An active queue management controller based on Hamiltonian energy theory for a class of nonlinear TCP network congestion system in the p resence of uncertain parameters and unknown external disturbances is derived. The restriction of inequality assumption is eliminated by introducing the MiniMax methods into dissipation Hamilton system. Sufficient conditions for the existence of MiniMax controller under circumstances the network system is attacked with maximum impact disturbance has been derived via Lyapunov stability theory. Furthermore, the nonlinear uncertainties presence i.s successfully ameliorated by employing a parameter projection mechanism. Simulation experiments have demonstrated this energy based control strategy ameliorates the pertinent control complexity and is considerably more effective in improving both transient stability and robustness.