期刊文献+
共找到326,337篇文章
< 1 2 250 >
每页显示 20 50 100
Informer-LSTM融合算法在蓝莓基质温湿度预测中的研究与应用
1
作者 胡玲艳 陈鹏宇 +6 位作者 郭占俊 徐国辉 秦山 付康 盖荣丽 汪祖民 张雨萌 《郑州大学学报(理学版)》 北大核心 2026年第1期78-86,共9页
为了精准预测温室蓝莓基质的温湿度变化趋势,提出一种融合Informer-LSTM算法的温湿度预测方法。以温室蓝莓现场环境数据为研究对象,使用LSTM算法捕捉时间序列数据中的依赖关系并与自注意力机制相结合,使模型在聚焦自注意力特征的同时兼... 为了精准预测温室蓝莓基质的温湿度变化趋势,提出一种融合Informer-LSTM算法的温湿度预测方法。以温室蓝莓现场环境数据为研究对象,使用LSTM算法捕捉时间序列数据中的依赖关系并与自注意力机制相结合,使模型在聚焦自注意力特征的同时兼顾LSTM特征,以增强其长期记忆力。在生成初步预测序列后,再应用LSTM算法修正模型的短期注意力,提高模型的反应速度。实验结果显示,Informer-LSTM预测模型在预测准确率、鲁棒性和响应速度等方面都有显著的优势。当温度湿度等时序输入数据发生明显变化时,模型能快速捕获短期内输入数据的动态模式变化。该模型在智慧温室管理中,对辅助人工决策及实现智能化控制具有较高实际价值。 展开更多
关键词 智慧农业 温室蓝莓 informer模型 LSTM模型 温湿度预测
在线阅读 下载PDF
基于Informer-SAO-LSTM的刀具磨损预测
2
作者 李昂 马俊燕 唐源斌 《组合机床与自动化加工技术》 北大核心 2026年第1期151-155,161,共6页
在产品加工过程中,准确预测刀具的磨损值既能避免过早更换造成的成本浪费,又可防止过度磨损影响加工精度,从而最大化发挥刀具寿命的价值。为了解决这个问题,提出了一种基于Informer、SAO与LSTM结合的深度学习网络模型,用于刀具磨损预测... 在产品加工过程中,准确预测刀具的磨损值既能避免过早更换造成的成本浪费,又可防止过度磨损影响加工精度,从而最大化发挥刀具寿命的价值。为了解决这个问题,提出了一种基于Informer、SAO与LSTM结合的深度学习网络模型,用于刀具磨损预测。Informer具有高效的编码器结构和稀疏自注意力机制,而LSTM网络具有较强的时间序列建模能力,通过SAO算法对超参数的调整,可以更准确高效地捕捉刀具磨损过程中长期的依赖关系,从而提取更有效的特征,提升了模型在处理长序列数据时的效率和准确性。使用PHM2010数据集进行对比实验,实验结果表明所提出的Informer-SAO-LSTM模型在MAE、RMSE等多项指标上均表现出色,最后设计了实验进行验证,进一步说明了所提出的方法比对比模型的预测准确率更高,泛化能力更好。 展开更多
关键词 LSTM informER SAO 刀具磨损 深度学习 时间序列预测
在线阅读 下载PDF
Information for Authors
3
《Journal of Geographical Sciences》 2026年第1期F0003-F0003,共1页
1 General information Journal of Geographical Sciences is an international academic journal that publishes papers of the highest quality in physical geography, natural resources, environmental sciences, geographic inf... 1 General information Journal of Geographical Sciences is an international academic journal that publishes papers of the highest quality in physical geography, natural resources, environmental sciences, geographic information sciences, remote sensing and cartography. Manuscripts come from different parts of the world. 展开更多
关键词 natural resources remote sensing environmental sciences physical geography geographic information sciences
原文传递
Deep Learning-Enhanced Human Sensing with Channel State Information: A Survey
4
作者 Binglei Yue Aili Jiang +3 位作者 Chun Yang Junwei Lei Heng Liu Yin Zhang 《Computers, Materials & Continua》 2026年第1期1-28,共28页
With the growing advancement of wireless communication technologies,WiFi-based human sensing has gained increasing attention as a non-intrusive and device-free solution.Among the available signal types,Channel State I... With the growing advancement of wireless communication technologies,WiFi-based human sensing has gained increasing attention as a non-intrusive and device-free solution.Among the available signal types,Channel State Information(CSI)offers fine-grained temporal,frequency,and spatial insights into multipath propagation,making it a crucial data source for human-centric sensing.Recently,the integration of deep learning has significantly improved the robustness and automation of feature extraction from CSI in complex environments.This paper provides a comprehensive review of deep learning-enhanced human sensing based on CSI.We first outline mainstream CSI acquisition tools and their hardware specifications,then provide a detailed discussion of preprocessing methods such as denoising,time–frequency transformation,data segmentation,and augmentation.Subsequently,we categorize deep learning approaches according to sensing tasks—namely detection,localization,and recognition—and highlight representative models across application scenarios.Finally,we examine key challenges including domain generalization,multi-user interference,and limited data availability,and we propose future research directions involving lightweight model deployment,multimodal data fusion,and semantic-level sensing. 展开更多
