Adaptive virtual queue (AVQ) algorithm is an effective method aiming to achieve low loss, low delay and high-link utilisation at the link. However, it is difficult to guarantee fast response, strong robustness and good trade-off over a wide range of network dynamics. The authors propose a stabilising active queue management (AQM) algorithm - effective-AVQ, as an extension of AVQ, to improve the responsiveness and robustness of the transmission control protocol (TCP)/AQM system. Specifically, a proportional integral derivative (PID) neuron is introduced to tune the virtual link capacity dynamically. Also we derive the parameter self-tuning mechanism for the PID neuron from the Hebbian learning rule and gradient descent approach. The stability condition of the closed-loop system is presented based on the time-delay control theory. The performance of effective-AVQ is validated in the NS2 platform. Simulation results demonstrate that effective-AVQ outperforms AVQ in terms of steady-state and transient performance. It achieves fast response, expected link utilisation, low queue size and small delay jitter, being robust against dynamic network changes.