陆空通话复诵语义自动化校验BiLSTM模型  

Automatic Semantic Verification of BiLSTM Model for Readbacks in Radiotelephony Communication

作　　者：李丹 贾桂敏[1] 程方圆 杨金锋[1] 郭晓静 LI Dan;JIA Gui-min;CHENG Fang-yuan;YANG Jin-feng;GUO Xiao-jing(Tianjin Key Lab for Advanced Signal Processing, Civil Aviation University of China, Tianjin 300300, China)

机构地区：[1]中国民航大学天津市智能信号与图像处理重点实验室,天津300300

出　　处：《信号处理》2019年第1期57-64,共8页Journal of Signal Processing

基　　金：国家自然科学基金(U1433120;61379102;61502498)

摘　　要：为保证航空运输安全,飞行员要对管制员发送的指令进行复诵,并且管制员要对复诵指令进行进一步的确认。而由于疲劳、紧张、疏忽等原因,管制员未能及时发现飞行员复诵错误的情况也时有发生,给民航运输安全带来巨大隐患。针对这个问题,本文提出一种陆空通话复诵语义自动化校验BiLSTM模型。首先,利用两个并行的长短时记忆网络(BiLSTM)对管制员发送的指令和飞行员复诵的指令分别进行语义特征提取;然后将两个BiLSTM网络各个时刻的输出进行交互得到一个指令和复诵指令间的语义匹配矩阵;最后经过一个动态k-Max池化层后输入到多层感知器中从而得到指令与复诵指令间最终的匹配分数来判别复诵语义是否一致。实验证明,该方法在解决陆空通话复诵语义自动校验任务中是有效的,平均测试精度达到了90. 53%。To ensure the safety of air transportation, the pilot must read back the instructions sent by the air traffic controller, and the controller has to further confirm it. However, due to fatigue, tension, negligence and other reasons, the controller often failed to find out the readback error in time, which is a huge hidden danger in the safety of civil aviation transportation. In order to solve this problem, an automatic semantic verification method based on bidirectional Long Short-Term Memory (BiLSTM) is proposed. Firstly, two parallel BiLSTM are used to extract the semantic features of controller ’ s instructions and the pilot ’ s readback respectively;then a matching matrix is generated by matching the output of the two BiLSTMs at each time step;finally, a k -Max pooling layer is applied to pick out the top- k matching features and input it into a Multi-Layer Perceptron (MLP) to obtain a final matching score to determine whether the readback is consistent in semantics with the instruction sent by controllers. The experiment shows that the method is effective in solving the automatic verification task of the aviation radiotelephony communication, and the average test accuracy is 90.53 %.

关 键 词：陆空通话 语义自动校验 双向长短时记忆网络 语义匹配矩阵 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]