
1.云南省红河州水利水电勘察设计研究院,云南 红河 661100
2.云南省水文水资源局红河分局,云南 红河 661100
3.云南省文山州水务局,云南 文山 663000
Received:05 October 2025,
Revised:2025-12-12,
Accepted:19 December 2025,
Published Online:06 January 2026,
移动端阅览
饶庆阳,杨琼波,崔东文.基于WPT二次分解与CPO优化的KAN地下水位预测模型[J].人民珠江,DOI:10.3969/j.issn.1001-9235.XXXX.XX.001.
RAO Qingyang,YANG Qiongbo,CUI Dongwen.KAN groundwater level prediction model based on WPT secondary decomposition and CPO[J].PEARL RIVER,
饶庆阳,杨琼波,崔东文.基于WPT二次分解与CPO优化的KAN地下水位预测模型[J].人民珠江,DOI:10.3969/j.issn.1001-9235.XXXX.XX.001. DOI:
RAO Qingyang,YANG Qiongbo,CUI Dongwen.KAN groundwater level prediction model based on WPT secondary decomposition and CPO[J].PEARL RIVER, DOI:10.3969/j.issn.1001-9235...001.
针对Kolmogorov-Arnold网络(KAN)中数据处理过拟合、时序建模弱以及超参数选取困难等问题,提出一种基于小波包变换(Wavelet Packet Transform,WPT)二次分解和中华穿山甲优化(Chinese Pangolin Optimizer,CPO)算法寻优KAN超参数的地下水水位预测模型,并构建WPT-CPO-Transformer等7种对比分析模型,通过云南省西城、文澜、临安、草坝站日均地下水位时间序列预测实例对8种模型进行验证。首先利用WPT二次分解技术对地下水位时序数据进行分解处理,划分训练集和验证集;然后利用CPO寻优KAN超参数,以克服人工调试繁琐低效、避免局部最优等问题;最后利用最佳超参数建立WPT-CPO-KAN模型对实例地下水位时间序列各分解分量进行训练、预测和重构。结果表明:①WPT-CPO-KAN模型相较于WPT-CPO-Transformer、WPT-CPO-LSTM、WPT-CPO-GRU、WPT-CPO-XGBoost、WPT-CPO-LSSVM、WPT-CPO-MLP、WPT-KAN模型,预测精度分别提高了15.6%、37.4%、26.5%、36.4%、18.6%、7.2%、26.7%以上(MAPE指标),具有更小的预测误差和较好的普适性;②)KAN能更好地捕捉地下水位时序数据中复杂的非线性空间-时间依赖性,更适应地下水位时序数据分布,性能优于Transformer、LSTM(Long Short-Term Memory)、GRU(Gated Recurrent Unit)、XGBoost模型和传统LSSVM(Least Squares Support Vector Machine)、MLP(Multilayer Perceptron)网络;③WPT-CPO-KAN模型预测误差随着预测步长的增加而增大,在3 d以内,WPT-CPO-KAN模型预测精度较高;④通过CPO寻优KAN超参数,显著提高KAN性能和预测自动化水平。优化方法可为改进KAN性能等研究提供参考。
To improve over-fitting of data processing
weak time series modeling
and difficult selection of hyperparameters in the Kolmogorov-Arnold network (Kan)
a groundwater level prediction model based on wavelet packet transform (WPT) secondary decomposition and Chinese Pangolin optimizer (CPO) algorithm was proposed to optimize KAN hyperparameters
and WPT-CPO-Transformer
WPT-CPO-LSTM
WPT-CPO-gated circulation unit (GRU)
WPT-CPO-least squares support vector machine (LSSVM)
WPT-CPO-extreme gradient ascent machine (XGBoost)
WPT-CPO-MLP
and WPT-KAN were constructed. These seven kinds of comparative analysis models were verified by the daily average groundwater level time series prediction examples of Xicheng
Wenlan
Lin'an
and Caoba stations in Yunnan Province. Firstly
the WPT secondary decomposition technology was used to decompose the groundwater level time series data and divide the training set and the verification set. Then
the CPO was used to optimize the hyperparameters of KAN to overcome the tedious and inefficient manual debugging and avoid local optimization. Finally
the WPT-CPO-KAN model was established by using the optimal hyperparameters to train
predict
and reconstruct the decomposed components of the groundwater level time series. The results show that: (1) compared with that of the WPT-CPO-Transformer
WPT-CPO-LSTM
