基于神经网络的磨削砂轮状态的在线监测

利用声发射(AE)传感器和功率传感器为信号源,固定时间间隔内的声发射信号幅值增量累加及砂轮碰撞破碎时电机功率信号的陡变为砂轮状态识别的特征值,应用BP神经网络建立信号特征值与砂轮状态之间的非线性关系模型,可以为小批量、多品种产品磨削加工中砂轮状态的智能化在线监测提供准确有效的途径·测试结果证明了该系统的可行性,为磨削加工实现智能控制奠定了基础,并能为砂轮修整确定最佳的周期

On-Line Monitoring for Grinding Wheel States Based on Neural Network

The acoustic emission signal and the power signal of grinding wheel motor will change when wheel is dulled or broken in grinding process. Using an acoustic emission (AE) sensor and a power sensor as signal recourses,summation of acoustic emission signal increments in fixed time interval and sudden change of power signal of grindiny wheel motor as signal feature values,nonlineer relationship model between signal features and grinding wheel states was built with BP neural network. This model was trained by samples. Based on the model,a system was built which can realize intelligent on line monitoring for grinding wheel states and be a base for intelligent control of grinding machining. The testing results show that the system is workable and can determine the best period for wheel dressing.