基于人工神经网络的超声加工振幅的在线监测

摘要

为了使超声振动加工发挥有效的作用，文章提出了利用人工神经网络分析影响系统频率变化的参数，来保证超声加工的振幅始终工作在最大值附近的方法。换能器的输入电流与其振动位移成比例，并且易于测量和控制，可作为反映系统振幅变化的参数。而影响系统振幅变化的因素中，系统温度的变化也和工件的温度的变化成比例，因此可将工件温度的变化作为测量参数。将可测量参数换能器的输入电流和加工件温度的变化作为人工神经网络的输入，进行训练，建立换能器输入电流与超声波发生器频率调整间的关系，达到实时调节超声波发生器频率使整个系统处于谐振的作用。最后，将此的方案应用在功率超声珩磨加工中，通过换能器输入电流和加工效果证明此方案的效果。%This paper raise the method that analyze current of transducer based on artificial neural network, so that ultrasonic machining can be effective. This method ensure the amplitude of ultrasonic machining vary in a range of maximum. The input current of transducer is proportional to its vibration displacement, and easy to measure and control, so it is regarded as a parameter to reflect the actual changes in system ampli-tude. The temperature variation is also regarded as a parameter to reflect the changes in system amplitude. Using artificial neural network training data, establish relationship between input current of transducer and adjustment frequency of the ultrasonic generator, make the whole system in the resonance state. Finally, the scheme is proved effective through applied to ultrasonic honing machine.