Power system blackout scenario generation under deliberate attacks can help identify high-risk operation condition, which is the foundation of resilience enhancement by pre-disaster defense and post-disaster restoration. Therefore, this paper proposes a blackout scenario generation method for cyber-physical power system under deliberate attacks based on incomplete-information attack-defense game. First, considering the asymmetry of information between attackers and defenders in the confrontation process, an incomplete-information attack-defense game framework is established to support blackout scenario generation under deliberate attacks in cyber-physical power systems (CPPS). Furthermore, from the attacker's perspective, a comprehensive line importance assessment model is developed, serving as the foundation for the mid-layer attack model. Finally, an Information-Based Incomplete Defender-Attacker-Defender (IIDAD) model is formulated and solved using a hybrid approach combining particle swarm optimization (PSO) and mixed-integer linear programming (MILP) to determine the optimal attack-defense resource allocation strategy under deliberate attacks. Subsequently, blackout scenarios are generated based on Monte Carlo simulation. Case studies on the IEEE 39-bus New England system demonstrate that, compared to the complete-information game model, the proposed method generates blackout scenarios that better align with practical system conditions. Additionally, it achieves more precise attack-defense resource allocation strategies, leading to a more reasonable simulation of the blackout process.