工业园区大气污染物排放总量核算模型优化算法反演性能对比

赵瀚森; 刘峰均; 张夏夏; 吴柏莹; 沈怡秀; 高逸飞; 陈高

doi:10.12153/j.issn.1674-991X.20230688

工业园区大气污染物排放总量核算模型优化算法反演性能对比

doi: 10.12153/j.issn.1674-991X.20230688

赵瀚森^1,,
刘峰均¹,
张夏夏¹,
吴柏莹¹,
沈怡秀¹,
高逸飞¹,
陈高^{2, 3, ,}

1.
江苏省生态环境监测监控有限公司
2.
江苏省环保集团有限公司
3.
江苏省生态环境大数据有限公司

基金项目: 江苏省环保集团科研课题（1004，2022-1003）

详细信息

作者简介:
赵瀚森（1993—），男，工程师，主要从事生态环境大数据分析研究，zhaohs12@163.com

通讯作者:
陈高（1982—），男，高级工程师，主要从事环境信息化研究，cg@jshb.gov.cn

中图分类号: X831
计量
- 文章访问数: 100
- HTML全文浏览量: 31
- PDF下载量: 12
- 被引次数: 0
出版历程
- 收稿日期: 2023-09-22
- 录用日期: 2024-03-20
- 修回日期: 2024-03-12

Comparative study on inversion performance of optimization algorithms for the total emission accounting model of air pollutants in industrial parks

1.
Jiangsu Ecological Environment Monitoring Co., Ltd.
2.
Jiangsu Environmental Protection Group Co., Ltd.
3.
Jiangsu Ecological Environment Big Data Co., Ltd.

摘要

摘要:
基于工业园区污染物排放总量核算模型，以东南沿海某重点石化产业基地VOCs的排放总量核算为例，对比研究了Nelder-Mead单纯形法（NM）、双模退火优化算法（DA）、粒子群优化算法（PSO） 3种优化算法和通过偏差平方和（目标函数1）、对数变换（目标函数2）、双曲余弦变换（目标函数3）构建的3种优化目标函数在不同随机误差强度和失效站点数量下的反演性能。结果表明：PSO算法的反演性能与粒子数量有关，但整体表现不适用于当前条件下的反演优化，NM和DA优化算法具有较好的反演准确性（MARE<30%），但NM优化算法的反演计算效率约为DA优化算法的11~20倍，NM优化算法反演性能最好；3种优化目标函数均适用于当前条件下的反演优化（MARE<30%），其中目标函数1和目标函数3在随机误差强度、失效站点数量较小的情况下反演性能更好，而目标函数2在随机误差强度、失效站点数量相对较大的情况下反演性能更好。
- 工业园区 /
- 污染排放总量核算 /
- 优化算法 /
- 优化目标函数 /
- 源参数反演
Abstract:
Based on the total pollutant emission accounting model of industrial parks and taking the total VOCs emission accounting of a key petrochemical industry park on the southeast coast of China as an example, a comparative study of inversion performances was compared on three optimization algorithms, including Nelder-Mead simplex method (NM), dual annealing optimization algorithm (DA) and particle swarm optimization algorithm (PSO), under different random error intensity and number of failure sites. Besides, the inversion performance of three optimization objective functions constructed by deviation sum of squares (Objective Function 1), logarithm transformation (Objective Function 2) and hyperbolic cosine transformation (Objective Function 3) was compared. The results showed that the inversion performance of PSO was related to the number of particles, but the overall performance implied that it was not suitable for the inversion optimization under the current conditions. NM and DA had better inversion accuracy (MARE<30%), but the inversion computation efficiency of NM was about 11-20 times higher than that of DA, and NM had the best inversion performance. All the three optimization objective functions were suitable for inversion optimization under current conditions (MARE<30%). Objective Function 1 and Function 3 performed better when the random error intensity was small and the number of failure sites was small, while Objective Function 2 performed better when the random error intensity was relatively large and the number of failure sites was relatively large.
- industrial parks /
- total pollutant emission accounting /
- optimization algorithm /
- optimization objective function /
- source parameter inversion

HTML全文

图 1 不同随机误差下3种优化算法的反演性能

Figure 1. Inversion performance of three optimization algorithms under different random errors