关键词 Channel State information(CSI) human sensing human activity recognition deep learning
在线阅读 下载PDF
Evolution of Smart Parks and Development of Park Information Modeling(PIM):Concept and Design Application 被引量:2
5
作者 YANG Kaixian ZHEN Feng ZHANG Shanqi 《Chinese Geographical Science》 2025年第5期982-998,共17页
With the development of smart cities and smart technologies,parks,as functional units of the city,are facing smart transformation.The development of smart parks can help address challenges of technology integration wi... With the development of smart cities and smart technologies,parks,as functional units of the city,are facing smart transformation.The development of smart parks can help address challenges of technology integration within urban spaces and serve as testbeds for exploring smart city planning and governance models.Information models facilitate the effective integration of technology into space.Building Information Modeling(BIM)and City Information Modeling(CIM)have been widely used in urban construction.However,the existing information models have limitations in the application of the park,so it is necessary to develop an information model suitable for the park.This paper first traces the evolution of park smart transformation,reviews the global landscape of smart park development,and identifies key trends and persistent challenges.Addressing the particularities of parks,the concept of Park Information Modeling(PIM)is proposed.PIM leverages smart technologies such as artificial intelligence,digital twins,and collaborative sensing to help form a‘space-technology-system’smart structure,enabling systematic management of diverse park spaces,addressing the deficiency in park-level information models,and aiming to achieve scale articulation between BIM and CIM.Finally,through a detailed top-level design application case study of the Nanjing Smart Education Park in China,this paper illustrates the translation process of the PIM concept into practice,showcasing its potential to provide smart management tools for park managers and enhance services for park stakeholders,although further empirical validation is required. 展开更多
关键词 smart park smart city Park information Modeling(PIM) smart technology Building information Modeling(BIM) City information Modeling(CIM)
在线阅读 下载PDF
Study on information utility 被引量:1
6
作者 张东戈 《Journal of Systems Engineering and Electronics》 SCIE EI CSCD 2005年第3期579-582,644,共5页
Information has two aspects. One aspect is the objective one; another aspect is the subjective one. Shannon has discussed the objective aspect of information in information theory. But the subjective aspect of informa... Information has two aspects. One aspect is the objective one; another aspect is the subjective one. Shannon has discussed the objective aspect of information in information theory. But the subjective aspect of information has not been fully discussed. Someone use “Bayesian approaches” to evaluate the value of information. But in some cases it does not meet the information user's need. This paper is focus on the subjective aspect of objectively measurable information and gives formal definitions for information, information utility, and marginal information utility, normalized calculation of information utility. The information discussed in the paper has interdisciplinary nature. This work can be the foundation of many application areas. 展开更多
关键词 analysis definition of information analysis definition of information utility analysis definition of marginal information utility normalized calculation of information utility mic^nomics analysis expression of information utility information utility functions C^4ISR system evaluation.
