Article
Construction & Building Technology
Jing Lin, Julian A. Fernandez, Rakiba Rayhana, Amirhossein Zaji, Ran Zhang, Omar E. Herrera, Zheng Liu, Walter Merida
Summary: This study introduces a novel predictive analysis method for building power demand, which can predict pattern profile of power demand and upcoming normal and abnormal behaviors. Experimental results show that power demand can be mapped to different foreseeable demand patterns, with an accuracy of anomaly prediction at 88%.
ENERGY AND BUILDINGS
(2022)
Article
Thermodynamics
Yejin Hong, Sungmin Yoon, Sebin Choi
Summary: This study proposes a novel symbolic hierarchical clustering method (HOS-SAX) to evaluate the efficiency and energy usage patterns of building systems. The analysis shows that approximately 71% of the entire operation period is characterized by inefficient operation, and a supply temperature reduction of 0.87 degrees C is expected in the most inefficient sections.
Article
Computer Science, Artificial Intelligence
Dongfang Zhao, Yesheng Chen, Shulin Liu, Jiayi Shen, Zhonghua Miao
Summary: Fault diagnosis is crucial for industrial equipment maintenance, and feature extraction plays a key role. Symbolic aggregate approximation (SAX) is a popular feature extraction approach with great potential, but it suffers from information aliasing. This study introduces a novel alternative method called parallel symbolic aggregate approximation (PSAX), which suppresses information aliasing and improves accuracy in fault diagnosis.
JOURNAL OF INTELLIGENT & FUZZY SYSTEMS
(2023)
Article
Chemistry, Analytical
Lei Hu, Ligui Wang, Yanlu Chen, Niaoqing Hu, Yu Jiang
Summary: This study proposes a rolling bearing fault diagnosis method that combines PAA with CEEMDAN, which can effectively separate fault vibration signals and improve diagnosis.
Article
Computer Science, Information Systems
Bin Lei, Peng Zhang, Yifei Suo, Na Li
Summary: This paper proposes a SAX-STGCN network for traffic flow prediction, which addresses the limitation of adjacency matrix in representing spatial correlation of traffic flow. By introducing a similarity matrix, this model captures the spatial and temporal correlation of traffic flow more accurately and demonstrates good long-term prediction ability.
Article
Chemistry, Analytical
Dong-Hyuk Yang, Yong-Shin Kang
Summary: Time-series representation is crucial in time-series analysis. Symbolic aggregate approximation (SAX) is widely used, but it only focuses on the mean value. We propose a new method, distance- and momentum-based symbolic aggregate approximation (DM-SAX), which considers time-series distribution and trend. Experimental results show that DM-SAX outperforms other methods in various tasks and can identify meaningful data points.
Proceedings Paper
Computer Science, Artificial Intelligence
Lamprini Pappa, Petros Karvelis, George Georgoulas, Chrysostomos Stylios
Summary: The introduced Slopewise Aggregate Approximation (SAA) efficiently describes the trend of a time series signal and achieves dimensionality reduction by incorporating shape and fluctuation information. Applying discretization transforms the problem into a symbolic space problem, using Intelligent Icons as features for Human Activity Recognition. The proposed method shows significant improvement in classification metrics compared to past implementations.
2021 IEEE SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE (IEEE SSCI 2021)
(2021)
Article
Computer Science, Information Systems
Jia Liu, Tianrui Li, Zhong Yuan, Wei Huang, Peng Xie, Qianqian Huang
Summary: This paper proposes a novel Symbolic Aggregate approXimation based Data Fusion model (SAX-DF) to accurately detect dangerous driving behavior. The method considers multiple influencing factors and designs a Positive Danger Mapping algorithm. Experimental results show that the proposed method is more effective than benchmark methods.
INFORMATION SCIENCES
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Mariem Taktak, Slim Triki
Summary: This paper presents an experimental evaluation of the SAX-based preprocessing technique for character recognition using gyroscope data. SAX is a popular and efficient symbolic dimensionality reduction technique, which is suitable for character recognition applications on low-cost mobile devices.
INTELLIGENT SYSTEMS DESIGN AND APPLICATIONS, ISDA 2021
(2022)
Article
Energy & Fuels
Zhaoyu Li, Yufan Zhang, Qian Ai
Summary: In this work, a multi-metric evaluation procedure is proposed to select suitable customers for demand response programs. The method utilizes clustering and analysis of load profiles to assess the variability, sensitivity, and quantity of electricity consumption, and models the dynamics using a one-step Markov chain.
SUSTAINABLE ENERGY GRIDS & NETWORKS
(2023)
Article
Engineering, Chemical
Gilseung Ahn, Hyungseok Yun, Sun Hur, Siyeong Lim
Summary: The study proposes a method to address the data insufficiency issue in RUL models by generating time-series data, converting training time series into alphabetical strings, identifying patterns, and generating new time series. Experiments show that this data-generation model can help prevent overfitting in RUL prediction models.
Article
Computer Science, Information Systems
Jia Liu, Wei Huang, Hao Li, Shenggong Ji, Yajun Du, Tianrui Li
Summary: Dangerous driving behaviors are the main cause of most traffic accidents, and detecting these behaviors accurately is a crucial research area in Intelligent Transportation System (ITS). This paper proposes a Symbolic Aggregate approXimation (SAX) and Long Short-Term Memory (LSTM)-based Attention Fusion method (SLAFusion) to improve the detection of dangerous driving behavior.
INFORMATION SCIENCES
(2023)
Article
Green & Sustainable Science & Technology
Xinlei Zhou, Wenye Lin, Ping Cui, Zhenjun Ma, Tishi Huang
Summary: This paper introduces an efficient data mining strategy to analyze the operational data of ground source heat pump systems, revealing inefficient operation patterns and energy conservation opportunities.
SUSTAINABLE ENERGY TECHNOLOGIES AND ASSESSMENTS
(2021)
Article
Engineering, Multidisciplinary
Bo Wang, Yi Ning, Yahu Zhang
Summary: Benefitting from the rapid development of artificial intelligence, the end-to-end fault diagnosis mode based on deep learning has become one of the most potential research directions. In this work, a novel fault diagnosis scheme for rolling bearings based on symbolic aggregate approximation (SAX) and a convolutional neural network with attention mechanism is developed to overcome the limitations of high-dimensional data redundancy and subjective feature extraction. Experimental results show that the proposed method achieves a fault classification accuracy of 98.4% on a common bearing fault data set.
MEASUREMENT SCIENCE AND TECHNOLOGY
(2022)
Article
Chemistry, Multidisciplinary
Jiancheng Yin, Yuqing Li, Rixin Wang, Minqiang Xu
Summary: This paper proposes an improved similarity trajectory method that can directly use monitoring data under multiple operating conditions for equipment life prediction without extracting the degradation trend of the monitoring data. Experimental results demonstrate that the proposed method can effectively improve the accuracy and effectiveness of life prediction.
APPLIED SCIENCES-BASEL
(2021)