Abstract: A PID-like Neural Network Controller(PIDNNC) and its design method are proposed.Based on the advantages of simple construction and good property in PID and self-regulation and adaptivity of neural networks,a multivariable adaptive neural network controller with PID structure is created.It is composed of a hybrid locally connected recurrent network with an activation feedback and an output feedback respectively in the hidden layer.By means of defining error function as the design objective,using resilient back-propagation(BP) algorithm and changing rate and sign in resilient BP algorithm,some differential relations are dealt with.The modified formula for on-line updating network weights is obtained.Lyapunov stability principle is used to derive the range of learning rate to ensure control system stability.Finally,numerical examples are given to show advantages of the proposed controller.