core technology of the engine has been aimed at the development of domestic aerospace,and the noise reduction of the engine is one of its core technologies.By controlling thej et,the engine wake is quickly mixed with the air,and it can quickly wake up.Jet control technology has been effectively applied in many industries,so it is very valuable to study the j et control technology.The main research content mainly includes two parts.The first part is the single-pulse minijet active control parameter manual optimization.Control parameters include minijet excitation frequency(5-150Hz),minijet mass flow(0-10L/min),and minijet duty cycle(0.1-1).Research the relationship between control parameters and scaling control parameters.The second part is using machine learning control to automatically optimize control parameters in the case of multiple Reynolds numbers.Research the effects of machine learning control.The experiment data is mainly measured by hot wire.Measuring the exit velocity of the minijet,the axial velocity decay of the main jet,and the velocity distribution of the main jet on the radial section of x/D=5.The method of machine learning control based on linear genetic programming.The optimization direction is to enhance the mixing of the axial j et center,thereby reducing the average velocity.A series of corrections have been made to the jet parameters such as jet velocity decay,minijet mass flow rate,and minijet duty cycle.The physical meaning of the modified scaling parameter Cm*/? is the effective per pulse flow ratio.There is a piecewise linear relationship between scaling parameters and jet velocity decay.As long as the scaling parameters are constant,the same control results can be obtained.The optimal excitation frequency increases with the increase of the Reynolds number in MLC experiment results,but the frequency ratio of the minijet to the main jet is close to 0.5.Under different Reynolds numbers the effective per pulse flow ratio Cm*/? of the scaling parameters of the optimal minijet are all close to 0.04,and J values are all close to 0.49.It proves that the optimal control effect has nothing to do with Reynolds number,and this machine learning method has good robustness.