对一类确定性Hammerstein系统 ,给出了基于神经网络的自适应控制算法。考虑到神经网络的非线性特点 ,特别是其自适应学习能力 ,控制系统采用两个神经网络分别作为估计器和控制器 ,通过在线训练网络的权重来获得模型参数和控制输入。神经网络的训练用Widrow -Hoff学习规则。对算法的全局收敛性进行分析表明系统具有总体收敛性 ,输入输出有界。 For a class of deterministic Hammerstein system, an adaptive control algorithm based on neural network is given. Taking into account the nonlinear characteristics of neural networks, especially its adaptive learning ability, the control system uses two neural networks as estimators and controllers, respectively, and obtains model parameters and control inputs by online training network weights. Neural network training Widrow-Hoff learning rules. The analysis of the global convergence of the algorithm shows that the system has overall convergence and the input and output are bounded.