基于IGWOPSO-SVM算法的水质监测及等级评定系统

蒋冬; 肖茂华; 张海军; 周俊博; 朱虹; 汪小旵; 陈爽

doi:10.7671/j.issn.1001-411X.202207034

基于IGWOPSO-SVM算法的水质监测及等级评定系统

蒋冬^1,,
肖茂华^1, ,,
张海军¹,
周俊博¹,
朱虹²,
汪小旵¹,
陈爽³

1.
南京农业大学工学院, 江苏南京 210031
2.
江苏省农机具开发应用中心, 江苏南京 210017
3.
江苏双木测控技术有限公司, 江苏镇江 212300

基金项目: 江苏省科技计划项目(BE2022385，BZ2021024)；江苏省现代农机装备与技术示范推广项目(NJ2021-03)；镇江市重点研发项目(NY2021018)；丹阳市重点研发项目(SNY202105)

详细信息

作者简介:
蒋　冬，硕士研究生，主要从事水产养殖装备研究，E-mail: 865551390@qq.com

通讯作者:
肖茂华，教授，博士，主要从事智能农机装备研究，E-mail: xiaomaohua@njau.edu.cn

中图分类号: S96；TP27
计量
- 文章访问数: 103
- HTML全文浏览量: 9
- PDF下载量: 16
出版历程
- 收稿日期: 2022-07-17
- 网络出版日期: 2023-09-03
- 发布日期: 2023-05-10
- 刊出日期: 2023-07-09

Water quality monitoring and grade judgment system based on IGWOPSO-SVM algorithm

1.
College of Engineering, Nanjing Agricultural University, Nanjing 210031, China
2.
Jiangsu Agricultural Machinery Test and Appraisal Station, Nanjing 210017, China
3.
Jiangsu Shuangmu Measurement and Control Technology Co., Ltd., Zhenjiang 212300, China

摘要

摘要:
目的
水污染监测是流域水污染防治工作的前提。为实现高精度的地表水水质监测及水体等级评定，本研究设计基于IGWOPSO-SVM(Improved grey wolf optimizer particle swarm optimization-support vector machine)模型的水质监测及等级评定系统。
方法
选用传感器组、STM32F103单片机、ESP8266WIFI无线通信模块搭建水质监测系统数据处理模块，利用WIFI无线通信将数据处理模块采集到的水质数据传输至服务器，设计水质监测系统服务器交互端，同时开发水质监测小程序对水质等级进行实时监测。基于改进粒子群算法(Improved particle swarm optimization，IPSO)及灰狼算法(Grey wolf optimizer，GWO)提出了IGWOPSO算法，对SVM进行优化，据此提出了IGWOPSO-SVM水质等级评定算法。基于南京市玄武湖、金川河、江浦水源地135组水质数据对本系统水质等级评定效果进行试验验证。
结果
相比于SVM，IGWOPSO-SVM水质等级评定算法的总样本分类准确率由86.67%上升至100.00%，上升了13.33个百分点；相比于粒子群算法(Particle swarm optimization，PSO)，IGWOPSO算法的最佳适应度由86.80上升至99.20，提高了14.29%。
结论
本研究解决了传统水体等级评定方法效率低、准确率低的问题，为地表水水质的精确监测提供了方法借鉴。
- 水质监测系统 /
- 支持向量机 /
- 粒子群算法 /
- 灰狼算法 /
- 改进粒子群算法 /
- IGWOPSO-SVM
Abstract:
Objective
Water pollution monitoring is a prerequisite for water pollution prevention and control in watersheds. In order to achieve high accuracy of surface water quality monitoring and water body rating judgement, we designed a water quality monitoring and rating system based on IGWOPSO-SVM (Improved grey wolf optimizer particle swarm optimization-support vector machine) model.
Method
We selected sensor group, STM32F103 microcontroller, ESP8266WIFI wireless communication module to build a water quality monitoring system data processing module. The WIFI wireless communication transmitted the water quality data collected by data processing module to the server. We designed water quality monitoring system server interactive end, while developing water quality monitoring applet for real-time monitoring of water quality grade. Based on the improved particle swarm optimization (IPSO) and grey wolf optimizer (GWO), the IGWOPSO algorithm was proposed to optimize SVM algorithm, according to which the IGWOPSO-SVM water quality rating algorithm was proposed. The water quality rating effect of this system was verified by experiment based on 135 groups water quality data of Nanjing Xuanwu Lake, Jinchuan River and Jiangpu water source.
Result
Compared with SVM, the total sample classification accuracy of IGWOPSO-SVM water quality rating algorithm increased from 86.67% to 100.00%, with an increase of 13.33 percent. Compared with particle swarm optimization (PSO), the best adaptation degree of IGWOPSO algorithm increased from 86.80 to 99.20, with an increase of 14.29%.
Conclusion
This study solves the problems of low efficiency and low accuracy of traditional water body rating methods, and provides a method reference for accurate monitoring of surface water quality.
- Water quality monitoring system /
- Support vector machine /
- Particle swarm optimization /
- Grey wolf optimizer /
- Improved particle swarm optimization /
- IGWOPSO-SVM

