針對水下潛器實際工作的需求和水下沉浮運動特征，提出一種基于自適應模糊控制的水下潛器自主沉浮控制方法。該方法從優化隸屬函數入手，采用多層前向神經網絡的誤差反向傳播(EBP)算法對它的參數進行在線修正，并采用Delta-Bar-Delta學習規則對學習速率進行在線調整，使EBP算法具有較快的收斂速度，同時避免了局部極小值問題。仿真實驗表明，對于水下潛器自主沉浮運動的不能精確建摸、干擾嚴重的非線性與時變情況，這種自適應模糊控制是一種較好的控制方式。

　　Abstract:

　　As for actual operation requirement and characteristics of ups and downs movement of underwater submersible vehicle， an independent ups and downs control method based on adaptive fuzzy control was advanced. In this method， on optimization of membership function， EBP algorithm of multiple layers forward neural network was used to correct parameters of membership function， and learning rule of Delta-Bar-Delta was used to adjust learning speed， which mended convergence speed of EBP algorithm， and the problem of sectional minimum was avoided at the same time. Simulation experiments show that independent ups and downs movement of underwater submersible vehicle could not be modeled precisely， and disturbance had characteristics of serious nonlinear and time change， which could be improved by this adaptive fuzzy control method.