WPT-CPO-GRU
WPT-CPO-XGBoost
WPT-CPO-LSSVM
WPT-CPO-MLP
and WPT-KAN models
the prediction accuracy of the WPT-CPO-KAN model is improved by 15.6%
37.4%
26.5%
36.4%
18.6%
7.2%
and 26.7%
respectively (MAPE index)
which has a smaller prediction error and better universality. (2) Under the same WPT secondary decomposition and CPO
KAN can better capture the complex nonlinear space and time dependence in groundwater level time series data and is more suitable for the distribution of groundwater level time series data. Its performance is better than that of the transformer
LSTM
GRU
XGBoost models
traditional LSSVM
and MLP network. (3) The prediction error of the WPT-CPO-KAN model increases with the increase in the prediction step. Within three days
the prediction accuracy of the WPT-CPO-KAN model is higher. (4) The reasonable selection of hyperparameters is of great significance to improve the performance of the KAN model. By using CPO to optimize KAN hyperparameters
the performance of KAN and the level of prediction automation are significantly improved. The optimization method can provide a reference for improving the performance of KAN. (5) KAN can reveal the variation characteristics of groundwater level time series data with fewer parameters
thus enhancing the interpretability of the WPT-CPO-KAN model.
田宇 , 崔东文 , 毛宗波 , 等 . 基于数据分解与十种“植物”算法优化的RELM地下水位预测 [J]. 水利水电技术(中英文) , 2025 , 56 ( 9 ): 118 - 130 .
TIAN Y , CUI D W , MAO Z B , et al . Groundwater level prediction based on data decomposition and ten “plant” algorithm optimization using RELM [J]. Water Resources and Hydropower Engineering , 2025 , 56 ( 9 ): 118 – 130 . (in Chinese)
刘小蝶 , 张红月 , 芮小平 , 等 . 引入注意力机制的多因素LSTM地下水位预测模型 [J]. 水文 , 2025 , 45 ( 2 ): 73 - 79 .
LIU X D , ZHANG H Y , RUI X P , SUN Wen , et al . A multifactor LSTM groundwater level prediction model introducing an attention mechanism [J]. Journal of China Hydrology , 2025 , 45 ( 2 ): 73 – 79 . (in Chinese)
徐祥森 , 郑方元 , 查元源 , 等 . 基于图神经网络的灌区地下水位预测研究 [J]. 中国农村水利水电 , 2025 ( 6 ): 147 - 156 .
XU X S , ZHENG F Y , ZHA Y Y , et al . Prediction research on groundwater level in irrigation area based on graph neural networks [J]. China Rural Water and Hydropower , 2025 ( 6 ): 147 – 156 . (in Chinese)
冯鹏宇 , 金韬 , 沈一选 , 等 . 基于CNN-Transformer的城区地下水位预测 [J]. 计算机仿真 , 2023 , 40 ( 4 ): 492 - 498 .
FENG P Y , JIN T , SHEN Y X , et al . Prediction of urban groundwater level based on CNN-transformer [J]. Computer Simulation , 2023 , 40 ( 4 ): 492 – 498 . (in Chinese)
刀海娅 , 程刚 , 崔东文 . 多极小波包变换与改进浣熊算法优化的混合核极限学习机径流预测 [J]. 中国农村水利水电 , 2024 ( 6 ): 1 - 9,20 .
DAO H Y , CHENG G , CUI D W . Multipole wavelet packet transform and improved Raccoon algorithm optimized hybrid kernel limit learning machine for runoff prediction [J]. China Rural Water and Hydropower , 2024 ( 6 ): 1 – 9, 20 . (in Chinese)
李菊 , 崔东文 . 基于WPT-IDBO-RELM和WPT-IDMO-RELM模型的日径流预测 [J]. 水利水电科技进展 , 2024 , 44 ( 6 ): 48 - 55,85 .
