当前位置：首页 > 过刊浏览->2023年45卷9期
一种基于高斯混合模型的主机运行工况构建方法
A method of constructing engine working conditions based on gaussian mixture model
- DOI:
- 作者:
- 乔继潘, 张焱飞, 陆思宇
QIAO Ji-pan, ZHANG Yan-fei, LU Si-yu
- 作者单位:
- 上海船舶运输科学研究所有限公司 航运技术与安全国家重点实验室, 上海 200135
State Key Laboratory of Navigation and Safety Technology, Shanghai Ship and Shipping Research Institute, Shanghai 200135, China
- 关键词:
- 高斯混合模型;船舶主机工况;相关性分析;置信区间;特征提取
Gaussian mixture model; ship main engine conditions; correlation analysis; confidence interval; feature extraction
- 摘要:
- 对复杂的船舶主机运行数据进行有效的工况划分可以提升主机设备的监测能力，基于工况下各个设备参数的特征值变化情况，进一步提升主机设备故障诊断和预估能力。为此，提出一种基于高斯混合模型（GMM）的主机工况构建方法。该方法首先对实船监测数据进行预处理和初步筛选，再对各个特征参数之间的相关性进行分析，从而确定转速和功率为聚类划分的特征参数。引入置信区间获取目标船主机在主要营运要求下的转速范围，提升工况划分的效率，结合GMM构建主机工况划分方法。基于GMM的工况划分方法可以对复杂的主机运行工况进行有效划分，能为各个工况下的主机设备监测提供可靠的数据支撑。
Effective work condition classification of complex ship main engine operation data can improve the monitoring capability of main engine equipment and further improve the fault diagnosis and prediction capability of main engine equipment based on the changes of the characteristic values of each equipment parameter under the work condition. In this regard, a Gaussian mixture model (GMM)-based method for constructing the main engine working conditions is proposed. The method firstly preprocesses and initially screens the real ship monitoring data and then analyzes the correlation between each characteristic parameter to determine the rotational speed and power as the characteristic parameters for clustering classification. Confidence intervals are introduced to obtain the speed range of the target ship's main engine under the main operating requirements, improve the working condition classification efficiency, and combine with GMM to build the main engine working condition classification method. The GMM-based condition classification method can effectively classify the complex main engine operating conditions and provide reliable data support for monitoring main engine equipment under each condition.
2023,45(9): 124-127 收稿日期：2022-07-21
DOI：10.3404/j.issn.1672-7649.2023.09.027
分类号：U664.12
作者简介：乔继潘(1990-),男,硕士,助理研究员,研究方向为智能船舶机舱数据分析
