【作者】 王凯；

【导师】 邬再新；

【作者基本信息】 兰州理工大学 ， 机械制造及其自动化， 2007， 硕士

【摘要】 数控加工技术是近代发展起来的一种自动控制技术，是用数字化的信息驱动机床运动从而加工出复杂型面零件轮廓形状的一种方法，是先进制造技术的重要组成部分。数控加工的目的是要提高产品加工的精度和加工效率，因此必须对加工过程进行一定的控制以满足生产的需要。但由于复杂型面零件的数控加工过程具有复杂的、非线性、不确定性等特点，用传统的基于被控对象精确数学模型的方法已经难以获得良好的控制效果。为了解决这一问题，本文将神经网络技术应用在复杂型面数控加工的控制中，利用神经网络拥有优秀的逼近能力、泛化能力和自学习能力的特点，以实现提高加工精度和加工效率的目的。将神经网络方法应用在数控加工过程的插补控制，利用神经网络的并行性能和可以模拟任意非线性函数的特性，使得插补运算的时间大幅度缩短，提高插补计算的速度，实现对空间任意曲线或空间离散点的直接插补。利用其具有良好的非线性逼近能力及隐式函数的构造能力，通过对数控系统进行网络辨识，并对误差补偿技术和误差控制的神经网络实现方法进行分析，建立误差补偿控制器的神经网络模型，实现较为精确的加工误差补偿技术，提高了工件的加工精度。利用神经网络技术的自学习能力和泛化能力，将其应用在复杂型面零件的模型建立上，这种方式不需要工程技术人员掌握相关的知识和进行大量的、复杂的数学计算，大大降低了零件模型建立的难度、减少了建模时间。通过仿真试验分析可以看出，将神经网络技术应用在复杂型面零件的数控加工控制中是可行的，与传统的控制方法相比较，基于神经网络的数控加工控制技术的控制方案效果较好，控制精度较高，具有实时性和稳定性的特点，能够提高数控加工的效率和加工精度，有较好的实际应用价值。更多还原

【Abstract】 The technology of computer numerical control (CNC) is a kind of modern automatic control technology, it is a kind of methods of controlling machine tools to machine complex curve workpieces which have been realized by using the digital information, it is an important component of advance manufacturing technology. The purpose of CNC machining is to improve precision and efficiency in manufacture, so the process of CNC should be controlled in order to achieve manufacture needs. But the process of CNC is a complicated, non-linear, unknown and variable dynamic process, and it almost can’t be described by precise mathematical model, so by using traditional control we can’t get a well result.In order to solve this problem, this article will use artificial neural network (ANN) technology in numerical control machining process control. By using its excellent approximation ability, generalization ability and self-learning ability, improving the precision and efficiency in manufacture. Making use of neural network technology in process of CNC interpolation control, in order to improve the calculation speed and make a direct interpolation for arbitrary or discrete curve by making use of its parallel performance and non-linear ability. ANN has an excellent capability of nonlinear approximation and reconstructing functions , we can make an improvement of workpieces machining accuracy by constructing a ANN controller model which is used for error’s compensation. Because of its excellent generalization ability and self-learning ability, using a neural network to reconstruct the model of workpieces of complex curve surface. This method does’t need engineers do lots of calculation and master interrelated knowledge, so it can improves the speed of reconstructing workpieces and reduces the difficulty of reconstruction.The research and the simulation calculation demonstrate this method has good ability to solve some problems in numerical control machining of complex curve surface. Comparing with traditional control methods, computer numerical control method that based on neural network has a better controllable effect, it can improve the efficiency and accuracy of machining process. This method has real-time and stability characteristic, it has an excellent practical value.更多还原