针对多入多出非仿射非线性系统,提出了一种径向基网络补偿逆模型误差的自适应控制方法.将难以求逆的非仿射项分解为可逆部分和不可逆部分,可逆部分作为理想逆来近似系统的直接逆,逆模型误差用径向基网络的自适应控制信号补偿,网络权值利用不可逆部分非仿射信息更新,应用均值理论和Lyapunov函数证明了自适应控制律的稳定性.仿真结果验证了该方法的有效性. Aiming at the multi-input and multi-output non-affine nonlinear system, an adaptive control method is proposed to compensate the inverse model error of radial basis network. The non-affine items which are difficult to invert are decomposed into reversible part and irreversible part. The direct inverse and inverse model error of ideal inverse approximation system is compensated by adaptive control signal of radial basis network. The network weights are updated by irreversible partial non-affine information. The mean value theory and Lyapunov function are used to prove the stability of adaptive control law The simulation results verify the effectiveness of this method.