基于语音识别技术的内燃机车运行安全数据分析系统设计
Design of an In-Service Safety Data Analysis System for Diesel Locomotives Based on Speech Recognition Technology
王高飞 1肖磊1
作者信息
- 1. 国能朔黄铁路发展有限责任公司机辆分公司,河北 沧州 062350
- 折叠
摘要
研究设计一个基于语音识别技术的内燃机车运行安全数据分析系统.通过采集内燃机车运行过程中不同工况下的音频信号,使用深度前馈序列记忆网络-连接时序分类(Deep Feedforward Sequential Memory Network-Connectionist Temporal Classification,DFSMN-CTC)模型进行语音识别,并结合单类支持向量机(Support Vector Machine,SVM)实现异常检测.实验结果表明,该系统在语音识别与异常检测方面均表现出较高的准确性与健壮性.
Abstract
This study designs a safety system for diesel locomotives.It uses speech recognition technology.Audio signals are collected from the diesel locomotive under different conditions.The Deep Feedforward Sequential Memory Network-Connectionist Temporal Classification(DFSMN-CTC)model is used to recognize speech.One-class Support Vector Machine(SVM)is used to detect anomalies.The system shows high accuracy and robustness.
关键词
内燃机车/语音识别/异常检测Key words
diesel locomotive/speech recognition/anomaly detection引用本文复制引用
出版年
2024
NETL
NSTL
万方数据