Dalarna University's logo and link to the university's website

This paper follows the principles of the Big Data Analysis (BDA) framework to construct a set of experiments using a photovoltaic (PV) system dataset, providing an empirical evaluation of these principles. We propose a novel two-dimensional approach to improving time-series clustering and anomaly detection. Firstly, we enhance the DBSCAN clustering algorithm with LSTM-based residual forecasting (LSTM+DBSCAN) to capture temporal dependencies. Secondly, we incorporate incremental data ingestion to assess the scalability and adaptability of the framework. Our methodology builds upon the Approximation Computation Approach (ACA), which emphasizes a Pattern-Method-Assertion (PMA) framework for data analysis. By leveraging LSTM to extract meaningful temporal patterns and DBSCAN for clustering, we validate and extend the principles of ACA to the context of time-series anomaly detection. Our results demonstrate that the framework could provide guidelines for experiments in both the design of the methods and the datasets. © 2025 Elsevier B.V., All rights reserved.

The purpose of this paper is to explore the feasibility and effectiveness of utilizing Vector Autoregression (VAR) model to predict the moisture content of construction materials. The moisture content of construction materials has an important impact on their performance, durability and safety. Reasonable control of moisture content can prolong the service life of materials, improve the quality and efficiency of construction, reduce energy consumption, lower economic costs, and improve environmental adaptability. However, traditional water content prediction methods mostly rely on empirical data and simple statistical models, which are difficult to accurately capture the complex relationship between multiple variables. For this reason, this paper proposes a prediction method based on the VAR model in order to overcome the limitations of the traditional methods. Experimental results show that the VAR model exhibits high accuracy in predicting the water content of construction materials, and its MAE, RMSE, and MAPE are better than those of the commonly used deep learning models such as CNN and LSTM. This may be attributed to the lower computational complexity of the VAR model and its effective capture of stable relationships. © 2025 IEEE.

Missing values is a crucial problem in the area of big data analysis, which hinders data integrity. Various regression methods have been employed for estimating missing values, but they exhibit significant prediction errors. To ensure the integrity of data collected from a wood sensor monitoring system and address the issue of data loss and anomaly, we propose a missing value estimation method based on the random forest regression model. This study focuses on the environmental data, including temperature, relative humidity, and absolute humidity surrounding the wood subjects. We simulate a number of methods on the data for comparison purpose. The experiment results in terms of prediction performance indicate that the random forest regression model algorithm we developed for estimating moisture content's missing values yields favourable outcomes with consistently low estimation errors. © 2024 IEEE.

This study conducts an in-depth verification and analysis of the authenticity of information in WeChat group chats by integrating Latent Dirichlet Allocation (LDA) topic modeling with advanced Natural Language Processing (NLP) techniques such as XLNet and BERT. Leveraging LDA, the thematic structure of group chat content is revealed, and through the integration of NLP technologies like XLNet and BERT, a comprehensive analysis of the information is achieved. Experimental results demonstrate that our developed model performs exceptionally well in identifying the authenticity of information, confirming the effectiveness of this method in the domain of social media information verification. This research not only deepens our understanding of the authenticity of information in WeChat group chats but also provides a more effective tool for social media platforms to detect and prevent the spread of false information. It opens up a new perspective on social media information authentication research and points out future research directions. © 2024 IEEE.

This study explores the complex relationship between wood moisture content and environmental factors, temperature and relative humidity. Utilizing a novel Autoregressive Polynomial Regression Model (APRM), data from sensors placed in reconstituted bamboo and pine planks at various positions were analyzed. The APRM, adept at handling polynomial and interaction terms, revealed a nuanced, non-linear relationship between moisture content and environmental conditions. The research findings underscore significant material-specific differences in response to environmental changes. This study not only contributes to the understanding of wood-environment interactions but also demonstrates the efficacy of APRM in environmental science, providing a foundational approach for future research in this field. © 2024 IEEE.

Building ventilation system needs to be controlled in a smart way to maintain indoor air quality while reducing energy use. Although many demand-controlled methods have been developed, the design of ventilation schedule has to be customized depending on local climate, occupant behavior and system capacity. This article introduces an easy-to-use control strategy based on mathematical modeling, clustering and genetic algorithm. Experimental results improve the performance of current system in an example house and provide a data-driven framework. © 2024 Elsevier Inc. All rights are reserved.

To maintain the life of building materials, it is critical to understand the hygrothermal transfer mechanisms (HTM) between the walls and the layers inside the walls. Due to the extreme instability of weather data, the actual data models of the HTM—the data being collected for actual buildings using modern sensor technologies—would appear to be a great difference from any theoretical models, in particular, for wood building materials. In this paper, we aim to consider a variety of data analysis tools for hygrothermal transfer features. A novel approach for peak and valley detection is proposed based on the discrete differentiation of the original data. Not to be limited to the measure of peak and valley delays for HTM, we propose a cross-correlation analysis to obtain the general delay between two daily time series, which seems to be representative of the delay in the daily time series. Furthermore, the seasonal pattern of the hygrothermal transfer combined with the correlation analysis reveals a reasonable relationship between the delays and the indoor and outdoor climates. © 2023 by the authors.

Due to the uncertainty and inconsistency of measurement data from multiple sensors in the same space, a multi-sensor data fusion algorithm is used to fuse the measurement data of multiple nodes. We propose a multi-Bayesian estimation method for fusing multi-sensor data, and combine Bayesian estimation with ARIMA model to predict the ambient temperature of bamboo and wood building materials. It can utilize the redundancy of data to reduce this uncertainty and improve the reliability of subsequent predictions. © 2023 ACM.

This review aims to comprehensively assess and synthesize the existing literature on the use of data-driven methods for studying hygrothermal transfer in building exterior walls. The review is conducted by an exhaustive search strategy to identify relevant articles from Web of Science and Scopus databases. There are 20 eligible studies included in this review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol. The most used data-driven methods are traditional neural networks, such as Multi-Layer Perceptrons and 2D Convolutional Neural Networks. Results suggested that neural network models hold potential for accurately predicting hygrothermal attributes of building exteriors. However, a conspicuous gap in the literature is the absence of studies drawing direct comparisons between data-driven methodologies and conventional simulation techniques. © 2023 ACM.

In the era of big data, it is indispensable to apply data science and technology for big data analysis to solve the big data problems. With the advancement of the big data technologies, we are also facing many problems when dealing with the big data and their studies. It is obvious that big data become "bigger"and "bigger", more complex than before, with a good number of attributes and features in various formats and styles. On the other hand, many data analysis techniques have been proposed for various application domain problems in different purposes. This worsens the situation of choosing an appropriate method for a right problem of right data. In this paper, the author intends to propose an approximation approach toward this problem, through discussing the ways of identification of patterns of the original data, be they of data features or analysis methods. The author attempts to apply the idea to a case of fault detection of a household photovoltaic system. © 2023 IEEE.