﻿ 基于支持向量机的船舶机舱火灾温度快速预测
 舰船科学技术  2018, Vol. 40 Issue (1): 148-152 PDF

Ship engine room fire temperature fast prediction based on support vector machine
ZHANG Xi, LI Yan
Jiangsu University of Science and Technology, College of Electronic Information, Zhenjiang 212003, China
Abstract: The fire temperature change process under various conditions are calculated by large eddy simulation model based on FDS platform. The research takes ship engine room as the object, the conditions include the fire power , the activation time of exhaust system, the exhaust flow and the are of make-up air inlets. The support vector machine of fire temperature prediction model in ship engine room is built on the basis of FDS numerical simulation results sample. In order to improve the accuracy of the model predictions, the genetic algorithm is selected to optimize parameters . The experiments show that, the prediction results of the GA-SVM model are basically identical with the FDS calculation results , and are better than the prediction results of BP neural network and the SVM model, an engineering calculation approach for fast predicting the temperature of the fire is brought forth.
Key words: ship engine room     fire     FDS     support vector machine     fast prediction
0 引　言

1 船舶机舱火灾数值模拟

1.1 船舶机舱火灾场景设计

 图 1 FDS场景的几何模型 Fig. 1 The geometric model of the FDS scene
1.2 船舶机舱火灾样本计算

2 船舶机舱火灾温度预测模型的建立 2.1 基于SVM模型

 ${y'} = w \cdot \phi (x) + b\text{。}$ (1)

2.2 遗传算法参数寻优

3 结果对比及分析 3.1 参数优化对比

 图 2 GA-SVM的适应度曲线和SVM的适应度曲线比较 Fig. 2 Comparing the fitness of GA-SVM and SVM
3.2 预测结果分析

 图 3 三种模型下的预测效果 Fig. 3 Prediction results based on different models

 图 4 GA-SVM模型下相对误差 Fig. 4 Relative error based on GA-SVM

 图 5 SVM模型下相对误差 Fig. 5 Relative error based on SVM

 图 6 BP神经网络下相对误差 Fig. 6 Relative error based on BP Neural Network

3种模型预测性能结果对比见表5，用MSEeMAPER三个指标进行预测性能评价。

4 结　语

 [1] 贾佳, 陆守香. 船舶舱室火灾风险分析方法研究[J]. 安全与环境学报, 2014, 14(3): 132–137. JIA Jia, LU Shou-xiang. Study of fire risk analysis method applied to ship compartment[J]. Journal of safety and environment, 2014, 14(3): 132–137. [2] 苌占星. VR技术在船舶舱室火灾处理中的应用研究[D]. 大连: 大连海事大学, 2015. [3] 周德闯. 基于虚拟现实平台的火灾场景计算和仿真研究[ D]. 合肥: 中国科学技术大学, 2010: 72–96. [4] 徐文强, 刘芳, 董龙洋, 等. 基于FDS的地下停车场火灾数值模拟分析[J]. 安全与环境工程, 2012, 19 (1): 73–76. XU Wenqiang, LIU Fang, DONG Longyang, et al. Analysis on the fire of underground parking based on FDS numerical simulation[J]. Safety and Environmental Engineering, 2012, 19 (1): 73–76. [5] 胡靖. 船舶封闭舱室火灾温度分布特性实验研究[D]. 合肥: 中国科学技术大学, 2010: 8–10. [6] 辛喆, 王顺喜, 云峰, 等. 基于火灾模拟软件(FDS)的草原火灾蔓延规律数值分析[J]. 农业工程学报, 2013, 29 (11): 156–163. XIN Zhe, WANG Shun-xi, YUN Feng, et al. Numerical analysis on spreading laws of grassland fire based on fire dynamics simulator (FDS)[J]. Transactions of the Chinese Society of Agricultural Engineering, 2013, 29 (11): 156–163. [7] 顾亚祥, 丁世飞. 支持向量机研究进展[J]. 计算机科学, 2011, 38 (2): 14–17. GU Yafei, DING Shifei. Advances of support vector machines(SVM)[J]. Computer Science, 2011, 38 (2): 14–17. [8] 陈伟根, 滕黎, 刘军, 等. 基于遗传优化支持向量机的变压器绕组热点温度预测模型[J]. 电工技术学报, 2014, 29 (1): 44–51. CHEN Wei-gen, TENG Li, LIU Jun, et al. Transformer winding hot-spot temperature prediction model of support vector machine optimized by genetic algorithm[J]. Transactions of China Electrotechnical Society, 2014, 29 (1): 44–51. [9] 杨钟瑾. 粒子群和遗传算法优化支持向量机的破产预测[J]. 计算机工程与应用, 2013, 49 (18): 265–270. YANG Zhong-jin. Bankruptcy prediction based on support vector machine optimized by particle swarm optimization and genetic algorithm[J]. Computer Engineering and Applications, 2013, 49 (18): 265–270. DOI: 10.3778/j.issn.1002-8331.1303-0352 [10] CHAUDHURI A, DE K. Fuzzy support vector machine for bankruptcy prediction[J]. Applied Soft Computing, 2011, 11 (2): 2472–2486. DOI: 10.1016/j.asoc.2010.10.003 [11] WANG Chunlin, ZHOU hao, LI Guoneng. Combining support vector machine and genetic algorithm to predict ash fusion temperature[J]. Proceedings of the CSEE, 2010, 27 (8): 11–15.