下载: 全尺寸图片幻灯片

图 2 不同失效站点数量下3种优化算法的反演性能

Figure 2. Inversion performance of three optimization algorithms under different number of failure sites

下载: 全尺寸图片幻灯片

图 3 不同粒子数量下PSO算法的反演性能

Figure 3. Inversion performance of PSO algorithm under different particle numbers

下载: 全尺寸图片幻灯片

图 4 不同优化目标函数的反演性能

Figure 4. Inversion performance of different optimization objective functions

下载: 全尺寸图片幻灯片

图 5 3种优化目标函数逐小时反演结果

注：试验条件是随机误差标准差为0.1 mg/m³、失效站点数量为0个。

Figure 5. The hourly inversion results of three optimization objective functions

下载: 全尺寸图片幻灯片

表 1 网格化监测点位空间坐标

Table 1. Grid monitoring point spatial coordinates m　

序号	x	y	序号	x	y
1	5 114.71	2 309.58	39	1 868.87	−2 886.01
2	−3 711.36	−1 446.63	40	−2 768.47	−3 123.10
3	−5 718.63	2 649.74	41	291.19	−1 494.39
4	5 114.71	2 309.58	42	616.39	761.73
5	−3 711.36	−1 446.63	43	−1 011.41	−523.11
6	−2 044.57	4 741.98	44	1 388.01	293.15
7	2 340.18	−3 994.98	45	−471.63	−905.61
8	3 117.35	1 792.32	46	1 676.55	−317.70
9	3 379.21	1 253.75	47	1 315.32	−844.27
10	4 144.75	952.74	48	−1 358.04	−293.28
11	4 731.31	1 690.47	49	−1 530.78	720.19
12	4 652.83	629.75	50	854.63	−1 434.43
13	2 335.90	2 123.29	51	3 838.09	2 784.23
14	1 237.91	−2 085.05	52	2 754.05	3 284.45
15	1 350.45	2 700.39	53	1 090.81	4 168.37
16	953.59	2 967.68	54	350.59	3 246.90
17	1 425.35	1 393.37	55	−2 232.65	4 130.03
18	1 688.50	3 137.02	56	−3 300.88	2 805.40
19	2 434.63	784.04	57	−3 706.65	397.61
20	2 117.51	285.32	58	−3 545.05	−2 042.49
21	2 458.64	−332.70	59	−3 999.08	−3 386.99
22	3 817.20	162.85	60	−2 074.68	−4 127.62
23	3 249.42	−682.20	61	−363.58	−4 275.04
24	2 822.69	−1 365.94	62	1 214.23	−4 274.61
25	2 400.04	−2 032.34	63	3 401.67	−2 631.37
26	−2 195.04	−1 306.65	64	4 143.97	−1 345.64
27	−897.94	−2 315.75	65	5 070.37	335.06
28	−3 149.97	−1 418.27	66	−3 711.36	−1 446.63
29	−2 998.00	−3 246.11	67	−3 711.36	−1 446.63
30	−397.45	2 926.92	68	−3 711.36	−1 446.63
31	−4.51	2 678.01	69	5 114.71	2 309.58
32	−1 077.26	1 304.27	70	−3 711.36	−1 446.63
33	−2 610.37	−2 021.12	71	−5 006.01	−344.31
34	−1 535.36	−1 827.48	72	−3 883.77	−775.66
35	−1 430.18	−2 948.00	73	−3 954.11	−1 265.81
36	−1 725.05	−815.43	74	2 340.18	−3 994.98
37	−2 738.78	−841.12	75	2 340.18	−3 994.98
38	−267.43	−2 797.06	76	2 340.18	−3 994.98

下载: 导出CSV

表 2 污染源排放点位空间坐标及模拟排放量

Table 2. Spatial coordinates of pollution source discharge points and simulated emissions

企业	排口	x/m	y/m	VOCs模拟排放量/（t/a）
1	1-1	−72.60	1 901.08	305.68
	1-2	727.24	1 474.78	305.68
	1-3	1 113.57	2 088.37	305.68
2	2-1	2 224.66	1 497.08	52.7
3	3-1	−534.00	1 120.58	0
4	4-1	3 340.73	693.67	2.46
5	5-1	2 361.30	2 911.65	11.1
5	5-2	3 838.94	1 827.13	0
6	6-1	−2 460.93	−1 056.16	0.84
7	7-1	1 638.99	−3 599.91	0.47
8	8-1	−3 000.27	−2 545.48	0.14
9	9-1	−912.64	185.32	1 016.33
	9-2	133.95	−610.70	1 016.33
	9-3	875.99	−11.24	1 016.33
10	10-1	2 521.40	−2 652.02	0
10	10-2	3 269.09	−1 700.12	0

