基于语音识别的空中交通态势评估方法

摘要

With the fast－booming development of civil aviation transportation industry ,air traffic flow also has rapid growth.As the first line of civil aviation ,the workload of air traffic controller may influence the safety of this traffic system.Therefore,this paper has developed a speech recognition system which based on iFLY talk for the ATC situation,it can transfer the air traffic controllers′voice into text accurately ,and the specific instruction information of speech text will be extracted by an improved regular expression which focuses on the prefix factor filtering.Then the flight models including flight speed ,flight height and flight course are proposed to describe flight processes ,and an identification method of air traffic situation based on a HMM－BP hybrid model is in combination of flight models to predict the air traffic situation in current ATC instruction.Last Kunming Changshui international airport′s radar data is used to validate the accuracy of the method.%近年来,随着我国民航运输业的飞速发展,空中交通流量的迅速增长,位居民航运输第一线的管制员承担的压力日益增大.管制员的决策和指令一旦出现细微的错误,都有可能会造成严重的安全事故.因此首先依据科大讯飞公司提供的讯飞开放平台开发适用于空中交通管制场景的语音识别系统,将管制语音通话储存为文本形式,利用基于前缀因子过滤的正则表达式指令匹配方法,从语音文本中提取出管制指令信息;随后结合航空器飞行特征构建航空器飞行速度、高度和航向状态变化模型,以此为基础提出基于隐马尔可夫－BP神经网络的空中交通运行态势识别方法,能够有效识别当前管制指令下的空域内航空器飞行状态即空域运行态势,从而为管制员发布管制指令提供参考.