Adaptive Architecture Research Articles

Two High Performance Global Tracking Composite Adaptive Controllers for Fully Actuated Euler-Lagrange Systems

Two adaptive global tracking controllers for fully actuated Euler-Lagrange systems with verifiable performance improvement with respect to existing designs are reported in the paper. The controllers are shown to ensure global exponential convergence under a weak interval excitation condition. Besides, one of the proposed controllers has a simple adaptive PID-like structure and—in contrast with the classical solution—its construction avoids additional filtering. We adopt a composite adaptation architecture, invoke a novel parameterization of the system dynamics and use a high performance estimation scheme recently reported in the literature. The validity of our claims is illustrated with extensive simulations of a two degrees-of-freedom robot manipulator operating in a realistic noisy environment.

High-Throughput Spike Detection in Greenhouse Cultivated Grain Crops with Attention Mechanisms-Based Deep Learning Models.

Detection of spikes is the first important step toward image-based quantitative assessment of crop yield. However, spikes of grain plants occupy only a tiny fraction of the image area and often emerge in the middle of the mass of plant leaves that exhibit similar colors to spike regions. Consequently, accurate detection of grain spikes renders, in general, a non-trivial task even for advanced, state-of-the-art deep neural networks (DNNs). To improve pattern detection in spikes, we propose architectural changes to Faster-RCNN (FRCNN) by reducing feature extraction layers and introducing a global attention module. The performance of our extended FRCNN-A vs. conventional FRCNN was compared on images of different European wheat cultivars, including "difficult" bushy phenotypes from 2 different phenotyping facilities and optical setups. Our experimental results show that introduced architectural adaptations in FRCNN-A helped to improve spike detection accuracy in inner regions. The mean average precision (mAP) of FRCNN and FRCNN-A on inner spikes is 76.0% and 81.0%, respectively, while on the state-of-the-art detection DNNs, Swin Transformer mAP is 83.0%. As a lightweight network, FRCNN-A is faster than FRCNN and Swin Transformer on both baseline and augmented training datasets. On the FastGAN augmented dataset, FRCNN achieved a mAP of 84.24%, FRCNN-A attained a mAP of 85.0%, and the Swin Transformer achieved a mAP of 89.45%. The increase in mAP of DNNs on the augmented datasets is proportional to the amount of the IPK original and augmented images. Overall, this study indicates a superior performance of attention mechanisms-based deep learning models in detecting small and subtle features of grain spikes.

Enterprise Architecture Design and Implementation for IoT Integration in Manufacturing Electrical Panels

Internet of Things technology has transformed manufacturing efficiency and optimization. Electrical panel manufacturing benefits from Internet of Things for better functionality, predictive maintenance, and smoother operations. This study examines the design and implementation of an Enterprise Architecture strategy for seamless Internet of Things integration in electrical panel manufacturing. This research aims to explain Enterprise Architecture and use it as a framework for Internet of Things integration in electrical panel manufacturing. This study examines the complex relationships between Internet of Things components, their connectivity, and a broad Enterprise Architecture framework needed to organize their functionality. This integration uses Enterprise Architecture principles to optimize resource use, reduce downtime, and improve manufacturing efficiency. This effort involves analyzing existing infrastructure, identifying Internet of Things deployment points, and creating an Enterprise Architecture plan that meets business goals. This research emphasizes the need for close IT-operations collaboration to achieve a unified vision and smooth Internet of Things integration. This research addresses Internet of Things implementation challenges in manufacturing, including security, data interoperability, and scalability. Strong governance and adaptable architecture are stressed to address these challenges within an Enterprise Architecture framework. This research aims to help electrical panel manufacturers harness the transformative power of the Internet of Things. Strategic Enterprise Architecture helps businesses navigate complexity, leverage Internet of Things, and create a more agile, connected, and optimized manufacturing landscape.

Assessment of Impacts and Sustainable Practices in Sediment Management via the Verbois Dam

This research thoroughly investigates the multifaceted challenges of sediment management at the Verbois Dam, a project that intersects technical, financial, social, and environmental dimensions. Recognizing the intricate nature of this task, the Services Industriels de Geneve (SIG) adopted a holistic approach that transcends traditional hydraulic methodologies. The primary aim was to develop a comprehensive decision support tool that encompasses two integral components: an analysis grid for assessing effects and a grid for evaluating durability. These elements work in tandem to facilitate a detailed scrutiny and assessment of current sediment evacuation tactics. Additionally, the tool’s adaptable architecture permits the exploration and refinement of novel management strategies. The findings of this study endeavor to make a substantial contribution to the sustainable handling of sediment in dam operations, effectively marrying technical efficacy with wider environmental and social considerations.