在线阅读 下载PDF
A Study on Information Systems Providing Local Government Information:Focusing on Website UI/UX Evaluation
7
作者 Xuan Fu Euitai Jung 《Psychology Research》 2025年第2期54-72,共19页
This study analyzes the User Interface(UI)and User Experience(UX)of information systems that provide local government information.The systems analyzed are the Local Administrative Comprehensive Information Disclosure ... This study analyzes the User Interface(UI)and User Experience(UX)of information systems that provide local government information.The systems analyzed are the Local Administrative Comprehensive Information Disclosure System(Zheripan),the Integrated Local Financial Disclosure System(Qinching Online),and the Local Regulations Information System(12348 Zhejiang Legal Network).The Local Administrative Comprehensive Information Disclosure System offers public service and personnel information,while the Integrated Local Financial Disclosure System provides financial information,and the Local Regulations Information System offers legal information as its main content.The analysis framework utilized three elements:objective data,psychological factors,and heuristic evaluation.The results of the first objective data analysis show that approximately 70%of visits to Zheripan and Qinching Online are through search,and the time spent on the homepage is short.In contrast,about 70%of visits to the 12348 Zhejiang Legal Network are direct visits,with users browsing multiple pages with a clear purpose.In terms of data provision methods,Zheripan provides two types of data in three formats,Qinching Online offers 28 types of data in five formats,and 12348 Zhejiang Legal Network provides one type of information in a single format.The second psychological factor analysis found that all three websites had a number of menus suitable for short-term cognitive capacity.However,only one of the sites had a layout that considered the user’s eye movement.Finally,the heuristic evaluation revealed that most of the evaluation criteria were not met.While the design is relatively simple and follows standards,feedback for users,error prevention,and help options were lacking.Moreover,the user-specific usability was low,and the systems remained at the information-providing level.Based on these findings,both short-term and long-term improvement measures for creating an interactive system beyond simple information disclosure are proposed. 展开更多
关键词 local government information regional information information systems WEBSITE UI UX
在线阅读 下载PDF
Geospatial Information Technology Innovations: From Earth Monitoring to Urban Planning
8
作者 Jiancheng Li Weiping Jiang 《Engineering》 2025年第4期1-2,共2页
