﻿ 信息融合技术在船舶动力系统状态评估中的应用
 舰船科学技术  2022, Vol. 44 Issue (15): 123-126    DOI: 10.3404/j.issn.1672-7649.2022.15.025 PDF

Application of information fusion technology in condition evaluation of ship power system
YUAN Jing-jin
Jiangsu Shipping College, Nantong 226010, China
Abstract: The working state of ship power system can be identified by vibration, oil detection, temperature detection and other methods, but the traditional state identification method is relatively simple. To solve this problem, this paper designs an online state evaluation platform of ship power system based on information fusion technology and SOM neural network technology, which uses the powerful processing ability of neural network, and integrates the vibration sensor, pressure sensor, the temperature sensor and other signals are analyzed and processed, and finally the accurate working state of the ship power system is obtained. This paper focuses on the overall structure of the state evaluation system, the principle of SOM neural network, and the entropy characteristics of vibration signals and power signals. Finally, the state evaluation of the dynamic system is carried out based on SOM and information fusion technology.
Key words: dynamic system     condition evaluation     information fusion     SOM neural network     vibration     entropy
0 引　言

1 船舶动力系统状态评估平台的整体设计

 图 1 船舶动力系统状态评估平台示意图 Fig. 1 Schematic diagram of ship power system condition evaluation platform

1）报警模块

2）数据采集模块

3）故障诊断模块

4）趋势预测模块

2 船舶动力系统状态评估的分析方法研究

1）油液分析法

2）温度分析法

3）振动分析法

 图 2 动力系统状态评估的信息融合原理图 Fig. 2 Schematic diagram of information fusion for power system state evaluation
3 基于SOM神经网络技术的船舶动力系统状态融合技术开发 3.1 SOM神经网络的研究

 图 3 SOM神经网络算法的拓扑结构 Fig. 3 Topological structure of SOM neural network algorithm

SOM神经网络算法的工作流程为：

 ${d_j} = \left\| {X - {W_j}} \right\| = \sqrt {\sum\limits_{i = 1}^m {{{\left( {{x_i}(t) - {w_{ij}}(t)} \right)}^2}} } \text{。}$

 ${{\text{w}}_{ij}}(t + 1) = \left\{ {\begin{array}{*{20}{c}} {{w_{ij}}(t) + \delta (t)\left[ {{x_i} - {w_{ij}}(t)} \right]}&{\left( {i \in {S_k}(t)} \right)} \;，\\ {{w_{ij}}}&{\left( {i \notin {S_k}(t)} \right)} \;。\end{array}} \right. \text{}$

 ${f_{{\rm{out}}}} = f\left( {\min \left\| {{X_j} - {W_j}} \right\|} \right) 。$
3.2 基于信号熵值的船舶动力系统信号分析

 $M = \sum\limits_{i = 1}^m {} \left\{ {{A_i}} \right\} \text{，} {A_i} \cap {A_j} = \varPhi 。$

 $H\left( A \right) = \sum\limits_{i = 1}^m {\mu \left( {{A_i}} \right)} \log \mu \left( {{A_i}} \right) \text{。}$

 图 4 不同振动频率下动力系统的熵值变化曲线 Fig. 4 Entropy change curve of dynamic system under different vibration frequencies

 $X = \left( {{x_1},{x_2},\cdots,{x_m}} \right) \text{。}$

 ${\boldsymbol{A}} = \left[ {\begin{array}{*{20}{c}} {{x_1}}&{{x_2}}& \cdots &{{x_M}} \\ {{x_2}}&{{x_3}}& \cdots &{{x_{M + 1}}} \\ \vdots & \vdots &{}& \vdots \\ {{x_{N - M}}}&{{x_{N - M + 1}}}& \cdots &{{x_N}} \end{array}} \right] \text{。}$

 ${H_s} = - \sum\limits_{i = 1}^K {{p_i}} \log {p_i} 。$

 图 5 不同柴油机转速下的振动信号图 Fig. 5 Vibration signal diagram under different diesel engine speeds
3.3 基于SOM神经网络和信息融合的船舶动力系统状态评估

1）SOM初始参数

2）训练集

 图 6 船舶动力系统状态评估平台的振动信号预测 Fig. 6 Prediction of vibration signal of ship power system condition evaluation platform
4 结　语

 [1] 阮羚, 谢齐家, 高胜友, 等. 人工神经网络和信息融合技术在变压器状态评估中的应用[J]. 高电压技术, 2014, 40(3): 822-828. RUAN Ling, XIE Qi-jia, GAO Sheng-you, et al. Application of artificial neural network and information fusion technology in transformer condition evaluation[J]. High Voltage Technology, 2014, 40(3): 822-828. DOI:10.13336/j.1003-6520.hve.2014.03.026 [2] 杨雪松, 刘勇. 基于BP和SOM神经网络的云平台数据合并技术[J]. 甘肃科技, 2013, 29(9): 24-26. YANG Xue-song, LIU Yong. Cloud platform data merging technology based on BP and SOM neural network[J]. Gansu Science and Technology, 2013, 29(9): 24-26. DOI:10.3969/j.issn.1000-0952.2013.09.009 [3] 赵雨薇, 马波, 刘锦南. 基于RBF神经网络和多信息融合技术的往复压缩机状态评估研究[J]. 压缩机技术, 2013(6): 7-11+15. ZHAO Yu-wei, MA Bo, LIU Jin-nan. Research on condition evaluation of reciprocating compressor based on RBF neural network and multi information fusion technology[J]. Compressor Technology, 2013(6): 7-11+15. DOI:10.3969/j.issn.1006-2971.2013.06.002 [4] 李冬辉, 贾巍等. 基于小波神经网络和数据融合的直流系统故障检测方法及实现[J]. 电力系统保护与控制, 2005, 33(22): 6-9. LI Dong-hui, JIA Wei. DC system fault detection method and implementation based on wavelet neural network and data fusion[J]. Power System Protection and Control, 2005, 33(22): 6-9. DOI:10.3969/j.issn.1674-3415.2005.22.002 [5] 袁英, 郁丰, 踪华, 等. 基于深度BP神经网络的智能信息融合技术[J]. 西北工业大学学报, 2021, 39(S1): 89-95. YUAN Ying, YU Feng, ZONG Hua, et al. Intelligent information fusion technology based on deep BP neural network[J]. Journal of Northwestern Polytechnical University, 2021, 39(S1): 89-95.