动态是系统最基本的特性．为了实现复杂动态系统的实时控制，文中对已有动态神经网络结构进行了改进，使之易于应用于一般未知系统；在此基础上研究了运用动态神经网络实现直接自适应控制的方法，并针对回转类复杂曲面车削过程的特点，提出了相应的动态神经网络直接自适应控制方案，以综合实时校正加工过程中参数缓慢时变、突变以及非线性等复杂动态因素的影响．针对典型零件的车削实验表明，该方法可有效地减小复杂轮廓加工误差 Dynamic is the most basic feature of the system. In order to realize the real-time control of complex dynamic system, the structure of the existing dynamic neural network has been improved to make it easy to be applied to the general unknown system. On this basis, the method of direct adaptive control using dynamic neural network is studied. The characteristics of the turning process of turning complex curved surface are analyzed and the corresponding dynamic neural network direct adaptive control scheme is proposed to comprehensively and real timely correct the influence of complex dynamic factors such as slow time-varying, sudden change and non-linearity in the machining process. Turning experiments on typical parts show that this method can effectively reduce the machining error of complex contours