Fault Diagnosis of Integrated Core Processor in Avionics System Based on CNN-Transformer

The structure and function of Integrated Core Processor (ICP) of the avionics system are relatively complex, while traditional fault diagnosis methods perform low accuracy and low efficiency. To address these issues, this paper proposes an enhanced CNN-Transformer algorithm for fault diagnosis of the integrated core processor. Initially, CNN model is employed to extract spatial features, reduce computational complexity, and preserve essential data. Subsequently, Transformer model utilizes self-attention calculations to capture relationships and characteristics within the input sequence. Finally, a fully connected neural network is applied to classify the faults. The method was validated using multiple datasets recorded by the Fiber Channel bus, achieving a diagnostic accuracy of 96.88%, outperforming other comparative approaches.