【摘要】 自治式水下机器人执行使命时，沿规划路径运动的导航，特别是围绕被搜索到的目标物作圆周运动的导航，是一种强非线性的控制问题、传统方法解决这一问题很困难，神经网络为非线性控制问题的实施提供了快速、准确的工具。利用神经网络构造的非线性控制器为自治式水下机器人的导航和定位提供了一种新的途径，网络经训练后能适应混有噪声的非线性环境。本文综合利用前惯和反馈控制器的优点，把GESA算法应用于自动增强的函数链网络，首先对自治式水下机器人实施前馈控制，再利用神经网络构成的PID控制器来提高控制系统的控制精度，使自治式水下机器人沿规划轨线运动。更多还原

【Abstract】 Navigation which AUV carries out its missions along planned path, especially around already located objects of under-water,is a highly nonlinear control problem. It is difficult to solve effectively this kind of problem withconventional method, neural network offers new approach for navigation and positioning of AUV. After training,neural network adapts itself to nonlinear ’noise’ environment. In the paper,we synthetically use merits of feedforwardand feedback control,and apply GESA algorithm to auto-enhancement functional-link net, and improve accuracy ofcontrol system,then ensure AUV move along planned path.更多还原