HTML全文

图 1 水质监测系统总体结构图

Figure 1. General structure of water quality monitoring system

下载: 全尺寸图片幻灯片

图 2 硬件布置方案

Figure 2. Hardware arrangement scheme

下载: 全尺寸图片幻灯片

图 3 数据采集节点软件设计的流程图

Figure 3. Flow chart of software design in data acquisition node

下载: 全尺寸图片幻灯片

图 4 WIFI无线通信模块软件设计的流程图

Figure 4. Flow chart of software design in WIFI wireless communication module

下载: 全尺寸图片幻灯片

图 5 服务器交互端设计流程图

Figure 5. Flow chart of server interactive end design

下载: 全尺寸图片幻灯片

图 6 水质监测程序界面

Figure 6. Interface of water quality monitoring procedure

下载: 全尺寸图片幻灯片

图 7 IGWOPSO-SVM模型算法流程

Figure 7. Flow chart of IGWOPSO-SVM model algorithm

下载: 全尺寸图片幻灯片

图 8 水质数据采样点

Figure 8. Sampling points of water quality data

下载: 全尺寸图片幻灯片

图 9 不同算法的分类结果

Figure 9. Classification results of different algorithms

下载: 全尺寸图片幻灯片

图 10 不同优化算法的适应度曲线

Figure 10. Fitness curve of different optimized algorithms

下载: 全尺寸图片幻灯片

表 1 地表水各指标不同等级的标准质量浓度

Table 1 Standard mass concentrations of different grades for each index in surface water ρ/(mg·L⁻¹)

等级 Grade	溶解氧 Dissolved oxygen	化学需氧量 Chemical oxygen demand	氨氮 NH₃-N	总磷 Total phosphorus	总氮 Total nitrogen	硫化物 Sulfide
Ⅰ	≥7.5	≤15	≤0.15	≤0.02	≤0.2	≤0.05
Ⅱ	≥6.0	≤15	≤0.50	≤0.10	≤0.5	≤0.10
Ⅲ	≥5.0	≤20	≤1.00	≤0.20	≤1.0	≤0.20
Ⅳ	≥3.0	≤30	≤1.50	≤0.30	≤1.5	≤0.50
Ⅴ	≥2.0	≤40	≤2.00	≤0.40	≤2.0	≤1.00

下载: 导出CSV

表 2 训练样本生成规则

Table 2 Generation rules of training samples

pH	ρ/(mg·L⁻¹)						样本数 Number of sample	期望输出 Expected output	等级 Grade
pH	溶解氧 Dissolved oxygen	化学需氧量 Chemical oxygen demand	氨氮 NH₃-N	总磷 Total phosphorus	总氮 Total nitrogen	硫化物 Sulfide	样本数 Number of sample	期望输出 Expected output	等级 Grade
6~9	≥7.5	≤15	≤0.15	≤0.02	≤0.2	≤0.05	50	1	Ⅰ
	6.0~7.5	≤15	0.15~0.50	0.02~0.10	0.2~0.5	0.05~0.10	50	2	Ⅱ
	5.0~6.0	15~20	0.50~1.00	0.10~0.20	0.5~1.0	0.10~0.20	50	3	Ⅲ
	3.0~5.0	20~30	1.00~1.50	0.20~0.30	1.0~1.5	0.20~0.50	50	4	Ⅳ
	2.0~3.0	30~40	1.50~2.00	0.30~0.40	1.5~2.0	0.50~1.00	50	5	Ⅴ