Nowadays,spatiotemporal information,positioning,and navigation services have become critical components of new infrastructure.Precise positioning technology is indispensable for determining spatiotemporal information ... Nowadays,spatiotemporal information,positioning,and navigation services have become critical components of new infrastructure.Precise positioning technology is indispensable for determining spatiotemporal information and providing navigation services. 展开更多
关键词 geospatial information technology determining spatiotemporal information positioning technology spatiotemporal information earth monitoring navigation services urban planning
在线阅读 下载PDF
Estimation of the probability of informed trading models via an expectation‑conditional maximization algorithm
9
作者 Montasser Ghachem Oguz Ersan 《Financial Innovation》 2025年第1期1860-1896,共37页
The estimation of the probability of informed trading(PIN)model and its extensions poses significant challenges owing to various computational problems.To address these issues,we propose a novel estimation method call... The estimation of the probability of informed trading(PIN)model and its extensions poses significant challenges owing to various computational problems.To address these issues,we propose a novel estimation method called the expectation-conditional-maximization(ECM)algorithm,which can serve as an alternative to the existing methods for estimating PIN models.Our method provides optimal estimates for the original PIN model as well as two of its extensions:the multilayer PIN model and the adjusted PIN model,along with its restricted versions.Our results indicate that estimations using the ECM algorithm are generally faster,more accurate,and more memory-efficient than the standard methods used in the literature,making it a robust alternative.More importantly,the ECM algorithm is not limited to the models discussed and can be easily adapted to estimate future extensions of the PIN model. 展开更多
关键词 Expectation conditional-maximization algorithm ECM PIN model MPIN Multilayer probability of informed trading Adjusted PIN model Maximum-likelihood estimation Private information information asymmetry
在线阅读 下载PDF
New Perspective for the Philosophy of Information:Re-construction&New Definition of Branches,and Hybrid Sciences&Systems
10
作者 Refet Ramiz 《Philosophy Study》 2025年第6期315-363,共49页
In this work,general definition and meaning of knowledge,information,data and symbol are expressed generally/specifically,and the differences/relationships between them are briefly discussed.The general definition of ... In this work,general definition and meaning of knowledge,information,data and symbol are expressed generally/specifically,and the differences/relationships between them are briefly discussed.The general definition of system is briefly interpreted,and the semantic contents of the concept“system”expressed with nine perspectives generally.The meaning and importance of philosophy of information are then defined according to the general approaches.Some of the important philosophers of information and their professional interests are evaluated.The