基于改进的混合蛙跳算法对旱地小麦籽粒生长模型参数的优化

doi:10.3864/j.issn.0578-1752.2023.12.004

摘要/Abstract

摘要：

【目的】作为农业智能化生产的核心决策模块，作物模型的精确模拟取决于模型参数的高效准确优化。为了提高调参效率，提升作物模型的性能和精确度，本研究通过改进优化算法对旱地春小麦籽粒生长子模型进行单目标参数优化，为中国西北部黄土丘陵区旱地春小麦的适应性研究提供参考，扩大模型的应用范围，以便于模型更好地指导农业生产。【方法】立足于2015—2021年甘肃省定西市安定区凤翔镇安家坡村的田间试验，结合1970—2021年的天气数据和年鉴产量数据，在发挥传统混合蛙跳算法（SFLA）全局交流、局部深度搜索的基础上，使用轮盘赌选择策略对旱地小麦籽粒生长阶段的6个参数进行进一步的优化，进行算法改进前后产量实测值与模拟值的误差计算与对比，对APSIM-Wheat模型进行检验。【结果】（1）在相同的迭代次数下，传统混合蛙跳算法在200次左右收敛，改进后的混合蛙跳算法在100次左右收敛；（2）旱地春小麦籽粒生长阶段的参数优化结果为小麦茎部分的每克籽粒数为26.0；开花至开始灌浆阶段的潜在籽粒灌浆速率为0.00119 grain/d；灌浆阶段的潜在籽粒灌浆速率为0.00174 grain/d；氮限制下的潜在籽粒灌浆速率为6.20×10^-5 g grain/d；氮限制下的最小籽粒灌浆速率为1.90×10^-5 g grain/d；单株小麦籽粒部分的最大干重值为0.0437 g；（3）分别使用传统混合蛙跳算法优化后所得参数值和改进混合蛙跳算法优化后所得参数值模拟小麦产量，参数优化后，产量实测值与模拟值的均方根误差（RMSE）从363.22 kg·hm^-2降至57.85 kg·hm^-2，标准均方根误差（NRMSE）从21.78%降至3.47%。【结论】相较于传统的混合蛙跳算法，改进后的混合蛙跳算法增加了种群和子群的多样性，收敛速度快，提高了优化效率和精度，优化后的结果符合旱地春小麦的生长发育过程，适用性较高，明显改善了中国西北部黄土丘陵农业区APSIM-Wheat模型的性能。

关键词: APSIM-Wheat模型, 参数优化, 混合蛙跳算法, 旱地春小麦, 小麦籽粒生长阶段

Abstract:

【Objective】As the core decision module for intelligent agricultural production, the accurate simulation of the crop model depends on efficient and accurate optimization of the model parameters. In order to improve the efficiency of tuning parameters and enhance the performance and accuracy of the crop model, this study optimized the single objective parameters of the dryland spring wheat grain growth sub-model by improving the optimization algorithm, so as to provide a reference for the adaptation study of dryland spring wheat in the loess hilly region of northwestern China, to expand the application of the model, and to facilitate the model to better guide agricultural production.【Method】Based on a field experiment in Anjiapo Village, Fengxiang Town, Anding District, Dingxi City, Gansu Province, from 2015 to 2021, this study combined weather data and yearbook yield data from 1970 to 2021, further optimized six parameters of dryland wheat grain growth stage using roulette selection strategy based on the global communication and local depth search of traditional shuffled frog leaping algorithm (SFLA), carried out error calculation and comparison between measured and simulated yield values before and after algorithm improvement, and tested the APSIM-Wheat model.【Result】(1) At the same number of iterations, the traditional shuffled frog leaping algorithm converged around 200 times, while the improved shuffled frog leaping algorithm converged around 100 times. (2) The optimized parameters for the dryland spring wheat grain growth stage were: the grain number per gram stem was 26.0; the potential rate of grain filling from flowering to start of grain filling period was 0.00119 grain/d; the potential rate of grain filling during grain filling period was 0.00174 grain/d; the potential rate of grain filling under N limitation was 6.20×10^-5 g grain/d; the minimum rate of grain filling under N limitation was 1.90×10^-5 g grain/d; the maximum grain dry weight per plant was 0.0437 g. (3) The wheat yield was simulated using the parameter values optimized by the traditional shuffled frog leaping algorithm and the parameter values optimized by the improved shuffled frog leaping algorithm, respectively. After parameter optimization, the root mean square error (RMSE) between the measured and simulated yield values decreased from 363.22 kg·hm^-2 to 57.85 kg·hm^-2, and the normalized root mean square error (NRMSE) decreased from 21.78% to 3.47%.【Conclusion】Compared with the traditional shuffled frog leaping algorithm, the improved shuffled frog leaping algorithm increased the diversity of populations and subpopulations, converged quickly, and improved the optimization efficiency and accuracy, so the optimized results conformed to the growth and development process of dryland spring wheat with higher applicability, which significantly improved the performance of the APSIM-Wheat model in the loess hilly agricultural area of northwestern China.

