Application of Probabilistic Neural Network on Aviation Magnetos Fault Diagnosis

Fault diagnosis of aviation magneto is critical to ensuring the safety and reliability of aviation systems. Traditional diagnostic methods often fall short when dealing with complex and evolving fault patterns. This study proposes a fault diagnosis approach based on the Probabilistic Neural Network (PNN), aimed at enhancing diagnostic accuracy through the analysis of simulated data. First, a simulation model of the aviation magneto was developed to replicate typical fault scenarios, such as stator winding fault, bearing fault and rotor eccentricity fault, thereby generating a comprehensive fault dataset containing multi-source signal features. Subsequently, a PNN-based diagnostic framework grounded in Bayesian decision theory was constructed to perform fault classification. The reliability and robustness of the PNN were systematically validated through experimental analysis. Furthermore, comparative experiments with other neural network models demonstrated the superior performance of PNN in terms of diagnostic accuracy and computational efficiency. The results indicate that the PNN holds significant promise for practical application in the fault diagnosis of aviation magneto.