meaning and importance of mind,and philosophy of mind are discussed due to general approaches.Some of the philosophers of mind and their interests are evaluated and compared with a table.Systems philosophy is defined in line with general approaches,and its relationships with four main areas are stated.The new perspective of philosophy is then defined by the author generally,and the eight basic branches of philosophy and hybrid philosophy,along with their relevant theories,are briefly outlined.R-Philosophy,R-Science,R-Information,R-Mind,and R-System new disciplines are shortly expressed.New perspective for philosophy of information is defined as complementary branch with other seven basic philosophies.Types of information due to method,size,and content are given with a table.The 23 sub-branches of philosophy of information are defined generally/specifically.Philosophy of basic senses and some other branches are new defined,and new perspectives for philosophy of mind and for some other branches are expressed specifically.18 hybrid philosophies for information are defined,and their relations with philosophy of information explained generally/specifically.General disciplines and concepts about information are defined shortly,and information science(s),2D-6D hybrid information sciences,information system(s),and information&communication systems are given with details.New perspective for philosophy of system is defined,and types of system due to methods,size,and content are given with a table.Hybrid philosophies for systems,and some disciplines and concepts about systems are shortly outlined.Systems science(s)are defined due to four categories and each of these categories is explained with detailed tables.Hybrid systems defined by the author are shortly interpreted. 展开更多
关键词 philosophy of information philosophy of mind philosophy of system New Era Philosophy information science(s) hybrid philosophies philosophy of basic senses information systems systems science(s) hybrid systems
在线阅读 下载PDF
基于DWT-CNN-Informer模型的液压支架压力多步长预测 被引量:3
11
作者 张传伟 张刚强 +1 位作者 路正雄 李林岳 《中国安全生产科学技术》 北大核心 2025年第4期57-63,共7页
为了实现液压支架压力多步长精准预测,提出1种基于DWT-CNN-Informer模型的压力多步长预测方法,该方法利用离散小波变换(discrete wavelet transform, DWT)将预处理后的压力时序数据分解为趋势项和周期项频率分量;各频率分量输入卷积神... 为了实现液压支架压力多步长精准预测,提出1种基于DWT-CNN-Informer模型的压力多步长预测方法,该方法利用离散小波变换(discrete wavelet transform, DWT)将预处理后的压力时序数据分解为趋势项和周期项频率分量;各频率分量输入卷积神经网络(CNN)模型提取频率特征;提取的频率特征输入Informer编码器,经位置编码和多头概率稀疏自注意力机制捕捉时序变化特征,并结合自注意力蒸馏减少特征冗余;将Informer解码器改为全连接层,直接输出各分量多步长预测结果;重构叠加各分量多步长预测结果得到液压支架压力多步长预测结果。研究结果表明:在预测步长分别为6,12,24时,DWT-CNN-Informer模型相比LSTM、Informer、CNN-Informer模型在平均绝对误差(MAE)、均方根误差(RMSE)、对称平均绝对百分比误差(SMAPE)指标上均表现出更高预测精度。研究结果为液压支架压力精准预测提供有效方法。 展开更多
关键词 液压支架压力 多步长预测 离散小波变换 CNN模型 informer模型
在线阅读 下载PDF
改进Informed RRT^(*)算法移动机器人路径规划 被引量:3
12
作者 鲁宇明 周羽逵 +2 位作者 郭鑫 池吕庭 戴骏 《计算机工程与应用》 北大核心 2025年第8期283-293,共11页
