Type-2 fuzzy logic systems have been applied in various control problems because of their abilities to model uncertainties in a more effective way than type-1 fuzzy logic systems. In this paper, a novel learning algorithm is proposed to train type-2 fuzzy neural networks. In the approach, instead of trying to minimize an error function, the weights of the network are tuned by the proposed algorithm in a way that the error is enforced to satisfy a stable equation. The parameter update rules are derived and the convergence of the weights is proved by Lyapunov stability method. To illustrate the applicability and the efficacy of the proposed method, the control problem of Duffing oscillator with uncertainties and disturbances is studied. The simulation studies indicate that the type-2 fuzzy structure with the proposed learning algorithm result in a better performance than its type-1 fuzzy counterpart.