Key words: APSIM-Wheat model, parameters optimization, shuffled frog leaping algorithm, dryland spring wheat, wheat grain growth stages

崔炜楠, 聂志刚, 李广, 王钧. 基于改进的混合蛙跳算法对旱地小麦籽粒生长模型参数的优化[J]. 中国农业科学, 2023, 56(12): 2274-2287.

CUI WeiNan, NIE ZhiGang, LI Guang, WANG Jun. Optimization of Dryland Wheat Grain Growth Model Parameters Based on an Improved Shuffled Frog Leaping Algorithm[J]. Scientia Agricultura Sinica, 2023, 56(12): 2274-2287.

0
/ / 推荐

导出引用管理器 EndNote|Reference Manager|ProCite|BibTeX|RefWorks

链接本文: https://www.chinaagrisci.com/CN/10.3864/j.issn.0578-1752.2023.12.004

https://www.chinaagrisci.com/CN/Y2023/V56/I12/2274

图/表 11

表1

表2

图1

图2

表3

图3

表4

表5

图4

图5

图6

参考文献 31

[1]	朱艳, 汤亮, 刘蕾蕾, 刘兵, 张小虎, 邱小雷, 田永超, 曹卫星. 作物生长模型(CropGrow)研究进展. 中国农业科学, 2020, 53(16): 3235-3256. doi: 10.3864/j.issn.0578-1752.2020.16.004
	ZHU Y, TANG L, LIU L L, LIU B, ZHANG X H, QIU X L, TIAN Y C, CAO W X. Research progress on the crop growth model CropGrow. Scientia Agricultura Sinica, 2020, 53(16): 3235-3256. (in Chinese) doi: 10.3864/j.issn.0578-1752.2020.16.004
[2]	王伟, 林团荣, 王真, 王玉凤, 焦欣磊, 张志成, 范龙秋, 王懿茜, 邢进, 尹玉和. 基于随机森林算法(RF)的马铃薯冠层叶绿素含量估测 //马铃薯产业与种业创新, 2022: 326-327.
	WANG W, LIN T R, WANG Z, WANG Y F, JIAO X L, ZHANG Z C, FAN L Q, WANG Y Q, XING J, YIN Y H. Estimation of chlorophyll content in potato canopy based on random forest algorithm (RF). //Potato Industry and Seed Innovation, 2022: 326-327. (in Chinese)
[3]	GASO D V, WIT A D, BERGER A G, KOOISTRA L. Predicting within-field soybean yield variability by coupling Sentinel-2 leaf area index with a crop growth model. Agricultural and Forest Meteorology, 2021, 308: 108553.
[4]	BEHESHTI Z, SHAMSUDDIN S M H. A review of population-based meta-heuristic algorithms. International Journal of Advances in Soft Computing and Its Applications, 2013, 5(1): 1-35.
[5]	尹起. 机器学习预测参考作物蒸散量在智能灌溉系统中的应用研究[D]. 乌鲁木齐: 新疆大学, 2021.
	YIN Q. Machine learning prediction of reference crop evapotranspiration in an intelligent irrigation system[D]. Urumqi: Xinjiang University, 2021. (in Chinese)
[6]	GOET G, SONKAR I, PRASAD K H. Estimation of non-linear water uptake parameter using genetic algorithm for sodic soils[R]. Copernicus Meetings, 2022.
[7]	聂志刚, 李广, 雒翠萍, 马维伟, 代永强. 利用混合蛙跳算法优化基于APSIM的旱地小麦产量形成模型参数. 作物学报, 2018, 44(8): 1229-1236. doi: 10.3724/SP.J.1006.2018.01229
	NIE Z G, LI G, LUO C P, MA W W, DAI Y Q. Parameter optimization in APSIM-Based simulation model for yield formation of dryland wheat using shuffled frog leaping algorithm. Acta Agronomica Sinica, 2018, 44(8): 1229-1236. (in Chinese) doi: 10.3724/SP.J.1006.2018.01229
[8]	聂志刚, 任新庄, 李广, 董莉霞, 马维伟, 唐洁, 刘小娥, 罗永忠. 基于APSIM的黄土丘陵区旱地小麦气候适宜性评价. 中国农业气象, 2017, 38(6): 369-377.
	NIE Z G, REN X Z, LI G, DONG L X, MA W W, TANG J, LIU X E, LUO Y Z. APSIM-based evaluation of climate suitability on wheat in dryland in the hill-gullied region of the Loess Plateau. Chinese Journal of Agrometeorology, 2017, 38(6): 369-377. (in Chinese)
[9]	李广, 黄高宝, BELLOTTI W, 陈文. APSIM模型在黄土丘陵沟壑区不同耕作措施中的适用性. 生态学报, 2009, 29(5): 2655-2663.