下载: 导出CSV

表 4 不同优化算法的最优解

Table 4 Optimal solutions of different optimized algorithms

优化算法 Optimized algorithm	核函数参数(c) Kernel function parameter	惩罚因子(g) Penalty factor
PSO	9.783 4	5.023 7
GWO	42.362 5	3.816 4
IPSO	82.436 5	0.659 1
GWOPSO	65.899 2	0.010 0
IGWOPSO	97.234 2	0.010 0

下载: 导出CSV

表 3 不同算法的分类准确率及性能

Table 3 Classification accuracy and performance of different algorithms

算法 Algorithm	样本准确率/% Sample accuracy				性能 Performance
算法 Algorithm	S₁	S₂	S₃	均值 Mean	绝对误差 Abosolute error	均方根误差 Root mean square error	纳什效率系数 Nash-Sutcliffe efficiency
SVM	100.00	75.56	84.44	86.67	23	0.49	0.63
PSO-SVM	100.00	86.67	86.67	91.11	15	0.39	0.77
GWO-SVM	100.00	88.89	88.89	92.59	10	0.27	0.89
IPSO-SVM	91.11	100.00	95.56	95.56	6	0.21	0.93
GWOPSO-SVM	100.00	100.00	93.33	97.78	3	0.15	0.97
IGWOPSO-SVM	100.00	100.00	100.00	100.00	0	0	1.00

下载: 导出CSV

参考文献(34)

