<p>Aircraft engine fault diagnosis is critical for ensuring flight safety and optimizing maintenance resources. Traditional single task learning approaches, based on uniform standards, often fail to account for engine-specific differences, which can lead to misdiagnosis in complex fault scenarios. This paper proposes a fault diagnosis method based on multi-task graph support vector machine inference, which constructs a graph structure to integrate individualized features into the fault identification process, thereby improving diagnostic accuracy. The key innovation of this method lies in leveraging the graph structure for engine similarity representation and using an individualized fine-tuning strategy to adapt to fault patterns. The experimental results demonstrate that the proposed method significantly improves accuracy in fault detection, especially under conditions of high individual variance and effectively reduces misdiagnosis rates. These findings highlight the importance of incorporating engine-specific differences into diagnostic models to enhance accuracy and reliability, particularly for similar fault types, and suggest potential for advancing intelligent aircraft engine health monitoring systems.</p>