下载: 导出CSV

参考文献(25)

[1]	江苏省打好污染防治攻坚战指挥部办公室. 江苏省工业园区(集中区)污染物排放限值限量管理工作方案(试行)[A/OL]. (2021-07-19)[2023-09-15]. https://huanbao.bjx.com.cn/news/20210811/1169162.shtml.
[2]	杨喆, 李涛, 杨松柏, 等. 同时考虑监测和物料衡算的环境保护税税基核定方法[J]. 环境污染与防治,2019,41(5):608-610. YANG Z, LI T, YANG S B, et al. Verification method of environmental protection tax base based on minitoring and material balance[J]. Environmental Pollution & Control,2019,41(5):608-610.
[3]	高新伟, 李瑶. 火电企业大气污染物排放量核算方法研究[J]. 环境科学与管理,2022,47(12):70-75. doi: 10.3969/j.issn.1673-1212.2022.12.015 GAO X W, LI Y. Calculation method and application of air pollutant emissions in thermal power companies[J]. Environmental Science and Management,2022,47(12):70-75. doi: 10.3969/j.issn.1673-1212.2022.12.015
[4]	白璐, 乔琦, 张玥, 等. 工业污染源产排污核算模型及参数量化方法[J]. 环境科学研究,2021,34(9):2273-2284. BAI L, QIAO Q, ZHANG Y, et al. Pollutant generation and discharge accounting model and parameters quantification method in industrial sector[J]. Research of Environmental Sciences,2021,34(9):2273-2284.
[5]	胡瑞, 张学伟. 环境统计中污染物产生量排放量核算方法的探讨[J]. 科技视界,2012(34):115.
[6]	李贵林, 路学军, 陈程. 物料衡算法在工业源污染物排放量核算中的应用探讨[J]. 淮海工学院学报(自然科学版),2012,21(4):66-69. LI G L, LU X J, CHEN C. Application of material balance algorithm to the accounting of industrial sources pollution emissions[J]. Journal of Huaihai Institute of Technology (Natural Science Edition),2012,21(4):66-69.
[7]	平措. 大气污染扩散长期模型的应用研究[D]. 天津: 天津大学, 2006.
[8]	董吉开, 杜文莉, 王冰, 等. 湍流状态下化学品扩散溯源中不同目标函数的影响分析[J]. 化工学报,2020,71(3):1163-1173. DONG J K, DU W L, WANG B, et al. Investigating impacts of cost functions to atmospheric dispersion modeling and source term estimation in turbulent condition[J]. CIESC Journal,2020,71(3):1163-1173.
[9]	MAO S S, LANG J L, CHEN T, et al. Impacts of typical atmospheric dispersion schemes on source inversion[J]. Atmospheric Environment,2020,232:117572. doi: 10.1016/j.atmosenv.2020.117572
[10]	程桂香. 非线性最优化问题的一族新的罚函数方法研究[D]. 北京: 首都师范大学, 2006.
[11]	肖宏峰. 基于单纯形多向搜索的大规模进化优化算法[D]. 长沙: 中南大学, 2009.
[12]	MA D L, DENG J Q, ZHANG Z X. Comparison and improvements of optimization methods for gas emission source identification[J]. Atmospheric Environment,2013,81:188-198. doi: 10.1016/j.atmosenv.2013.09.012
[13]	胡峰, 郎建垒, 毛书帅, 等. 典型优化目标函数下源参数反演性能对比研究[J]. 中国环境科学,2021,41(5):2081-2089. doi: 10.3969/j.issn.1000-6923.2021.05.011 HU F, LANG J L, MAO S S, et al. Comparative study on source parameters inversion performance of typical cost functions[J]. China Environmental Science,2021,41(5):2081-2089. doi: 10.3969/j.issn.1000-6923.2021.05.011
[14]	毛书帅, 郎建垒, 陈添, 等. 多情景源排放参数反演下典型优化算法性能对比[J]. 北京工业大学学报,2020,46(4):369-376. doi: 10.11936/bjutxb2019100019 MAO S S, LANG J L, CHEN T, et al. Performance of typical optimization algorithms on inversing multi-scene source parameters[J]. Journal of Beijing University of Technology,2020,46(4):369-376. doi: 10.11936/bjutxb2019100019
