Research On Vibration Effect And Thickness Detection:Application To Wire Electrical Discharge Machining

Wire Electrical Discharge Machining(WEDM)is a non-contact cutting process which has the advantages of high precision,no cutting force,high flexibility etc.It has been widely used in the fields of mechanical electronics,aerospace,automobile,household appliances,ships and military industry.It also becomes a frontier subject for scholars of various countries.With the development of industry,the requirements for quality and precision of WEDM are becoming more demanding.However,in parts with difficult geometries(such as parts with low circular radius and parts with variable thickness),the conventional strategy of using a single set of process parameters for the whole machining path does not lead to optimum results.The vibration effect of electrode wire and the cutting error caused by thickness variation of workpiece are discussed in this paper.In order to improve the stability and efficiency of processing,the researches about identification and correlation of WEDM process signals with wire vibration effects on part accuracy(wire lag effect and concavity effect)and identification of part thickness directly from process variables using techniques of Artificial Intelligence(Artificial Neural Networks)are carried out.Firstly,this thesis summarized the State of the Art in conventional WEDM at home and abroad,carried out the elaboration of the process precision and explained the causes of the vibration effect of the wire resulting in the cutting error during the machining process.Meanwhile,it also points out that the development prospect of artificial intelligence in this field.In order to determine the correlation between cutting parameters and electrode wire vibration effect,it is necessary to extract the 3D cutting surface topography of different machining conditions and get the numerical value.Afterwards,the regression curve of vibration effect is constructed by MATLAB programming method.The research shows that the wire lag and concavity effects of the workpiece with small cutting radius are much more obvious,which leads to the greater accuracy error.The technical requirements for the processing of such parts need to be improved accordingly,and the interpolation value is set more consistently with the actual processing.WEDM is a complex cutting process with multiple parameters and uncertainty.Based on the 9 representative attributes of process recorded by recorder and artificial neural networks with strong nonlinear modeling ability,a BP neural network workpiece thickness prediction model were trained and validated in the MATLAB Neural Network Toolbox.And verifications indicate that the relative error between the predictive thickness and actual one is from 2.30%to 2.93%,the mean standard deviation is 1.422.The most experimental values of thickness prediction obtained are in the allowance 5mm range for quantity of thickness target,while the extreme short period of obvious deviation can be neglected.The ANNs thickness prediction model exhibited significant goodness of fit in the present study,meeting the technical requirements of machining,reflected the nonlinear relation between machining parameters and workpiece thickness.