基于声音识别的多阀门泄漏检测系统

阀门在石油工业中使用频繁且出现泄漏现象不易检测。论文阐述了通过软件方式根据声音信号识别阀门泄漏的方法。系统利用声音传感器采集有效时间段的声音信号,提取其MFCC特征参数作为观测序列,并利用隐马尔可夫模型对其进行建模训练,使其能够识别故障信号,及时发现阀门泄漏,并进行报警处理,从而保证输油泵正常工作。实验证明,系统能够有效地识别阀门泄漏故障。

Multi-valve leakage detective system based on acoustics

A method of detecting valve leakage based on acoustics was developed in this study.The built system could pick up sound signals within effective time by acoustic sensor and the MFCC characteristic parameter was chosen as observation sequence.After trained by Hidden Markov Model,the system could identify breakdown signals,detect valve leakage in time and give an alarm to ensure the oil pump work well.Testing results showed that the system could effectively identify the problem of valve leakage.