【摘要】 对水下机器人运动控制神经网络的结构与学习方法进行了研究 ,提出了带有局部回归结构的控制器网络结构、分步计算的动作过程、教师样本的生成方法以及变系数学习方法 .水下机器人“Twin_Burger”的水池实验结果 ,验证了本文所提方法的有效性更多还原

【Abstract】 Theoretical research on the structure and studying method of AUV control neural network was carried out.A not_all recurrent controller’s neural network was proposed.The process of step_by_step calculation,the creating method of teaching data and the variable standardization studying method were proposed also.The result of the experiment in pool by AUV“Twin_Burger”proves that the performance methods presented are reasonable.更多还原