	LI G, HUANG G B, BELLOTTI W, CHEN W. Adaptation research of APSIM model under different tillage systems in the Loess hill-gullied region. Acta Ecologica Sinica, 2009, 29(5): 2655-2663. (in Chinese)
[10]	HOLZWORTH D P, HUTH N I, DEVOIL P G, ZURCHER E J, HERRMANN N I, MCLEAN G, CHENU K, OOSTEROM E J, SNOW V, MURPHY C, MOORE A D, BROWN H, WHISH J P, VERRALL S, FAINGES J, BELL L W, PEAKE A S, PEAKE P L, HOCHMAN Z, THORBURN P J, GAYDON D S, DALGLIESH N P, RODRIGUEZ D, COX H, CHAPMAN S, DOHERTY A, TEIXEIRA E, SHARP J, CICHOTA R, VOGELER I, LI F Y, WANG E, HAMMER G L, ROBERTSON M J, DIMES J P, WHITBREAD A M, HUNT J, REES H, MCCLELLAND T, CARBERRY P S, HARGREAVES J N G, MACLEOD N, MCDONALD C, HARSDORF J, WEDGWOOD S, KEATING B A. APSIM-evolution towards a new generation of agricultural systems simulation. Environmental Modelling & Software, 2014, 62: 327-350.
[11]	HOLZWORTH D, HUTH N I, FAINGES J, BROWN H, ZURCHER E, CICHOTA R, VERRALL S, HERRMANN N I, ZHENG B, SNOW V. APSIM next generation: Overcoming challenges in modernising a farming systems model. Environmental Modelling & Software, 2018, 103: 43-51.
[12]	ZHENG B, CHENU K, DOHERTY A, CHAPMAN S. The APSIM- wheat module (7.5 R3008). Agricultural Production Systems Simulator (APSIM) Initiative, 2014: 615.
[13]	WANG E, OOSTEROM E V, MEINKE H, ASSENG S, ROBERTSON M, HUTH N, KEATING B, PROBERT M. The new APSIM-Wheat model-performance and future improvements//Proceedings of the 11th Australian Agronomy Conference. Geelong, Victoria. Australian Society of Agronomy, 2003: 6.
[14]	郭林祥, 李广, 肖鸿民, 闫丽娟. 甘肃省定西市安定区1970-2010年下垫面气候变化特征分析. 草业科学, 2014, 31(11): 2015-2020.
	GUO L X, LI G, XIAO H M, YAN L J. Climate change characteristics of underlying surface in Anding District, Dingxi in Gansu Province from 1970 to 2010. Pratacultural Science, 2014, 31(11): 2015-2020. (in Chinese)
[15]	杨文雄. 中国西北春小麦. 北京: 中国农业出版社, 2016: 155-161.
	YANG W X. Spring Wheat in Northwest China. Beijing: China Agriculture Press, 2016: 155-161. (in Chinese)
[16]	吕东灿, 袁帅, 赵仲麟, 顿文涛, 李建伟, 袁志华, 袁超. Origin软件在实验数据拟合中的应用. 农业网络信息, 2016(2): 38-40.
	LÜ D C, YUAN S, ZHAO Z L, DUN W T, LI J W, YUAN Z H, YUAN C. Application of Origin software in experimental data fitting. Agriculture Network Information, 2016(2): 38-40. (in Chinese)
[17]	张静, 高尚. 基于轮盘赌反向选择机制的果蝇优化算法. 计算机与数字工程, 2020, 48(7): 1595-1600, 1742.
	ZHANG J, GAO S. Fruit fly optimization algorithm based on roulette reverse selection mechanism. Computer and Digital Engineering, 2020, 48(7): 1595-1600, 1742. (in Chinese)
[18]	HO-HUU V, NGUYEN-THOI T, TRUONG-KHAC T, LE-ANH L, VO-DUY T. An improved differential evolution based on roulette wheel selection for shape and size optimization of truss structures with frequency constraints. Neural Computing and Applications, 2018, 29(1): 167-185. doi: 10.1007/s00521-016-2426-1
[19]	EUSUFF M M, LANSEY K E. Optimization of water distribution network design using the shuffled frog leaping algorithm. Journal of Water Resources Planning and Management, 2003, 129(3): 210-225. doi: 10.1061/(ASCE)0733-9496(2003)129:3(210)
