Traditional intrusion detection systems have certain limitations in dynamic traffic adaptation and real-time processing efficiency. Spiking neural networks (SNNs) have demonstrated unique advantages in temporal information processing. Aiming at the problem that synaptic delays are difficult to be dynamically optimized in the training of traditional SNNs, we propose a delay learning algorithm based on spike train kernels for feedforward SNNs and applies it to network intrusion detection tasks. The proposed algorithm improves the network traffic temporal feature extraction capability of SNNs by co-optimizing the synaptic weights and delays to achieve efficient network intrusion detection. The performance of the proposed algorithm is verified by the NSL-KDD dataset, and compared with the static synaptic delay learning algorithm and other mainstream network intrusion detection algorithms. The experimental results show that the proposed algorithm can effectively improve the network intrusion detection performance of SNNs, achieving detection results comparable to other mainstream network intrusion detection algorithms.