基于GA-PSO-Otsu算法的质子交换膜燃料电池催化层孔结构自适应识别

doi:10.19562/j.chinasae.qcgc.2023.09.018

摘要/Abstract

摘要：

车载质子交换膜燃料电池催化层的孔结构识别效率低、精度差且实验要求严格，无法适应日趋规模化的行业发展体系，因此针对该问题，本文提出基于遗传粒子群的最大化类间方差（GA-PSO-Otsu）优化算法，实现对催化层扫描电镜图孔径分布和孔隙率高效、精确且自适应的识别和测算。首先，协同引入高斯卷积核与二值化阈值最大化类间方差，有效降低噪声和手动调参对精度和效率的影响，实现自动化去噪和孔结构识别；其次，进一步提出遗传粒子群算法，有效解决传统方法遍历参数耗时长和易陷入局部优化的问题，兼具高精度和高效率的优点；最后，通过对催化层结构和灰度分布差异明显的扫描电镜图的对比实验验证，表明该方法具备良好的鲁棒性、自适应性和实用性，与遍历所有参数的传统Otsu算法的孔隙率误差小于0.5%，测算耗时降低约26.2%。

关键词: 质子交换膜燃料电池（PEMFC）, 催化层孔径分布, 催化层孔隙率, 遗传算法（GA）, 粒子群算法（PSO）, 最大化类间方差法（Otsu）

Abstract:

The low efficiency， low accuracy and strict experimental requirements for the detection of the proton exchange membrane fuel cell （PEMFC） catalyst layer porous structure can’t adapt to the increasingly large-scale industry development system. Therefore， to address the problem， this paper innovatively proposes the genetic algorithm-particle swarm optimization-Otsu （GA-PSO-Otsu） algorithm to realize efficient， accurate and self-adaptive identification of pore size distribution and porosity calculation of the scanning electron microscope （SEM） of the catalyst layer. Firstly， Gaussian convolution and binary threshold are combined to maximize the inter-class variance between the foreground and background to effectively reducethe impact of noise and manual adjustment of parameters on accuracy and efficiency， which ensures automatic noise reduction and pore structure detection. Furthermore， the genetic algorithm based particle swarm optimization method is proposed to solve the problem of long time consuming caused by traverse parameters and to avoidlocal optimization， with the advantages of high accuracy and high efficiency. Lastly， the comparative analysis of different algorithms applied on various PEMFC catalyst SEMs with different component ratios and gray scales indicate that the proposed method has high robustness， self-adaptiveness and practicability. Compared with the traditional Otsu approach which traverses all parameters， the porosity error of the proposed method is less than 0.5% and the calculation time is significantly reduced by 26.2%.

Key words: proton exchange membrane fuel cell （PEMFC）, catalyst layer pore size distribution, catalyst layer porosity, genetic algorithm （GA）, particle swarm optimization （PSO）, Otsu

袁新杰,刘芳,侯中军. 基于GA-PSO-Otsu算法的质子交换膜燃料电池催化层孔结构自适应识别[J]. 汽车工程, 2023, 45(9): 1702-1709.

Xinjie Yuan,Fang Liu,Zhongjun Hou. Self-adaptive Porous Structure Detection of the Catalyst Layer in PEMFCs Based on GA-PSO-Otsu Algorithm[J]. Automotive Engineering, 2023, 45(9): 1702-1709.

图/表 10

图1

图2

图3

图4

表1

GA-PSO-Otsu算法的核心参数［15，19-20］"

参数名称	参数符号	参数值
高斯卷积核半径上下限	$[r g, ? m i n, ? r g, ? m a x]$	$[1, ? 10]$
二值化阈值上下限	$[t h r e s m i n, ? t h r e s m a x]$	$[0, ? 255]$
交叉概率	$p c r o s s o v e r$	$0.6$
变异概率	$p m u t a t i o n$	$0.04$
种群规模	$N p o p$	$300$
初始惯性权重值	$ω s$	$0.9$
最大迭代次数的惯性权重值	$ω e$	$0.4$
最大迭代次数	$s m a x$	$50$
学习因子1	$c 1$	$2$
学习因子2	$c 2$	$2$

