
基于改进GM-LSSVR模型的郑州市用水量预测
李彦彬, 闫文晶, 张海涛, 杜军凯
基于改进GM-LSSVR模型的郑州市用水量预测
Forecast of Water Consumption in Zhengzhou City Based on Improved GM-LSSVR Model
准确的用水量预测是响应国家高质量发展的重要抓手,也是城市水资源优化配置的基础。针对用水量序列存在波动性、灰色模型与所需因素存在线性关系等问题,提出了一种基于HP滤波分解的GM-LSSVR预测模型,即先采用灰色关联分析法筛选合适的用水量影响因素,再利用HP滤波分解法将筛选的用水量及影响因素分解为长期趋势序列和短期波动序列,最后通过GM-LSSVR组合模型来预测用水量。以郑州市为例,使用该模型预测2001-2019年用水量,并与GM(1,N)模型、BP神经网络模型的预测结果进行对比。结果表明,基于HP滤波分解的GM-LSSVR预测模型预测精度大大提高,具有可行性与实用性,可以较好地应用于城市用水量预测研究中。
Accurate water consumption prediction is an important starting point for responding to the country’s high-quality development, and it is also the basis for the optimal allocation of urban water resources. In view of the volatility of the water consumption sequence and the linear relationship between the gray model and the required factors, this paper proposes a GM-LSSVR prediction model based on HP filter decomposition, that is, the gray correlation analysis method is first used to screen suitable water consumption influencing factors. And then the HP filter decomposition method is used to decompose the selected water consumption and influencing factors into a long-term trend sequence and a short-term fluctuation sequence, and finally the GM-LSSVR combined model is used to predict water consumption. Taking Zhengzhou City as an example, the model is used to predict water consumption from 2001 to 2019, and compared with the prediction results of the GM(1,N) model and the BP neural network model. The results show that the prediction accuracy of the GM-LSSVR prediction model based on HP filter decomposition is greatly improved, and it is feasible and practical, and can be better applied to the study of urban water consumption prediction.
灰色关联分析 / HP滤波分解 / GM-LSSVR模型 / 用水量预测 {{custom_keyword}} /
grey correlation analysis / HP filter decomposition / GM-LSSVR model / water consumption forecast {{custom_keyword}} /
表1 2012-2019年郑州总用水量统计 (亿m3)Tab.1 Statistics of total water consumption in Zhengzhou from 2012 to 2019 |
年份 | 农业用水 | 工业用水 | 生活用水 | 生态用水 | 总用水量 |
---|---|---|---|---|---|
2011 | 4.43 | 6.51 | 4.79 | 1.96 | 17.69 |
2012 | 4.44 | 6.93 | 5.04 | 2.05 | 18.46 |
2013 | 4.78 | 5.77 | 4.95 | 2.13 | 17.63 |
2014 | 5.01 | 5.40 | 5.39 | 2.04 | 17.84 |
2015 | 5.10 | 5.46 | 5.64 | 2.01 | 18.22 |
2016 | 5.50 | 5.47 | 5.88 | 2.70 | 19.55 |
2017 | 4.71 | 5.45 | 6.43 | 3.94 | 20.53 |
2018 | 4.23 | 5.27 | 6.60 | 4.61 | 20.71 |
2019 | 4.24 | 4.99 | 7.30 | 5.13 | 21.66 |
表2 用水量影响因子灰色关联度Tab.2 Grey correlation degree of water consumption influencing factor |
影响因素 | 灰色关联度r | 影响因素 | 灰色关联度r |
---|---|---|---|
总人口/万人 | 0.925 | 污水处理率/% | 0.785 |
平均气温/℃ | 0.909 | 粮食总产值/万t | 0.782 |
绿化覆盖率/% | 0.878 | 工业用水重复利用率/% | 0.781 |
建成区面积/km2 | 0.825 | 人均生产总值/元 | 0.769 |
表3 用水量预测结果 (亿m3)Tab.3 Water consumption forecast results |
年份 | 实际值 | 模拟值 | 相对误差/% | |
---|---|---|---|---|
训练样本 | 2011 | 17.69 | 17.66 | 0.002 |
2012 | 18.46 | 18.01 | 0.024 | |
2013 | 17.63 | 17.96 | 0.019 | |
2014 | 17.84 | 17.94 | 0.006 | |
2015 | 18.22 | 18.32 | 0.006 | |
2016 | 19.55 | 19.57 | 0.001 | |
2017 | 20.53 | 20.4 | 0.006 | |
平均相对误差/% | - | - | 0.009 | |
测试样本 | 2018 | 20.71 | 20.69 | 0.001 |
2019 | 21.66 | 21.663 | 0 | |
平均相对误差/% | - | - | 0.001 |
表4 3种模型预测结果 (亿m3)Tab.4 The prediction results of the three models |
年份 | 实际值 | 灰色GM(1,N)模型 | BP神经网络模型 | GM-LSSVR模型 | ||||
---|---|---|---|---|---|---|---|---|
模拟值 | 相对误差 | 模拟值 | 相对误差 | 模拟值 | 相对误差 | |||
训练样本 | 2011 | 17.69 | 17.22 | 0.027 | 17.64 | 0.003 | 17.66 | 0.002 |
2012 | 18.46 | 17.29 | 0.063 | 17.82 | 0.035 | 18.01 | 0.024 | |
2013 | 17.63 | 18.34 | 0.040 | 17.16 | 0.027 | 17.96 | 0.019 | |
2014 | 17.84 | 18.65 | 0.046 | 18.13 | 0.016 | 17.94 | 0.006 | |
2015 | 18.22 | 19.09 | 0.048 | 18.22 | 0 | 18.32 | 0.006 | |
