在网络异常检测中,为了提高对异常状态的检测率,降低对正常状态的误判率,该文提出利用TSK模糊控制系统进行网络异常检测的新方法。在对TSK模糊控制系统的训练中采取梯度下降算法,充分发挥梯度下降局部细致搜索优势。实验数据采用KDDCUP99数据集,实验结果表明,基于梯度下降的模糊控制系统提高了异常检测的准确性。 In the network anomaly detection, in order to improve the detection rate of the abnormal state and reduce the false positive rate of the normal state, this paper proposes a new method of network anomaly detection using the TSK fuzzy control system. In the training of TSK fuzzy control system to take a gradient descent algorithm, give full play to the gradient descent local meticulous search advantage. Experimental data using KDDCUP99 dataset, the experimental results show that the fuzzy control system based on gradient descent improves the accuracy of anomaly detection.