Research On Output Energy Model Of Rotary Ultrasonic Machining Based On Elman Neural Network

With the rise of high-end materials science and precision manufacturing industry,the rapid development of high-performance hard and brittle materials have been widely used in aerospace,automotive,medical,optical,semiconductor and consumer electronics and other fields.Especially in the trend of 5G,high-performance hard and brittle ceramics will be more widely used.Rotary ultrasonic machining is internationally recognized as one of the important processing technologies in the field of hard and brittle materials processing.Compared with the conventional processing methods,it has obvious advantages in terms of precision,tool wear,surface finish and the like.Ultrasonic processing is suitable for almost all of hard and brittle materials.Ultrasonic processing has been widely used and rapid development in hard and brittle materials processing.In rotary ultrasonic machining,the initial ultrasonic amplitude has an important effect on the machining effect.In traditional rotary ultrasonic machining,manual testing of the amplitude of ultrasonic transducers is a common me thod of calibration.The ceramics will be affected by temperature,humidity and processing impact in complex field processing environment.These factors can lead to inaccurate calibration of artificial relationships and affect the processing results.In the field of processing,the different length of tools will lead to ultrasonic transducer amplitude output unpredictable.At the same time these factors also led to the instability of the processing effect.Rotary ultrasonic machining is an intermittent machining method.This processing method puts forward higher requirements on the consistency of the amplitude.In order to solve the existing problems of rotary ultrasonic machining,an improved ultrasonic machining system based on Elman neural network is des igned.According to the vibration characteristics of the ultrasonic transducer,a model of the vibration energy of the transducer is established and the relationship between the ultrasonic vibration energy and the ultrasonic amplitude is analyzed.The equivalent circuit of the transducer is established,and the admittance circle of the transducer and the relevant parameters of the transducer are obtained.At the same time,according to the equivalent principle of force,electricity and sound analogy,the transducer's equivalent model of electromechanical force is established.The transducer's amplitude model is further investigated according to the equivalent model.The relationship among transducer output amplitude and transducer material parameters and driving current is obtained.Due to the complex dynamics of the transducer and the complex material properties of the piezoelectric ceramic,it is difficult to obtain a precise mathematical model of the transducer's characteristic parameters and amplitude.Based on these difficulties,this paper presents Elman neural network method to predict the driving current and amplitude.The Elman neural network model is established by using the neural network tools in MATLAB.At the same time,the feasibility of this scheme is verified through simulation,and the scheme is implemented in the actual system.Finally,the ultrasonic machining system based on Elman neural network can effectively solve the above problems.This system can obtain better consistency of amplitude during intermittent processing.