The problem of finite-time stabilization for a class of nontriangular nonlinear system is studied based on fuzzy logic systems. By applying the sign functions in the control functions, a power form of the error is improved to ensure a continuous stabilizer and a continuous differentiable Lyapunov function, thereby ensuring finite time stability. Then, by introducing the global Barrier Lyapunov functions and finite time stability theory, a novel approach is proposed to construct a reliable controller when all the state variables are constrained. The results show that the system output can well follow the specified reference signal in finite time, and all signals in the system are guaranteed within fixed limits. One simulation example demonstrates the feasibility of the approach developed in this paper.