Research On Fault Classification Of Airborne Software

In the context of the continuous development of airborne systems,the proportion of software in airborne systems is increasing,and its quality has an increasingly significant impact on system functions and system safety,and the risks caused by software failures are also increasing.Software fault classification is of great significance for the repair and management of airborne software faults.An effective software fault classification method can help developers to quickly understand and repair faults and improve the overall quality of airborne software systems.At present,although there are standards and methods related to software fault classification,the existing fault categories can not comprehensively and accurately classify airborne software faults due to the complex operating environment and fault causes of airborne software.In addition,the traditional airborne software fault classification usually adopts manual classification,and the software fault classification effect is not good.For this reason,this paper focuses on three aspects:fault type,fault standardized description and fault automatic classification of airborne software.Specifically,according to the characteristics of airborne software faults,a new method to define the types of airborne software faults is proposed.On this basis,the standardized description of airborne software faults is given.Combined with the characteristics of airborne software fault data,an automatic classification model of airborne software fault based on machine learning is discussed.The self-attention mechanism is introduced into the classification model to realize the automatic classification of airborne software fault,which can effectively assist the repair and management of airborne software fault and improve the overall reliability of airborne system.In order to verify the effectiveness of the proposed method,software fault classification experiments are conducted on real data sets and public data sets.The results show that the proposed model based on self-attention mechanism can improve the effectiveness of fault classification.