Influence Of Nozzle Jet On Performance Of Nozzle Flapper Valve And Research On Jet Image Detection

Nozzle is an important part of twin flapper-nozzle electro-hydraulic servo valve.Because of its high design requirements and great processing difficulty,it is necessary to carry out special quality inspection after processing.One of the commonly used detection methods is the jet column measurement method.The operation method is to absorb hydraulic oil with rubber ear washing ball,pressurize the hydraulic oil at the end of the nozzle to spray it from the nozzle.Observing the jet shape,if the jet keeps a certain distance of horizontal injection,the nozzle processing quality is considered qualified,and if other shapes such as oblique,scattering and spiral appear,it is considered unqualified.(1)In order to explore the theoretical basis of the influence of jet shape on the whole valve,and improve the nozzle jet detection technique,the corresponding nozzle models are established for horizontal jet,oblique jet,scattering jet and spiral jet respectively.The effects of four jet patterns on the performance of the whole valve in noload flow characteristics,pressure characteristics,vortex and cavitation phenomena are simulated and analyzed.According to the four jet patterns,the corresponding detection qualified indexes are put forward.The influence of nozzle jet on the whole valve under different index values is simulated and analyzed,and the critical value of qualified indexes is determined.(2)Based on deep learning algorithm,YOLO v3 convolution neural network model is selected,and the image detection of nozzle jet type and position is realized through data acquisition,data preprocessing,algorithm improvement and model training.According to the prediction information of the convolution neural network model,the detection index values corresponding to the four jet patterns are calculated,and the qualified conditions of nozzle processing quality are judged accordingly.(3)Comparing the visualization experiment of nozzle flapper flow field and simulation results,the correctness of the jet qualified index is verified.Through the test experiment of image detection model,the difference between the model prediction index and the actual index is analyzed,which proves that there is no obvious difference between the model prediction index and the actual index.