[15]	马思琦. 高炉和二次污染物的对流扩散反问题[J]. 复旦学报(自然科学版),2009,48(2):231-237. MA S Q. Inverse convection-diffusion problems of plumes and second contamination[J]. Journal of Fudan University (Natural Science),2009,48(2):231-237.
[16]	殷凤兰, 李功胜, 贾现正. 一个多点源扩散方程的源强识别反问题[J]. 山东理工大学学报(自然科学版),2011,25(2):1-5. YIN F L, LI G S, JIA X Z. An inverse problem of determining magnitudes of multi-point sources in the diffusion equation[J]. Journal of Shandong University of Technology (Natural Science Edition),2011,25(2):1-5.
[17]	WANG J L, LIU J J, WANG B, et al. A new method for multi-point pollution source identification[J]. Atmospheric and Oceanic Science Letters,2021,14(6):100098. doi: 10.1016/j.aosl.2021.100098
[18]	张俊明, 袁鹏, 郭继香, 等. 基于MATLAB模拟的石化企业蒸馏装置泄漏扩散环境风险分析[J]. 环境工程技术学报,2013,3(3):259-265. doi: 10.3969/j.issn.1674-991X.2013.03.041 ZHANG J M, YUAN P, GUO J X, et al. Analysis of environmental risk of leakage and diffusion of distillation column in petrochemical enterprises based on MATLAB[J]. Journal of Environmental Engineering Technology,2013,3(3):259-265. doi: 10.3969/j.issn.1674-991X.2013.03.041
[19]	国家技术监督局, 国家环境保护局. 制定地方大气污染物排放标准的技术方法: GB/T 3840—91[S]. 北京: 中国标准出版社, 1991.
[20]	陈增强, 高艺博, 陈成功, 等. 基于差分进化-NM单纯形法的危化品泄漏源定位[J]. 中国安全生产科学技术,2022,18(5):90-95. CHEN Z Q, GAO Y B, CHEN C G, et al. Location for leakage source of hazardous chemicals based on differential evolution-NM simplex method[J]. Journal of Safety Science and Technology,2022,18(5):90-95.
[21]	蒋璐, 李奇, 陈维荣, 等. 基于Nelder-Mead优化的PEMFC三阶RQ等效电路参数辨识研究[J]. 电源学报,2019,17(2):12-18. JIANG L, LI Q, CHEN W R, et al. Research on parameter identification of third-order RQ equivalent circuit of PEMFC based on nelder-mead optimization[J]. Journal of Power Supply,2019,17(2):12-18.
[22]	YANG F W, CHEN J J, LIU Y C. Improved and optimized recurrent neural network based on PSO and its application in stock price prediction[J]. Soft Computing,2023,27(6):3461-3476. doi: 10.1007/s00500-021-06113-5
[23]	杨俊祺, 范晓军, 赵跃华, 等. 基于PSO-BP神经网络的山西省碳排放预测[J]. 环境工程技术学报,2023,13(6):2016-2024. doi: 10.12153/j.issn.1674-991X.20230190 YANG J Q, FAN X J, ZHAO Y H, et al. Prediction of carbon emissions in Shanxi Province based on PSO-BP neural network[J]. Journal of Environmental Engineering Technology,2023,13(6):2016-2024. doi: 10.12153/j.issn.1674-991X.20230190
[24]	苗建杰, 李德波, 李慧君, 等. 改进粒子群算法在燃煤电厂中的应用现状及展望[J]. 环境工程,2023,41(增刊1):354-362.
[25]	甘秋云, 李兢思, 杨佳翰. 冷却调度参数对模拟退火算法性能的影响分析[J]. 蚌埠学院学报,2023,12(5):47-52. doi: 10.3969/j.issn.2095-297X.2023.05.009 GAN Q Y, LI J S, YANG J H. Analysis of cooling scheduling parameters on performance of simulated annealing algorithm[J]. Journal of Bengbu University,2023,12(5):47-52. □ doi: 10.3969/j.issn.2095-297X.2023.05.009