表1

图5

图6

图7

图8

图9

参考文献 20

1	国家发改委，国家能源局. 氢能产业发展中长期规划（2021-2035 年）［M/OL］. ［2022-03-24］. http：//www.gov.cn/xinwen/2022-03/24/content_5680975.htm.
	National Development and Reform Commission， National Energy Administration. Plan on the development of hydrogen energy for the 2021-2035 period［M/OL］. ［2022-03-24］. http：//www.gov.cn/xinwen/2022-03/24/content-5680975.htm.
2	KONGKANAND A， MATHIAS M F. The priority and challenge of high-power performance of low-platinum proton-exchange membrane fuel cells［J］. The Journal of Physical Chemistry Letters， 2016（7）： 1127-1137.
3	SASSIN M B， GARSANY Y， ATKINSON R W， et al. Understanding the interplay between cathode catalyst layer porosity and thickness on transport limitations en route to high-performance PEMFCs［J］. International Journal of Hydrogen Energy， 2019（44）： 16944-16955.
4	EIKERLING M. Water management in cathode catalyst layers of pem fuel cells［J］. Journal of Electrochemical Society， 2006（153）： 58-70.
5	SOBOLEVA T， MALEK K， XIE Z， et al. PEMFC catalyst layers： the role of micropores and mesopores on water sorption and fuel cell activity［J］. ACS Applied Materials and Interfaces， 2011（3）： 1827-1837.
6	DIM P E， FLETCHER R S， RIGBY S P. Improving the accuracy of catalyst pore size distributions from mercury porosimetry using mercury thermoporometry［J］. Chemical Engineering Science， 2016（140）： 291-298.
7	KATAYANAGI Y， SHIMIZU T， HASHIMASA Y， et al. Cross-sectional observation of nanostructured catalyst layer of polymer electrolyte fuel cell using FIB/SEM［J］. Journal of Power Sources， 2015（280）： 210-216.
8	LIU X C， PARK K， SO M， et al. 3D generation and reconstruction of the fuel cell catalyst layer using 2D images based on deep learning［J］. Journal of Power Sources， 2022（14）： 10084.
9	ZHAO Y Q， YU X Y， WU H B， et al. A fast 2-D Otsu lung tissue image segmentation algorithm based on improved PSO［J］. Microprocessers and Microsystems， 2020（80）： 103527.
10	JOY C A， UMAMAKESWARI A. A novel percentage split distribution method for image thresholding［J］. Optik （Stuttg）， 2020（218）： 164953.
11	ISIET M， GADALA M. Self-adapting control parameters in particle swarm optimization［J］. Applied Software Computing Journal， 2019（83）： 105653.
12	REGASAMY S， MURUGESAN P. PSO based data clustering with a different perception［J］. Swarm and Evolutionary Computing， 2020（64）： 100895.
13	GONZALEZ-JIMENEZ A， LOMAZZI L， CADINI F， et al. On the mitigation of the RAPID algorithm uneven sensing network issue employing averaging and Gaussian blur filtering techniques［J］. Composite Structure， 2021（278）： 114716.
14	CAO Y， SONG H B， KAIWARTYA O， et al. Electric vehicle charging recommendation and enabling ict technologies： recent advances and future directions［J］. IEEE COMSOC MMTC Communications - Frontiers， 2016（11）： 1-10.
15	DUTTA K， TALUKDAR D， BORA S S. Segmentation of unhealthy leaves in cruciferous crops for early disease detection using vegetative indices and Otsu thresholding of aerial images［J］. Measurement： Journal of the International Measurement Confederation， 2022（189）： 110478.
16	ZHANG R， LU S， YU H， et al.Recognition method of cement rotary kiln burning state based on Otsu-Kmeans flame image segmentation and SVM［J］. Optik （Stuttg）， 2021（243）： 167418.
17	YUAN X J， LIU Y C， BUCKNALL R. Optimised MOPSO with the grey relationship analysis for the multi-criteria objective energy dispatch of a novel SOFC-solar hybrid CCHP residential system in the UK［J］. Energy Conversion and Management， 2021（243）： 114406.
18	SOMEYA H， YAMAMURA M. A genetic algorithm for function optimization［J］. IEEJ Transactions on Electronics， Information and Systems， 2002（122）： 363-373.
19	WANG Z Z， CAO L H， SI H Y. An improved genetic algorithm for determining the optimal operation strategy of thermal energy storage tank in combined heat and power units［J］. Journal of Energy Storage， 2021（43）： 103313.
20	CAI J J， MA X Q， LI Q， et al. A multi-objective chaotic particle swarm optimization for environmental/economic dispatch［J］. Energy Conversion and Management， 2009（50）： 1318-1325.