LI J , CUI D W . Daily runoff prediction based on WPT-IDBO-RELM and WPT-IDMO-RELM models [J]. Advances in Science and Technology of Water Resources , 2024 , 44 ( 6 ): 48 – 55, 85 . (in Chinese)
邓智予 , 崔东文 . 基于二次分解技术与十种“鸟”群算法优化的OSELM月径流预测 [J]. 人民珠江 , 2025 , 46 ( 11 ): 44 - 54 .
DENG Z Y , CUI D W . Improved monthly runoff prediction of OSELM based on secondary decomposition technique and optimization of ten “bird” swarm algorithms [J]. Pearl River , 2025 , 46 ( 11 ): 44 – 54 . (in Chinese)
GUO Z Q , LI , G W , JIANG F . Chinese pangolin optimizer: a novel bio-inspired metaheuristic for solving optimization problems [J]. The Journal of Supercomputing 81 , 2025 ( 4 ). DOI: 10.1007/s11227-025-07004-4 http://dx.doi.org/10.1007/s11227-025-07004-4 .
郭辰星 , 李自成 , 徐瑞瑞 . 基于GRU-NN预测模型的压电作动器MPC-KAN控制方法 [J]. 压电与声光 , 2025 , 47 ( 1 ): 157 - 162,171 .
GUO C X , LI Z C , XU R R . MPC-KAN control method for piezoelectric actuators based on GRU-NN prediction model [J]. Piezoelectrics & Acoustooptics , 2025 , 47 ( 1 ): 157 – 162, 171 . (in Chinese)
陈思宇 , 李肖男 , 花续 , 等 . Kolmogorov-Arnold网络在长江中下游水位预报中的应用 [J]. 水力发电学报 , 2025 , 44 ( 4 ): 97 - 107 .
CHEN S Y , LI X N , HUA X , et al . Application of Kolmogorov-Arnold networks to water level forecasting in middle and lower Yangtze River [J]. Journal of Hydroelectric Engineering , 2025 , 44 ( 4 ): 97 – 107 . (in Chinese)
王毓灿 , 元海文 , 孙齐 , 等 . 不同地质条件下盾构机掘进速度预测方法 [J]. 仪器仪表学报 , 2025 , 46 ( 3 ): 30 - 40 .
WANG Y C , YUAN H W , SUN Q , et al . Prediction of tunneling speed of shield machine under varying geological conditions [J]. Chinese Journal of Scientific Instrument , 2025 , 46 ( 3 ): 30 – 40 . (in Chinese)
袁立宁 , 冯文刚 , 刘钊 . 基于Kolmogorov-Arnold网络的节点分类算法 [J]. 计算机科学与探索 , 2025 , 19 ( 3 ): 645 - 656 .
YUAN L N , FENG W G , LIU Z . Node classification algorithm based on Kolmogorov-Arnold networks [J]. Journal of Frontiers of Computer Science and Technology , 2025 , 19 ( 3 ): 645 – 656 . (in Chinese)
杨坪宏 , 胡奥 , 崔东文 , 等 . 基于数据处理与若干群体算法优化的GRU/LSTM水质时间序列预测 [J]. 水资源与水工程学报 , 2023 , 34 ( 4 ): 45 - 53 .
YANG P H , HU A , CUIi D W , et al . Prediction of GRU/LSTM water quality time series based on data processing and optimization of several swarm intelligence algorithms [J]. Journal of Water Resources and Water Engineering , 2023 , 34 ( 4 ): 45 – 53 . (in Chinese)
陈金红 , 崔东文 . 基于小波包分解的GJO-XGBoost水面蒸发量预测 [J]. 三峡大学学报(自然科学版) , 2023 , 45 ( 3 ): 1 - 7 .
CHEN J H , CUI D W . GJO-XGBoost prediction of water surface evaporation based on wavelet packet decomposition [J]. Journal of China Three Gorges University (Natural Sciences Edition) , 2023 , 45 ( 3 ): 1 – 7 . (in Chinese)
0
Views
0
下载量
0
CSCD
Publicity Resources
Related Articles
Related Author
Related Institution
京公网安备11010802024621