{"title":"Tem2-KAN:基于改进的Kolmogorov-Arnold网络的数据驱动时间温度预测。","authors":"Yongxiang Lei, Bin Deng, Ziyang Wang","doi":"10.1016/j.isatra.2025.07.014","DOIUrl":null,"url":null,"abstract":"<p><p>Accurate temperature forecasting relies on traditional meteorological parameters that are essential for monitoring weather informatics and guiding forecasting efforts. This study introduces a deep learning architecture for high-precision climate temperature forecasting via an improved Kolmogorov-Arnold Networks, named Tem<sup>2</sup>-KAN. Grounded in the Kolmogorov-Arnold representation theorem, Tem<sup>2</sup>-KAN explores replacing conventional linear weights in neural networks with spline-parameterized univariate functions, enabling dynamic learning of nonlinear climate patterns while maintaining intrinsic interpretability. The proposed framework uniquely integrates the universal approximation capabilities of Multi-Layer Perceptrons (MLPs) with physically meaningful feature visualization through its adaptive activation functions, addressing critical limitations of black-box climate models. A temperature prediction pipeline is established that first preprocesses raw meteorological data from UK monitoring stations, then trains Tem<sup>2</sup>-KAN to map historical trends to multi-horizon forecasts. Rigorous evaluations on real-world climate datasets demonstrate Tem<sup>2</sup>-KAN's dual advantage achieving state-of-the-art prediction accuracy while utilizing fewer trainable parameters. In addition, a systematic ablation study quantifies the sensitivity of key Tem<sup>2</sup>-KAN-specific hyperparameters (spline order k, grid resolution grid) on forecasting performance. Finally, we theoretically prove Tem<sup>2</sup>-KAN's universal approximation capacity through function space analysis, and practically, we demonstrate its interpretability and prediction performance. These innovations position Tem<sup>2</sup>-KAN as a paradigm-shifting tool for climate informatics, offering meteorologists both high predictive performance and mechanistic insight into temperature dynamics. The framework's reduced hyperparameter complexity further enhances its viability for operational forecasting systems.</p>","PeriodicalId":94059,"journal":{"name":"ISA transactions","volume":" ","pages":""},"PeriodicalIF":6.5000,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Tem<sup>2</sup>-KAN: Data-driven temporal temperature prediction via an improved Kolmogorov-Arnold network.\",\"authors\":\"Yongxiang Lei, Bin Deng, Ziyang Wang\",\"doi\":\"10.1016/j.isatra.2025.07.014\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><p>Accurate temperature forecasting relies on traditional meteorological parameters that are essential for monitoring weather informatics and guiding forecasting efforts. This study introduces a deep learning architecture for high-precision climate temperature forecasting via an improved Kolmogorov-Arnold Networks, named Tem<sup>2</sup>-KAN. Grounded in the Kolmogorov-Arnold representation theorem, Tem<sup>2</sup>-KAN explores replacing conventional linear weights in neural networks with spline-parameterized univariate functions, enabling dynamic learning of nonlinear climate patterns while maintaining intrinsic interpretability. The proposed framework uniquely integrates the universal approximation capabilities of Multi-Layer Perceptrons (MLPs) with physically meaningful feature visualization through its adaptive activation functions, addressing critical limitations of black-box climate models. A temperature prediction pipeline is established that first preprocesses raw meteorological data from UK monitoring stations, then trains Tem<sup>2</sup>-KAN to map historical trends to multi-horizon forecasts. Rigorous evaluations on real-world climate datasets demonstrate Tem<sup>2</sup>-KAN's dual advantage achieving state-of-the-art prediction accuracy while utilizing fewer trainable parameters. In addition, a systematic ablation study quantifies the sensitivity of key Tem<sup>2</sup>-KAN-specific hyperparameters (spline order k, grid resolution grid) on forecasting performance. Finally, we theoretically prove Tem<sup>2</sup>-KAN's universal approximation capacity through function space analysis, and practically, we demonstrate its interpretability and prediction performance. These innovations position Tem<sup>2</sup>-KAN as a paradigm-shifting tool for climate informatics, offering meteorologists both high predictive performance and mechanistic insight into temperature dynamics. The framework's reduced hyperparameter complexity further enhances its viability for operational forecasting systems.</p>\",\"PeriodicalId\":94059,\"journal\":{\"name\":\"ISA transactions\",\"volume\":\" \",\"pages\":\"\"},\"PeriodicalIF\":6.5000,\"publicationDate\":\"2025-07-10\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"ISA transactions\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1016/j.isatra.2025.07.014\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"ISA transactions","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1016/j.isatra.2025.07.014","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Tem2-KAN: Data-driven temporal temperature prediction via an improved Kolmogorov-Arnold network.
Accurate temperature forecasting relies on traditional meteorological parameters that are essential for monitoring weather informatics and guiding forecasting efforts. This study introduces a deep learning architecture for high-precision climate temperature forecasting via an improved Kolmogorov-Arnold Networks, named Tem2-KAN. Grounded in the Kolmogorov-Arnold representation theorem, Tem2-KAN explores replacing conventional linear weights in neural networks with spline-parameterized univariate functions, enabling dynamic learning of nonlinear climate patterns while maintaining intrinsic interpretability. The proposed framework uniquely integrates the universal approximation capabilities of Multi-Layer Perceptrons (MLPs) with physically meaningful feature visualization through its adaptive activation functions, addressing critical limitations of black-box climate models. A temperature prediction pipeline is established that first preprocesses raw meteorological data from UK monitoring stations, then trains Tem2-KAN to map historical trends to multi-horizon forecasts. Rigorous evaluations on real-world climate datasets demonstrate Tem2-KAN's dual advantage achieving state-of-the-art prediction accuracy while utilizing fewer trainable parameters. In addition, a systematic ablation study quantifies the sensitivity of key Tem2-KAN-specific hyperparameters (spline order k, grid resolution grid) on forecasting performance. Finally, we theoretically prove Tem2-KAN's universal approximation capacity through function space analysis, and practically, we demonstrate its interpretability and prediction performance. These innovations position Tem2-KAN as a paradigm-shifting tool for climate informatics, offering meteorologists both high predictive performance and mechanistic insight into temperature dynamics. The framework's reduced hyperparameter complexity further enhances its viability for operational forecasting systems.