Informed RRT^(*)算法对初始解不敏感,规划出的路径太接近障碍物,导致路径不平滑。提出一种改进的Informed RRT^(*)路径规划算法,该算法改进了约束采样空间和引导策略。在采样初期,将采样区域限制在一个圆形区域,加快初始解收敛,在算法... Informed RRT^(*)算法对初始解不敏感,规划出的路径太接近障碍物,导致路径不平滑。提出一种改进的Informed RRT^(*)路径规划算法,该算法改进了约束采样空间和引导策略。在采样初期,将采样区域限制在一个圆形区域,加快初始解收敛,在算法规划的过程中引入人工势场中引力场和斥力场的思想,使机器人与障碍物保持安全距离,并向目标位置行进。对Informed RRT^(*)算法和基于目标偏置的Informed RRT^(*)算法(Goal-bias-Informed RRT^(*))以及改进后的Informed RRT^(*)算法进行比较实验,实验结果验证了改进后Informed RRT^(*)算法的有效性和优越性及稳定性。该算法较Informed RRT^(*)算法和Goal-bias-Informed RRT^(*)效率更高、更容易得到初始解、更安全、更平滑、更稳定。 展开更多
关键词 移动机器人 路径规划 随机采样 informed RRT^(*)算法 目标偏置 约束采样空间
在线阅读 下载PDF
基于多任务Informer模型的船舶轨迹预测及行为识别研究
13
作者 李世刚 刘克中 +3 位作者 陈立家 周乃祺 周阳 黄嘉韬 《中国航海》 北大核心 2025年第3期157-165,共9页
为有效预判航行风险,并为船舶避碰、交通管理等决策提供重要依据,研究了一种基于多任务Informer模型的船舶轨迹预测及行为识别模型。该模型以Informer框架为基础,并引入多任务学习模式,通过设计多任务损失函数将船舶行为识别与轨迹预测... 为有效预判航行风险,并为船舶避碰、交通管理等决策提供重要依据,研究了一种基于多任务Informer模型的船舶轨迹预测及行为识别模型。该模型以Informer框架为基础,并引入多任务学习模式,通过设计多任务损失函数将船舶行为识别与轨迹预测并联训练,解决了AIS数据中船舶行为不准确无法作为模型输入的问题;在模型训练时,并设计基于同方差不确定性的损失函数自适应更新策略,自适应分配两个任务的损失权重。利用太仓航段水域中的真实AIS数据进行试验中多任务的Informer船舶轨迹预测模型在轨迹预测中的损失比LSTM和Informer模型分别降低了40.2%和14.7%;在行为识别任务中多任务模型的识别准确率比LSTM和Informer模型分别提升了11.7%和5.95%。表明了多任务模型能在有效提升船舶轨迹预测的性能的同时实现船舶对行为的准确识别。 展开更多
关键词 轨迹预测 行为识别 AIS数据 informer模型 多任务学习
在线阅读 下载PDF
基于融合劣化指标和VMD-Informer的水电机组劣化趋势预测
14
作者 宋阿妮 陈亦真 +2 位作者 詹云峰 李超顺 付波 《中国农村水利水电》 北大核心 2025年第5期90-96,共7页
水电机组长期运行在恶劣环境下,异常振动更加频繁,逐渐出现疲劳、磨损,导致机组性能劣化。为保障机组的安全稳定运行,需要准确直观地反映水电机组运行并预测机组未来劣化状况,为机组状态检修提供重要依据。提出了一种基于融合劣化指标和... 水电机组长期运行在恶劣环境下,异常振动更加频繁,逐渐出现疲劳、磨损,导致机组性能劣化。为保障机组的安全稳定运行,需要准确直观地反映水电机组运行并预测机组未来劣化状况,为机组状态检修提供重要依据。提出了一种基于融合劣化指标和VMD-Informer的机组劣化趋势预测方法。首先构建KAN健康模型拟合工况参数与振摆值之间的映射关系,然后通过对比模型输出值与实测振摆值在不同指标下的差异得到多个劣化序列,运用遗传算法对多个劣化序列进行寻优获取融合劣化指标,兼顾多个指标的优势,更为准确地反映机组劣化趋势。之后用变分模态分解(VMD)将融合劣化序列分解为多个分量,最后利用Informer预测模型对分解后的各个分量进行多步预测并重构得到最终的预测结果,从而实现对机组运行状况的准确评估和预测。实例分析表明,所提方法能够生成可靠的劣化趋势,同时在预测上能学习劣化趋势序列的长期趋势和局部特征,预测精度更高。 展开更多
关键词 水电机组 劣化评估 退化预测 Kolmogorov-Arnold Network 遗传算法 informER
在线阅读 下载PDF
融合LSTM与Informer网络的锂离子电池SOH估计
15
作者 胡远志 陈飞 +2 位作者 刘西 冯仁华 舒星 《武汉大学学报(理学版)》 北大核心 2025年第6期813-820,共8页
锂离子电池在长期使用过程中易发生容量衰退,显著影响电动汽车的续航性能和安全。为精准评估电池的健康状态(SOH),建立了一种融合长短期记忆神经网络(LSTM)和Informer网络的锂离子电池SOH估计方法。采用局部离群因子(LOF)算法对实验电... 锂离子电池在长期使用过程中易发生容量衰退,显著影响电动汽车的续航性能和安全。为精准评估电池的健康状态(SOH),建立了一种融合长短期记忆神经网络(LSTM)和Informer网络的锂离子电池SOH估计方法。采用局部离群因子(LOF)算法对实验电池的充电数据进行预处理,提取电池充电电压曲线的最大切线角及其对应时间作为健康特征,并基于Spearman相关系数分析了其与SOH之间的相关性。结合LSTM在局部特征提取方面的优势与Informer捕捉全局依赖关系方面的能力,构建了融合LSTM与Informer的串联网络,并将提取的健康特征作为模型输入,实现了对电池SOH的精确估算。实验结果表明,该方法在估算精度上具有显著优势,最大绝对误差可控制在2.5%以内,均方根误差(RMSE)和平均绝对值误差(MAE)均控制在1%以内。与传统单一网络对比,该方法具有更优的估算性能与泛化能力。 展开更多
关键词 锂离子电池 电池健康状态(SOH) 异常值处理 长短期记忆(LSTM)网络 informer网络
原文传递
基于RF-Informer模型的月径流遥相关预报
16
作者 李继清 谢宇韬 +1 位作者 徐学军 吴亮 《水资源保护》 北大核心 2025年第3期39-45,共7页
为延长中长期径流预报的预见期,提高预报精度,从物理成因上考虑径流的影响因素,在前期降水径流的基础上增加遥相关因子,通过随机森林(RF)模型进行因子选择,引入长时间序列预报中表现良好的Informer模型,构建了月径流预报的RF-Informer模... 为延长中长期径流预报的预见期,提高预报精度,从物理成因上考虑径流的影响因素,在前期降水径流的基础上增加遥相关因子,通过随机森林(RF)模型进行因子选择,引入长时间序列预报中表现良好的Informer模型,构建了月径流预报的RF-Informer模型,并利用该模型对雅砻江流域两河口、锦西、二滩3个水库的入库月径流进行了预报。结果表明:将遥相关因子引入流域月径流预报可以延长预见期,提高预报精度;相较于线性相关法,基于RF模型选择预报因子可以挖掘因子间非线性关系,提升预报效果;与RF-LSTM、RF-SVM、RF-BP神经网络模型相比,RF-Informer模型的误差最小,预报精度最高。 展开更多