[1]	YAROSHENKO I, KIRSANOV D, MARJANOVIC M, et al. Real-time water quality monitoring with chemical sensors[J]. Sensors, 2020, 20(12): 3432. doi: 10.3390/s20123432.
[2]	IMRAN L B, LATIF R M A, FARHAN M, et al. Smart city based autonomous water quality monitoring system using WSN[J]. Wireless Personal Communications, 2020, 115(2): 1805-1820. doi: 10.1007/s11277-020-07655-x
[3]	AMOATEY P, BAAWAIN M S. Effects of pollution on freshwater aquatic organisms[J]. Water Environment Research, 2019, 91(10): 1272-1287. doi: 10.1002/wer.1221
[4]	徐显阳. 多层水质数据监测系统的采集与控制研究[D]. 大连: 大连理工大学, 2021.
[5]	曹守启, 葛照瑞, 张铮. 基于物联网的浮标水质监测系统与溶解氧浓度预测模型[J]. 农业机械学报, 2021, 52(11): 210-218.
[6]	肖茂华, 周俊博, 王伟臣, 等. 一种新型拖拉机故障诊断系统及故障诊断方法: CN202110892020.4[P]. 2021-10-29.
[7]	SILLBERG C V, KULLAVANIJAYA P, CHAVALPARIT O. Water quality classification by integration of attribute-realization and support vector machine for the Chao Phraya River[J]. Journal of Ecological Engineering, 2021, 22(9): 70-86. doi: 10.12911/22998993/141364
[8]	杨望灿, 张培林, 任国全, 等. 基于模糊熵与LS-SVM的轴承故障诊断[J]. 机械强度, 2014, 36(5): 666-670.
[9]	周金. 基于自适应VMD和PSO-SVM的电能质量扰动检测与分类[D]. 长沙: 湖南大学, 2019.
[10]	孙俊, 莫云南, 戴春霞, 等. 基于介电特性与IRIV-GWO-SVR算法的番茄叶片含水率检测[J]. 农业工程学报, 2018, 34(14): 188-195.
[11]	SHAHEEN M A M, HASANIEN H M, ALKUHAYLI A. A novel hybrid GWO-PSO optimization technique for optimal reactive power dispatch problem solution[J]. Ain Shams Engineering Journal, 2021, 12(1): 621-630. doi: 10.1016/j.asej.2020.07.011
[12]	MATOS J, FARIA R P V, NOGUEIRA I B R, et al. Optimization strategies for chiral separation by true moving bed chromatography using particles swarm optimization (PSO) and new Parallel PSO variant[J]. Computers & Chemical Engineering, 2019, 123: 344-356.
[13]	郜云波. 纤维制备生产参数与质量能耗评估预测模型及应用研究[D]. 哈尔滨: 东北林业大学, 2019.
[14]	GOHIL B N, PATEL D R. A hybrid GWO-PSO algorithm for load balancing in cloud computing environment[C]//2018 Second International Conference on Green Computing and Internet of Things, 2018: 185-191.
[15]	刘双印, 徐龙琴, 李振波, 等. 基于PCA-MCAFA-LSSVM的养殖水质pH值预测模型[J]. 农业机械学报, 2014, 45(5): 239-246.
[16]	陈俊英, 邢正, 张智韬, 等. 基于高光谱定量反演模型的污水综合水质评价[J]. 农业机械学报, 2019, 50(11): 200-209.
[17]	李鑫星, 王聪, 田野, 等. 基于ZigBee的多参数水质在线监测系统[J]. 农业机械学报, 2015, 46(S1): 168-173.
[18]	江先亮, 尚子宁, 金光, 等. 基于无人船的水产养殖水质动态监测系统设计与实验[J]. 农业机械学报, 2020, 46(9): 175-185.
[19]	程淑红, 刘洁, 朱丹丹. 基于计算机视觉与SVM的水质异常监测方法[J]. 光电工程, 2014, 41(5): 28-33.
[20]	李昕, 陈泽君, 李立君, 等. 基于偏好免疫网络和SVM算法的油茶果多特征识别[J]. 农业工程学报, 2020, 36(22): 205-213. doi: 10.11975/j.issn.1002-6819.2020.22.023
[21]	周建民, 王发令, 张臣臣, 等. 基于特征优选和GA-SVM的滚动轴承智能评估方法[J]. 振动与冲击, 2021, 40(4): 227-234.
[22]	LIU W, GUO G, CHEN F, et al. Meteorological pattern analysis assisted daily PM2.5 grades prediction using SVM optimized by PSO algorithm[J]. Atmospheric Pollution Research, 2019, 10(5): 1482-1491. doi: 10.1016/j.apr.2019.04.005
[23]	杨柳, 孙金华, 冯仲科, 等. 基于PSO-LSSVM的森林地上生物量估测模型[J]. 农业机械学报, 2016, 47(8): 273-279.
[24]	周俊博, 朱烨均, 肖茂华, 等. 基于LWD-QPSO-SOMBP神经网络的拖拉机柴油机故障诊断[J]. 农业工程学报, 2021, 37(17): 39-48.
[25]	苏中滨, 张磊磊, 马铮, 等. 基于改进PSO的无人自转旋翼机二自由度PID飞行控制[J]. 农业机械学报, 2022, 53(1): 151-158.
[26]	周强, 周永权. 一种基于领导者策略的狼群搜索算法[J]. 计算机应用研究, 2013, 30(9): 2629-2632.
[27]	TATSUMI K, IBUKI T, TANINO T. A chaotic particle swarm optimization exploiting a virtual quartic objective function based on the personal and global best solutions[J]. Applied Mathematics and Computation, 2013, 219(17): 8991-9011. doi: 10.1016/j.amc.2013.03.029
[28]	曹旺. MATLAB智能优化算法: 从写代码到算法思想[M]. 北京: 北京大学出版社, 2021.
[29]	MIRJALILI S, MIRJALILI S M, LEWIS A. Grey wolf optimizer[J]. Advances in Engineering Software, 2014, 69: 46-61. doi: 10.1016/j.advengsoft.2013.12.007
[30]	国家环境保护总局, 国家质量监督检验检疫总局. 地表水环境质量标准: GB 3838—2002[S]. 北京: 中国标准出版社, 2002.
[31]	ZHOU Q, ZHOU Y. Wolf colony search algorithm based on leader strategy[J]. Application Research of Computers, 2013, 30(9): 2629-2632.
[32]	刘冬, 李帅, 付强, 等. 基于KHA优化BP神经网络的地下水水质综合评价方法[J]. 农业机械学报, 2018, 49(9): 275-284.
[33]	成玉宁, 王雪原. 拟自然化: 城市湖泊水环境治理的生态智慧与途径: 以南京玄武湖为例[J]. 中国园林, 2021, 37(7): 19-24.
[34]	南京市生态环境局. 2022年5月南京市区县级以上集中式生活饮用水水源水质状况报告[EB/OL]. http://www.nanjing.gov.cn/zdgk/202205/t20220526_3428993.html.