摘要
在理解线性预测分析(LPC)方法及其自相关法和自协方差法两种高效解法基础上,根据舱音短时平稳特点,提出了基于LPC原理的Durbin迭代的舱音信号特征提取与重构方法,构建了舱音信号特征提取与重构方法的步骤,得到了详细图形和结果。获得了最佳预测阶数P、满足线性预测解稳定条件等有益结论。研究表明:基于线性预测分析方法得到的参数和结论为舱音信号特征识别提供了一条比较好的新途径。
Aircraft black box cockpit sound, recorded by CVR （cockpit voice recorder）, is one of the main evidence to analyze accident cause for airplane crash. And the characteristics of cockpit sound are the key in wreckage aircraft accident investigation. Some researches are made in the paper. Firstly, two high effect computing solution of autocorrelation ＆ auto covarianee and linear prediction coding （LPC） are applied in the research according to the characteristics of cockpit sound short-term stationary.; Then methods of acquirement and reconstruction of cockpit sound signal characteristics are put forward through Durbin arithmetic of LPC and the approach of acquirement and reconstruction for cockpit sound signal are formed; Thirdly, the relative figures and results in details are gained. Finally, the best predictive order number P is gotten through LPC, which meets the beneficial conclusion of LPC＇s stable conditions. The study shows： parameters and conclusions based on principle of LPC provide a better new method for the identification of the characteristics of cockpit sound signals.
出处
《微计算机信息》
2012年第10期61-63,共3页
Control & Automation
基金
基金颁发部门:国家自然科学基金委
项目名称:基于舱音特征融合及飞行状态预测诊断安全技术研究
基金编号:(61071203)
基金申请人:程道来
关键词
计算机
声信号
飞机
舱音特征提取与重构
线性预测分析(LPC)
Computer
sound signal
aircraft
cockpit voice characteristics＇ acquirement and reconstruction
Linear prediction coding （LPC）
