The Mechanism Of Multiple Parameters Composite Effect To Zero Feature Of Deflector Jet Servo Valve And Experimental Study

With high precision and fast response,the Electro-hydraulic servo valve is regarded as an important part in the Electro-hydraulic servo control system and widely applied in military equipment.In manufacturing,despite correct processing on the slide valve couples(proper radial dimensions of sliding spool and sleeve and appropriate lapping matching),it is problematic for the zero feature to meet the requirements during testing.The solution is to conduct adjustment and testing based on specific situations,causing low efficiency and quality risks,which cannot satisfy the increasing demand of manufacturing.The reason is that there is no influence relationship developed between zero feature and multiple effect parameters in manufacturing.For the problem above,this thesis introduces the research on the effect mechanism of the zero feature with multiple parameters,and its detailed processes are shown as follows.First,the definition of zero feature and representative curve are defined.All the effect parameters are listed and then filtered by the gradation – gradient method,then the effect parameters indicating main effects are selected and investigated.An intermediate flow area is selected to develop the relations between multiple parameters and flow area as well as between flow area and curve.The influence formula for these two relations are derived respectively.Secondly,servo valve fluid path model is developed and simulated.Ranging from single factor to multiple factors,the analysis first evaluates the influence of single effect factor on the curve of zero feature and then expands to that with multiple effect factors.In terms of simulation data,with different combinations of effector factors,curves of no-load characteristics,pressure characteristics and internal leakage are plotted.The influence of different effect parameters on opening shape is investigated,then the model is modified.Later six characteristic indexes are extracted to develop backpropagation neural network model to generate predictions based on different combinations of the rest effect parameters.The weights of each parameter for extracted characteristic indexes are calculated by combining data from prediction andsimulation,which conducts a quantitative investigation of influence of multiple effect parameters on zero feature.Finally,the experiments are conducted and characteristic indexes are extracted based on measured curves.In terms of the experimental data of effect factors,the predicted characteristic indexes are calculated and compared to those from the experiments.The reasons of errors are investigated.