2016 | 19.55 | 19.60 | 0.002 | 19.26 | 0.015 | 19.57 | 0.001 | |
2017 | 20.53 | 20.13 | 0.020 | 20.81 | 0.014 | 20.40 | 0.006 | |
平均相对误差 | - | - | 0.035 | - | 0.016 | - | 0.009 | |
测试样本 | 2018 | 20.71 | 20.67 | 0.002 | 20.72 | 0 | 20.69 | 0.001 |
2019 | 21.66 | 21.233 | 0.020 | 21.36 | 0.014 | 21.663 | 0 | |
平均相对误差 | - | - | 0.011 | - | 0.007 | - | 0.001 |
1 |
高磊,刘英杰.构建节水型城市 确保城市可持续发展:以北京市为例[J].水利科技与经济,2011,17(12):37-39.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
2 |
徐伟铭,鱼京善,王崴,等.基于神经网络模型的全国用水量“四维”模拟[J].南水北调与水利科技,2020,18(1):11-17.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
3 |
常浩娟,刘卫国,吴琼.玛纳斯河流域用水分析、需水预测和影响因素[J].节水灌溉,2017(7):88-93.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
4 |
侯薇. 基于GIS的关中地区水土资源优化配置研究[D].陕西咸阳:西北农林科技大学,2012.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
5 |
高学平,陈玲玲,刘殷竹,等.基于PCA-RBF神经网络模型的城市用水量预测[J].水利水电技术,2017,48(7):1-6.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
6 |
张陈俊,章恒全,陈其勇,等.中国用水量变化的影响因素分析:基于LMDI方法[J].资源科学,2016,38(7):1 308-1 322.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
7 |
刘裕辉,王雪茹,王志雄,等.吉安市用水结构演变驱动因子分析及用水预测[J].江西水利科技,2019,45(5):376-381.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
8 |
朱世垚,宋松柏,王小军,等.基于LMDI和STIRPAT模型的区域用水影响因素定量分析研究[J].水利水电技术(中英文),2021,52(2):30-39.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
9 |
邓权龙,蒋仲安,韩硕.基于小波分析与GM(1,1)-ARMA(p,q)组合的矿井防尘用水量预测[J].矿业安全与环保,2018,45(4):75-79.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
10 |
崔远来,张新,罗玉峰,等.稻田回归水模拟及其评价[J]. 灌溉排水学报, 2005(5):1-4.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
11 |
杨雪菲,粟晓玲,马黎华.基于人工神经网络模型的关中地区用水量的预测[J].节水灌溉,2009(8):4-6.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
12 |
杜懿,麻荣永.不同改进灰色模型在广西年用水量预测中的应用研究[J].水资源与水工程学报,2017,28(3):87-90.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
13 |
孙丽芹,常安定,位龙虎.基于AM残差修正的GM(1,1)模型的用水量预测[J].统计与决策,2016(17):79-82.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
14 |
白鹏,龙秋波.3种用水量预测方法在京津冀地区的适用性比较[J].水资源保护,2021,37(2):102-107.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
15 |
秦长海,孙素艳,张小娟.宁夏社会经济及生态需水量预测[J].水资源保护,2008(5):24-29.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
16 |
甘月云,陈星,付军,等.基于灰色预测GM(1,1)模型的城镇生活需水量预测[J].水电能源科学,2012,30(9):40-42.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
17 |
陈金翠. 最小二乘支持向量回归组合预测模型的应用[D].乌鲁木齐:新疆大学, 2011.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
18 |
冯天梅,张鑫.基于修正组合模型的包头市用水量预测分析[J].西北农林科技大学学报(自然科学版),2014,42(3):226-234.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
19 |
刘思峰,蔡华,杨英杰,等.灰色关联分析模型研究进展[J].系统工程理论与实践,2013,33(8):2 041-2 046.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
20 |
李国祥,李永清,马天骄.基于HP滤波法的我国CPI波动规律研究[J].经济问题,2017(10):60-65,125.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
21 |
曾亮.改进的灰色多变量GM(1,N)模型及其应用[J].西南大学学报(自然科学版),2019,41(9):68-76.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
22 |
温冬琴,王建东,张霞.基于GM-LSSVR机场噪声时间序列预测模型[J].计算机科学,2013,40(9):198-200,220.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
23 |
黄德治.变化环境下城市需水量影响因子识别[J].中国农村水利水电,2012(3):140-145.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
24 |
刘怀丕,李文哲.郑州:引领中原城市群发展[J].宁波经济(财经视点),2019(3):18-19.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
25 |
河南省人民政府. 高温之下郑州日供水量“节节攀升” 135.9万m3创历史之最[EB/OL]. 2017-07-11
{{custom_citation.content}}
{{custom_citation.annotation}}
|
26 |
赵华.郑州市城区地下水漏斗现状分析[J].河南水利与南水北调,2010(12):45-46.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
27 |
郑州发布. 市委关于制定郑州十四五规划和2035年远景目标的建议(全文):中共郑州市委关于制定郑州市国民经济和社会发展第十四个五年规划和二〇三五年远景目标的建议[EB/OL]. 2021-01-11.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
28 |
张振杰,郝志萍,侯煜堃.郑州市自来水“一户一表”改造的探索与思考[J].净水技术,2017,36(10):114-118.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_ref.label}} |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
/
〈 |
|
〉 |