【摘要】 提出了一种基于松散型小波网络的切削表面粗糙度预测方法。结合切削参数和切削振动理论,建立了预测网络结构,为避免频域混叠,采用小波包改进算法来实现振动信号去噪。根据振动加速度及切削参数,利用Elman网络的非线性映射和学习机制,实现切削表面粗糙度的实时在线预测。为减少训练时间,用遗传算法对网络权重进行预先优化。实验表明,该方法的预测误差小于3%。更多还原

【Abstract】 A forecast method based on relax-type wavelet network for cutting surface toughness was indicated.The forecasting network structure was established by considering the influence of cutting parameters and vibration.The noise in cutting vibration signals was filtered with the reformed wavelet pack algorithm to avoid aliasing in frequency domain.The real-time forecast was achieved by the nonlinear mapping and learning mechanism in Elman network according to the vibration acceleration and cutting parameters.The weights in network were optimized with genetic algorithm in advance to reduce learning time.The forecasting error of this method is less than 3% in experiments.更多还原