关键词 月径流预报 遥相关因子 随机森林模型 informer模型 雅砻江流域
在线阅读 下载PDF
基于Informer模型的航班延误预测
17
作者 杨新湦 游超 朱承元 《科学技术与工程》 北大核心 2025年第19期8282-8288,共7页
为能更加精准预测不同时段的航班延误态势,选用美国亚特兰大机场2023年全年的运行数据与相关气象数据进行实验,提出基于相关系分析(correlation analysis,CA),主成分分析(principal components analysis,PCA)和Informer模型的CA-PCA-Inf... 为能更加精准预测不同时段的航班延误态势,选用美国亚特兰大机场2023年全年的运行数据与相关气象数据进行实验,提出基于相关系分析(correlation analysis,CA),主成分分析(principal components analysis,PCA)和Informer模型的CA-PCA-Informer航班延误预测模型,采用MAE(mean absolute error)和RMSE(root mean square error)作为模型的评价指标进行预测误差分析。结果表明,CA-PCA-Informer模型比简单的组合模型预测效果更好,与CA-PCA-LSTM和CA-PCA-GRU模型相比模型误差最低,MAE和RMSE分别降低了20.2%~20.7%和12.7%~14.1%;CA-PCA-Informer模型对预测步长为1 h时预测更为精准,该模型可以为决策者提供更加准确的航班延误态势以保证航班的高效运行。 展开更多
关键词 民航交通运输 航班延误预测 informer模型 主成分分析 神经网络
在线阅读 下载PDF
基于数字孪生和Informer-LSTM的滚动轴承剩余寿命预测 被引量:1
18
作者 王雅君 崔海峰 +1 位作者 刘云松 鞠鸿宇 《组合机床与自动化加工技术》 北大核心 2025年第7期135-140,共6页
为了能够精准地实现滚动轴承剩余使用寿命的预测,提出了一种使用数字孪生技术和时间序列预测结合的预测方法,并且对Informer的解码器进行修改,引入LSTM以增强序列建模能力,构建出Informer-LSTM混合模型进行预测。通过数字孪生技术建立... 为了能够精准地实现滚动轴承剩余使用寿命的预测,提出了一种使用数字孪生技术和时间序列预测结合的预测方法,并且对Informer的解码器进行修改,引入LSTM以增强序列建模能力,构建出Informer-LSTM混合模型进行预测。通过数字孪生技术建立轴承的虚拟模型,实现对轴承物理状态的监控和模拟,并为剩余使用寿命的预测提供数据,通过提取历史数据和数字孪生数据的时域和频域特征,建立数据集,并利用Informer-LSTM模型对收集到的时间序列数据进行处理,实现轴承的剩余寿命的预测。通过实验看出,该方法在预测滚动轴承剩余寿命方面具有较高的准确性和稳定性,可以为实现设备的智能维护和故障预防提供了有力支持。 展开更多
关键词 数字孪生 时间序列预测 informer-LSTM 剩余使用寿命
在线阅读 下载PDF
融合改进Informer与迁移学习的风电功率预测
19
作者 郭利进 孙淼 衡安阳 《太阳能学报》 北大核心 2025年第7期371-377,共7页
为克服风电功率序列的不稳定性导致预测精度低以及一些风电场历史数据有限的问题,提出一种特征交互Informer与迁移学习(FIITL)的风电功率预测模型。首先提出特征交互(FI)机制用双通道输入进一步提取信息,并将迁移学习(TL)引入到预测模型... 为克服风电功率序列的不稳定性导致预测精度低以及一些风电场历史数据有限的问题,提出一种特征交互Informer与迁移学习(FIITL)的风电功率预测模型。首先提出特征交互(FI)机制用双通道输入进一步提取信息,并将迁移学习(TL)引入到预测模型中,提出循环微调迁移学习,将模型从源监测站迁移到目标站,实现在有限历史数据情况下预测性能的提升。最后,通过与传统Informer模型及其他基线预测方法比较,FIITL模型展现了在有限数据情况下的性能优势。 展开更多
关键词 迁移学习 风电功率 预测 informER 特征交互
原文传递
基于SDAE-EEMD降噪分解与改进Informer-BiLSTM模型的电力短期负荷预测方法
20
作者 蔡子龙 李嘉棋 +3 位作者 沈赋 王健 徐潇源 杨宇林 《电网技术》 北大核心 2025年第12期5009-5018,I0010-I0015,共16页
当前短期负荷预测模型在电价与负荷动态融合机制、负荷数据降噪与时序特征提取环节仍存在不足,制约了预测精度的提升。该文提出了一种集成电价及气象多维特征的短期电力负荷预测框架。首先,结合堆叠降噪自编码器(stacked denoising auto... 当前短期负荷预测模型在电价与负荷动态融合机制、负荷数据降噪与时序特征提取环节仍存在不足,制约了预测精度的提升。该文提出了一种集成电价及气象多维特征的短期电力负荷预测框架。首先,结合堆叠降噪自编码器(stacked denoising autoencoders,SDAE)和集合经验模态分解(ensemble empirical mode decomposition,EEMD)构建混合降噪分解模块,有效抑制负荷序列中的噪声干扰和模态混叠问题。EEMD将去噪后负荷序列分解为固有模态函数(intrinsic mode functions,IMFs)分量。其次,基于最大信息系数(maximum information coefficient,MIC)分析,将电价和气象特征分别融入高、低频IMFs分量中,实现差异化的特征动态融合。在此基础上,提出分频预测策略。针对高频分量,引入全局时间戳编码与稀疏注意力机制的改进Informer模型,以捕捉短时剧烈波动特征;对低频分量,采用Bi LSTM网络捕捉长期趋势与周期性。最后,基于澳大利亚国家电力市场公开数据集的实证结果表明,在平均绝对百分比误差和均方误差两个指标上均显著优于未引入电价特征或未采用分频策略的对比模型。通过高质量数据预处理、关键特征动态融合与针对性分频结构建模的协同优化,有效提升了短期负荷预测的精度与稳定性,可为电力市场动态定价机制下的负荷预测提供高效可靠的技术支撑。 展开更多
关键词 短期负荷预测 电价 SDAE EEMD 改进informer BiLSTM 分频预测
原文传递
上一页 1 2 250 下一页 到第
使用帮助 返回顶部