[20]	EUSUFF M, LANSEY K, PASHA F. Shuffled frog-leaping algorithm: A memetic meta-heuristic for discrete optimization. Engineering Optimization, 2006, 38(2): 129-154. doi: 10.1080/03052150500384759
[21]	NERI F, COTTA C. Memetic algorithms and memetic computing optimization: A literature review. Swarm and Evolutionary Computation, 2012, 2: 1-14. doi: 10.1016/j.swevo.2011.11.003
[22]	GUO Y, MA Y, ZHAN Z, LI B, DINGKUHN M, LUQUET D, DE REFFYE P. Parameter optimization and field validation of the functional-structural model GREENLAB for maize. Annals of Botany, 2006, 97(2): 217-230. pmid: 16390847
[23]	蒋纪芸, 苏珮. 小麦灌浆期间营养器官对籽粒氮积累的影响. 西北农业学报, 1993, 2(3): 43-47.
	JIANG J Y, SU P. The effect of nutrition organs of winter wheat to the accumulation of N substance in seed during the fill period. Acta Agriculturae Boreali-occidentalis Sinica, 1993, 2(3): 43-47. (in Chinese)
[24]	LIPOWSKI A, LIPOWSKA D. Roulette-wheel selection via stochastic acceptance. Physica A: Statistical Mechanics and Its Applications, 2012, 391(6): 2193-2196. doi: 10.1016/j.physa.2011.12.004
[25]	鲁建厦, 翟文倩, 李嘉丰, 易文超, 汤洪涛. 基于改进混合蛙跳算法的多约束车辆路径优化. 浙江大学学报 (工学版), 2021, 55(2): 259-270.
	LU J X, ZHAI W Q, LI J F, YI W C, YANG H T. Multi-constrained vehicle routing optimization based on improved hybrid shuffled frog leaping algorithm. Journal of Zhejiang University (Engineering Science), 2021, 55(2): 259-270. (in Chinese)
[26]	CHAI T, DRAXLER R R. Root mean square error (RMSE) or mean absolute error (MAE)? - Arguments against avoiding RMSE in the literature. Geoscientific Model Development, 2014, 7(3): 1247-1250. doi: 10.5194/gmd-7-1247-2014
[27]	SHCHERBAKOV M V, BREBELS A, SHCHERBAKOVA N L, TYUKOV A. A survey of forecast error measures. World Applied Sciences Journal, 2013, 24(24): 171-176.
[28]	陶惠林, 徐良骥, 冯海宽, 杨贵军, 杨小冬, 苗梦珂, 代阳. 基于无人机数码影像的冬小麦株高和生物量估算. 农业工程学报, 2019, 35(19): 107-116.
	TAO H L, XU L J, FENG H K, YANG G J, YANG X D, MIAO M K, DAI Y. Estimation of plant height and biomass of winter wheat based on UAV digital image. Transactions of the Chinese Society of Agricultural Engineering, 2019, 35(19): 107-116. (in Chinese)
[29]	COLLINS B, NAJEEB U, LUO Q, TAN D K. Contribution of climate models and APSIM phenological parameters to uncertainties in spring wheat simulations: Application of SUFI-2 algorithm in northeast Australia. Journal of Agronomy and Crop Science, 2022, 208(2): 225-242. doi: 10.1111/jac.v208.2
[30]	彭慧文, 赵俊芳, 谢鸿飞, 房世波. 作物模型应用与遥感信息集成技术研究进展. 中国农业气象, 2022, 43(8): 644-656.
	PENG H W, ZHAO J F, XIE H F, FANG S B. Progresses of crop model application and its integration with remote sensing technology. Chinese Journal of Agrometeorology, 2022, 43(8): 644-656. (in Chinese)
[31]	靳霞菲, 陈先冠, 宫志宏, 冯利平. 小麦模型WheatSM参数调试优化方法研究. 农业大数据学报, 2021, 3(3): 13-22.
	JIN X F, CHEN X G, GONG Z H, FENG L P. Parameter adjustment and optimization methods for the WheatSM. Journal of Agricultural Big Data, 2021, 3(3): 13-22. (in Chinese)

