结合 Word2vec 和 BiLSTM 的民航非计划事件分析方法

安全是民航业的核心主题。针对目前民航非计划事件分析严重依赖专家经验及分析效率低下的问题, 文章提出一种结合 Word2vec 和双向长短期记忆(bidirectional long short-term memory, BiLSTM) 神经网络模型的民航非计划事件分析方法。首先采用 Word2vec 模型针对事件文本语料进行词向量训练, 缩小空间向量维度; 然后通过 BiLSTM 模型自动提取特征, 获取事件文本的完整序列信息和上下文特征向量; 最后采用 softmax 函数对民航非计划事件进行分类。实验结果表明, 所提出的方法分类效果更好, 能达到更优的准确率和 $ F_{1} $ 值, 对不平衡数据样本同样具有较稳定的分类性能, 证明了该方法在民航非计划事件分析上的适用性和有效性。

Safety is the core theme of the civil aviation industry. Aiming at the problem that the analysis of civil aviation unplanned events heavily depends on expert experience and the low analysis efficiency, a civil aviation unplanned event analysis method is proposed, which combines Word2vec and bidirectional long short-term memory (BiLSTM) neural network model. Firstly, Word2vec is used to train word vectors for event text corpus, reducing the dimension of the space vector. Then, the features are automatically extracted by BiLSTM model to obtain the complete sequence information and context feature vector of the event text. Finally, the softmax function is used to classify civil aviation unplanned events. The experimental results show that the proposed method has better classification effect and can achieve better accuracy and $ F_{1} $ value, and also has more stable classification performance for unbalanced data samples, which proves the applicability and effectiveness of this method in the analysis of civil aviation unplanned events.