Fault Diagnosis Of Avionics Based On Clustering Method

The degree of automation and intelligence of avionics equipment has been significantly improved with the flourish of aviation industry.However,due to its increasingly complex structural design,various problems have also followed.For example,problems such as the aging of the original equipment and insufficient maintenance have begun to seriously haunt the improvement of production efficiency,and even cause equipment failures,causing catastrophic consequences.Fault diagnosis for avionics is the key to ensure the smooth operation of avionics.This thesis summarizes the fault types of avionics by combining expert experience,and applies the clusteringbased method to the fault diagnosis of avionics.The research contents are as follows:1.The thesis conducts statistical analysis on the failure data of avionics,the fault types of avionics equipment are classified by comprehensively considering the risk degree,fault frequency and fault causes,combined with expert experience.It also summarizes a total of 10 fault types,and briefly analyzes the impact of these fault types on avionics system.2.Deep Clustering Network(DCN)is introduced to solve the clustering problem of high-dimensional fault data sets.In order to reduce the loss of information in the deep neural network learning,that is,to ensure the long distance dependence in the network training,the self attention mechanism is introduced into the self coding network of deep clustering network.In addition,residual blocks are added into the coding network to solve the network degradation problem caused by the increase of network depth in network training.Experimental results on the dataset in this thesis show that the proposed algorithm performs significantly better than the DCN algorithm.3.The thesis analyzes the fault diagnosis method based on data mining in detail,and proposes to apply the clustering method to the fault diagnosis of avionics equipment.The fault diagnosis experiments of avionics equipment are carried out by using the traditional clustering model,the traditional model of dimension reduction before clustering and the model based on deep clustering algorithm,and the experimental results are analyzed Finally,an avionics fault diagnosis system based on clustering method is implemented by decomposing the system into front end and back end.