参数 Parameter	土壤深度Soil depth (mm)
参数 Parameter	0-50	50-100	100-300	300-500	500-800	800-1100	1100-1400	1400-1700	1700-2000
容重Bulk density (g·cm^-3）	1.29	1.23	1.33	1.20	1.14	1.14	1.25	1.12	1.11
风干含水量Air-dried moisture (mm·mm^-1)	0.01	0.01	0.05	0.07	0.09	0.10	0.11	0.12	0.13
萎蔫系数Wilting coefficient (mm·mm^-1)	0.09	0.09	0.09	0.09	0.09	0.11	0.11	0.12	0.13
田间持水量Field capacity (mm·mm^-1)	0.27	0.27	0.27	0.27	0.26	0.27	0.26	0.26	0.26
饱和含水量Saturated moisture (mm·mm^-1)	0.46	0.49	0.45	0.50	0.52	0.52	0.48	0.53	0.53
有效水分下限Lower available moisture (mm·mm^-1)	0.09	0.09	0.09	0.09	0.10	0.12	0.13	0.18	0.22

名称 Name		单位 Unit	值 Value	来源 Reference
种植密度 Sowing density	SD	plant/hm²	400×10⁴	数据来源于研究区 Data from the study area
开花期茎干重 Stem dry weight at flowering	W_s	g	1.31	数据通过烘干法测得 Data measured by oven-drying method
开花至开始灌浆阶段茎的氮含量 Nitrogen concentration of stem parts from flowering to start of grain filling	C_{N_stem}	%	1.19	数据通过半微量凯氏定氮法测得 Data measured by semi-micro Kjeldahl method
开花至开始灌浆阶段叶片的氮含量 Nitrogen concentration of leaf parts from flowering to start of grain filling	C_{N_leaf}	%	2.66	数据通过半微量凯氏定氮法测得 Data measured by semi-micro Kjeldahl method
灌浆期茎的氮含量 Nitrogen concentration of stem parts during grain filling	C_{N_stem}	%	0.65	数据通过半微量凯氏定氮法测得 Data measured by semi-micro Kjeldahl method
灌浆期叶片的氮含量 Nitrogen concentration of leaf parts during grain filling	C_{N_leaf}	%	1.23	数据通过半微量凯氏定氮法测得 Data measured by semi-micro Kjeldahl method
籽粒含水量 Grain water content	C_w	%	20	数据来源于《中国西北春小麦》^[15]Data from Spring Wheat in Northwest China^[15]

图 Figure	函数表达式 Function expression	区间 Interval	函数类型 Function type
图1 Fig. 1	h_{grain_gfr_Tmean} =$\frac{1}{26}$T_mean	T_mean∈[0, 26)	单一变量的线性函数 Linear functions of a single variable
图1 Fig. 1	h_{grain_gfr_Tmean} = 1	T_mean∈[26, +∞)	单一变量的线性函数 Linear functions of a single variable
