An improved model-free adaptive control approach(iMFAC) with virtual reference feedback tuning(VRFT) is proposed for a class of discrete-time nonlinear systems. Firstly, the original nonlinear system is transformed into the compact form dynamic linearization(CFDL) data model by using dynamic linearization method. Then, the control algorithm and pseudo-partial-derivative(PPD) estimation algorithm are designed based on the optimization technique. Finally, estimation algorithms of the initial value and the reset value of PPD are designed based on virtual reference feedback tuning, respectively. The controller design depends merely on the measured input and output(I/O) data of the controlled plant and guarantees BIBO stability and tracking error convergence of the closed-loop system. Numerical simulation results show the effectiveness of  the proposed approach.