为通过舰船水下噪声评估及时发现舰船存在的异常，采用基于隐马尔可夫模型的评估方法。分别选取舰船正常状态下和异常状态下的水下噪声数据作为两类训练样本，送进隐马尔可夫模型进行训练，得到对应的模型参数；把待评估样本送进经过训练的两类隐马尔可夫模型进行概率估计，概率明显大的模型对应的状态，即为评估判断结果。选取202个数据样本进行试验，测试结果正确率超过95%，表明隐马尔可夫模型有较好的分类判断能力，可用于舰船水下噪声评估。
In order to detect the abnormality of the ship in time by underwater noise assessment, the evaluation method based on Hidden Markov model (HMM) was adopted. The data of underwater noise under normal and abnormal conditions were selected as two kinds of training samples, which were sent to the hidden Markov model for training, and the corresponding model parameters were obtained. Those were the results to evaluate. A total of 202 samples of underwater noise data ship working conditions were tested. The results show that the HMM has a good ability of classification and judgment of underwater noise, and can be used to estimate the underwater noise of ships.
2019,41(9): 121-124 收稿日期：2018-09-03
DOI：10.3404/j.issn.1672-7649.2019.09.023
分类号：O429
作者简介：吴毅斌(1992-),男,硕士研究生,研究方向为水声工程及信号处理
参考文献：
[1] 汤晓迪, 程一超. 外军舰船声磁监测主要做法及启示[J]. 船舶电子工程, 2016, 267(9):14-17
[2] HAWKINS D. Identification of Outliers[M]. Chapman and Hall, London, 1980.
[3] 管景崇, 胡金华. 基于隐马尔可夫模型的线谱跟踪技术[J]. 舰船科学技术, 2014, 36(5):110-115
GUAN Jing-chong, HU Jin-hua. Line spectrum tracking technology based on Hidden Markov model[J]. Ship Science and Technology, 2014, 36(5):110-115
[4] 上海航空测控技术研究所. 航空故障诊断与健康管理技术[M]. 北京:航空工业出版社, 2013:239-240.
[5] 许丽红, 余小清, 万旺根. 基于HMM算法的语音识别系统设计及其混合编程实现[J]. 计算机工程, 2002, 28(5):139
XU Li-hong, YU Xiao-qing, WAN Wang-gen. Speech recognitionstudy based on MFCC and CDHMM and its realization using a mixed program method[J]. Computer Engineering, 2002, 28(5):139
[6] 陆振波, 章新华, 朱进. 基于MFCC的舰船辐射噪声特征提取[J]. 舰船科学技术, 2004(2):51-54
LU Zhen-bo, ZHANG Xin-hua, ZHU Jin. Feature extraction of ship radiated noise based on MFCC[J]. Ship Science and Technolog, 2004(2):51-54
[7] 张帅林. 基于HMM的关键词语音识别技术在智能家居中的应用研究[D]. 兰州, 兰州大学, 2017.
[8] 张杰, 黄志同, 王晓兰. 语音识别中隐马尔可夫模型状态数的选取原则及研究[J]. 计算机工程与应用, 2000(1):67-69, 133
[9] 王强. 深度学习方法及其在水下目标识别中的应用[A]. 中国声学学会水声学分会. 中国声学学会水声学分会2015年学术会议论文集[C]//中国声学学会, 2015:3.
WANG Qiang. Deep Learning Method and its application in underwater target recognition[A]. Underwater acoustic branch of Chinese acoustic society. Proceeding of the 2015 Symposium of the Underwater Acoustic Branch of Chinese Acoustic Society[C]//Acoustic Society of China, 2015:3.
[10] CHEN J, WANG Y, WANG DL. A feature study for classification-based speech separation at low signal-to-noise ratios[J]. IEEE/ACM Transactions on Audio, Speech and Language Processing, 2014, 22(12):1993-2002
[11] JIANG Y, WANG DL, LIU RS, et al. Binaural classification for reverberant speech segregation using deep neural networks[J]. IEEE/ACM Transactions on Audio, Speech and Language Processing, 2014, 22(12):2112-2121
