High-precision terminal guidance for unmanned combat aerial vehicles (UCAVs) requires integrated guidance and control (IGC) strategies capable of managing nonlinear dynamics, subsystem interactions, and time-varying disturbances. Conventional design approaches treat guidance and control separately, often resulting in reduced accuracy and degraded stability in operational scenarios. This study addresses this limitation by investigating whether a hybrid adaptive controller—combining an offline-trained deep neural network (DNN) with a fuzzy inference system—can significantly improve interception performance in loitering munition missions. The hypothesis is that integrating DNN-based gain scheduling with fuzzy adaptation will reduce interception time and engagement range while maintaining robust stability under uncertainties. The IGC model is fully derived for a 3‑D engagement scenario, and three controllers are implemented: a classical PID, an adaptive offline DNN, and the proposed adaptive offline DNN‑fuzzy controller. High‑fidelity simulations under identical initial conditions were conducted to quantify performance. Results show that, compared with the PID controller (interception time ≈ 50 s, range ≈ 8000 m) and the adaptive offline DNN controller (35 s, 4600 m), the proposed controller intercepts the target in just 10.6 s at a range of about 610 m, corresponding to a 78.8% reduction in time-to-impact and a 92.4% reduction in range. These improvements enhance mission responsiveness, reduce target evasion opportunities, and broaden applicability to time-critical defense and high-speed interception scenarios.