Cyber-Aware Quantum Network Control with Hybrid Quantum Anomaly Diagnostics and Risk-Aware Purification

Quantum networks depend on a classical control plane for heralding, synchronization, routing, purification, and feedforward, creating an operational coupling between cyber state and entanglement quality. We present a modular simulation study of cyber-aware purification in quantum-repeater chains, alongside hybrid quantum anomaly diagnostics. First, a leakage-free hybrid classical--quantum autoencoder (HCQAE) is evaluated on five unseen CICDDoS2019 DrDoS families using a six-qubit, two-layer variational latent module, a VQE-inspired latent-energy observable, Bloch-space geometry, local R\'enyi-2 entropy, and quantum Fisher information (QFI). Across three seeds, the fixed composite score achieves mean ROC-AUC 0.9903 and mean F1 0.9621, while reconstruction remains the dominant operational ranking signal and the quantum latent layer provides complementary diagnostics. Second, CUDA-Q noisy kernels validate purification and swapping primitives that are embedded in a SeQUeNCe-style event-layer Monte Carlo model. In an eight-node stationary chain, resource-aware predictive purification raises above-target delivery from $0.311\pm0.006$ to $0.362\pm0.007$ relative to fixed purification while using fewer purifications. Finally, under a benign-to-SSDP trace, IDS-aware control increases attack-period above-target delivery from $0.098\pm0.007$ to $0.344\pm0.011$, close to the oracle-aware $0.335\pm0.011$. A matched 6D classical control reached ROC-AUC 0.9937, so no quantum advantage is claimed. Hence, hybrid quantum models should be treated as representation-rich diagnostic tools whose anomaly signals can become actionable inputs for fidelity-aware quantum-network control.