图2 Fig. 2	C_{N_crit_stem} = -0.00253s + 0.02485	s∈[6, 8]	单一变量的线性函数 Linear functions of a single variable
	C_{N_min_stem} = -0.00007s + 0.00327
	C_{N_crit_leaf} = -0.00344s + 0.03527
	C_{N_min_leaf} = -0.00254s + 0.02495
	C_{N_crit_stem} = -0.000715s + 0.01033	s∈(8, 9]	单一变量的线性函数 Linear functions of a single variable
	C_{N_min_stem} = -0.00024s + 0.00463
	C_{N_crit_leaf} = -0.0023s + 0.02615
	C_{N_min_leaf}= -0.00132s + 0.01519

名称 Name		wheat.xml中的定义 Definition in wheat.xml	默认值 Default value	下限 Lower bound	上限 Upper bound
小麦茎部分的每克籽粒数 Grain number per gram stem	R_g	grains_per_gram_stem	26.0	19.0	32.0
开花至开始灌浆阶段的潜在籽粒灌浆速率 Potential rate of grain filling from flowering to start of grain filling period (grain/d)	R_{grain_gfr}	potential_grain_growth_rate	0.00112	0.00077	0.00129
灌浆阶段的潜在籽粒灌浆速率 Potential rate of grain filling during grain filling period (grain/d)	R_{grain_gfr}	potential_grain_growth_rate	0.00249	0.00172	0.00346
氮限制下的潜在籽粒灌浆速率 Potential rate of grain filling under nitrogen limitation (g grain/d)	h_{N_poten}	potential_grain_n_filling_rate	6.70×10^-5	5.50×10^-5	8.60×10^-5
氮限制下的最小籽粒灌浆速率 Minimum rate of grain filling under nitrogen limitation (g grain/d)	h_{N_min}	minimum_grain_n_filling_rate	1.80×10^-5	1.50×10^-5	2.30×10^-5
单株小麦籽粒部分的最大干重值 Maximum dry weight of a single plant (g)	W_gm	max_grain_size	0.0437	0.0362	0.0519

名称 Name	值 Value	单位 Unit	wheat.xml中的定义 Definition in wheat.xml
灌浆期至成熟期积温 Thermal time required from start grain filling to maturity	580	℃	tt_startgf_to_mat
穗粒数 Number of grains produced by unit weight of stem	30.0	number/spike	grain_num_coeff
最大灌浆速率 Maximum grain filling rate of individual grains	2.30	mg grain/d	max_grain_fill_rate
单蘖重 Single tiller weight when elongation ceases	1.22	g	Dm_tilller_max
单株重 Plant weight	4.00	g	x_stem_wt
株高 Plant height for above weights	1.00	m	Y_height