No abstract

The reconfiguration technology, which is the significant feature of the newly designed Integrated Modular Avionics (IMA) system, enables the transfer of avionics functions from the failed module to the residual normal module, thereby enhancing the robustness of the whole system. The basic target of the IMA reconfiguration is to ensure the safe flight and correct execution of the mission. To solve the problem of lack of effective management mechanism for the IMA system development and safety assessment, a safety analysis method based on STAMP/STPA and UPPAAL for IMA reconfiguration is proposed. The method focuses mainly on system characteristics and multiparty interactions. On the basis of this approach, some studies and analyses have been carried out. Firstly, the STAMP/STPA principle is studied and used to identify unsafe control actions in the reconfiguration process. Secondly, a formal model of IMA reconfiguration is developed using UPPAAL. Finally, the accessibility analysis of the formal model is used to analyze UCAs and the corresponding loss scenarios. The method enables a detailed description of the interactions between the components and a rigorous mathematical analysis of the system, thereby diluting the effect of human factors while ensuring the accuracy and reliability of the safety constraints.

With the continual enhancement of the onboard avionics, the minimum flight crew has been downsized from five to two-person crew mode, and reduced crew operation has drawn extensive attention from aviation experts. Single-pilot operation (SPO) mode warrants careful account and research. This study investigated the intention modeling of commercial aviation single pilot based on the bidirectional long short-term memory (BiLSTM), mining the intention tendency of pilots’ behavior through artificial intelligence technology. This was done to avoid safety hazards caused by different intents and inconsistent operations of the single pilot and the cockpit automation system. The classification task of a single pilot’s behavior is the core of intention recognition. Various operation items contribute differently to the classification. To construct the interaction dataset and encode it into time series features, a single-pilot experiment is specifically performed, wherein the experience of an expert is summarized into single-pilot intent labels. The deep information in the feature vector of a single-pilot operation item is captured by the BiLSTM network, and the neural weight is adaptively assigned by the training mechanism. The operation sequence with the feature data is finally loaded into the softmax layer for intention classification. The proposed method is evaluated against long short-term memory (LSTM), term frequency-inverse document frequency (TF-IDF), convolutional neural network (CNN), Naive Bayesian (NB), and distributed representation’s intention modeling techniques. Because the proposed methods have higher F1 scores, the model can effectively share real-time information about the single-pilot intention with the cockpit automation system.

Defects and errors in code are different in that they are not detected by editors or compilers but pose a potential risk to software operation. For safety-critical software such as airborne software, the code review process is necessary to ensure the proper operation of software applications and even an aircraft. The traditional manual review method can no longer meet the current needs with the dramatic increase in code sizes and variety. To this end, we propose Deep Reviewer, a general and flexible code review framework that automatically detects code defects and correlates the review comments of the defects. The framework first preprocesses the data using several methods, including the proposed D2U flow. Then, features are extracted and matched by the detector, which contains a pair of twin LSTM models, one for code defect type detection and the other for review comment retrieval. Finally, the review comment output function is implemented based on the masks generated by the code defect types. The method is validated using a large public dataset, SARD. For the binary-classification task, the test results of the proposed are 98.68% and 98.67% in terms of precision and F1 score, respectively. For the multi-classification task, the proposed framework shows a significant advantage over other methods.