Integrating asset-specific flood vulnerability assessments with value-based preservation processes to develop the Heritage Building Flood Robustness Toolkit

Climate change is posing a growing threat to cultural heritage buildings, and many cities lack the capacity to manage these risks. One significant challenge is flooding, which can cause significant damage to heritage buildings. A literature review identified gaps in current research on heritage adaptation, including: a lack of tools for assessing the vulnerability of asset-specific cultural heritage buildings to flooding and providing recommendations for adaptation; limited examples of successful architectural adaptation of cultural heritage buildings to flooding impacts; and insufficient consideration of values, such as aesthetic, in adaptation tools. In response to these gaps, we developed a Heritage Building Flood Robustness Toolkit that incorporates both architectural and engineering considerations. The Toolkit is designed as a decision support tool to assist asset owners and key stakeholders, such as community groups, to make more quantified and informed decisions on heritage adaptation. The Toolkit includes three key advancements: an enhanced component-based methodology for individual cultural heritage building flood vulnerability assessments; improved recommendation of adaptation strategies aided by component-based risk assessments; and integration of user values, including aesthetic implications, into adaptation recommendations through adopting a Structured Decision-Making process. The Toolkit was tested on two buildings in Cockatoo Island, Sydney (Australia) with different materials and construction techniques. The results highlight the Toolkit's ability to support decision-making through more precisely assessed adaptation strategies based on asset-specific flood damage assessments and its sensitivity to user preferences, particularly aesthetic impacts from adaptation implementation. The Toolkit can be used as an instrument by planners, designers, and councils to enhance the protection of cultural heritage against flooding in the future.

An adaptive ensemble deep learning framework for reliable detection of pandemic patients

Nurses, often considered the backbone of global health services, are disproportionately vulnerable to COVID-19 due to their front-line roles. They conduct essential patient tests, including blood pressure, temperature, and complete blood counts. The pandemic-induced loss of nursing staff has resulted in critical shortages. To address this, robotic solutions offer promising avenues. To solve this problem, we developed an ensemble deep learning (DL) model that uses seven different models to detect patients. Detected images are then used as input for the soft robot, which performs basic assessment tests. In this study, we introduce a deep learning-based approach for nursing soft robots, and propose a novel deep learning model named Deep Ensemble of Adaptive Architectures. Our method is twofold: firstly, an ensemble deep learning technique detects COVID-19 patients; secondly, a soft robot performs basic assessment tests on the identified patients. We evaluate the performance of various deep learning-based object detectors for patient detection, examining implementations of You Only Look Once (YOLO), Single Shot MultiBox Detector (SSD), Region-Based Convolutional Neural Network (RCNN), and Region-Based Fully Convolutional Network (R-FCN) on a proprietary dataset comprising 32,668 hospital surveillance images. Our results indicate that while YOLO and VGG facilitate rapid detection, Faster-RCNN (Inception ResNet-v2) and our proposed Ensemble-DL achieve the highest accuracy. Ensemble-DL offers accurate results in a reasonable timeframe, making it apt for patient detection on embedded platforms. Through real-world experiments, our method outperforms baseline approaches (including Faster-RCNN, R-FCN variants, CNN+LSTM, etc.) in terms of both precision and recall. Achieving an impressive accuracy of 98.32%, our deep learning-based model for nursing soft robots presents a significant advancement in the identification and assessment of COVID-19 patients, ultimately enhancing healthcare efficiency and patient care.

Near-unity emission in zero-dimensional Sb(III)-based halides intervened by hydrogen bonds towards efficient solid-state lighting technology

Zero-dimensional (0D) metal halides have garnered considerable interest owing to their superb optical characteristics, and adaptable architectures. Two 0D organic-inorganic metal halides (OIMHs) (C20H20P)2SbCl5·(EA) ([C20H20P]+ = ethyltriphenylphosphine) and (C20H20P)2SbCl5·(IPA) were successfully synthesized and investigated using a simple solvent-evaporation approach. These OIMHs possess a similar structure, except for the solvent molecules in the lattice, however, their photoluminescence quantum yields (PLQYs) exhibit a considerable difference, with 99.82% in (C20H20P)2SbCl5·(EA) and 73.7% in (C20H20P)2SbCl5·(IPA). Bond-order and crystal orbital Hamilton population (COHP) calculations reveal that (C20H20P)2SbCl5·EA system exhibits stronger hydrogen bonds, leading to stronger supramolecular interactions, which suppress non-radiative transition. The experimental results and density functional theory (DFT) calculations suggest an increase in PLQY, which could be attributed to the stronger hydrogen bonding in (C20H20P)2SbCl5·(EA) resulting in strong interactions between the organic and inorganic components, good mobility, and improved exciton utilization. This study introduces a rational strategy for microstructure optimization, which builds upon solvent modulation to achieve stronger hydrogen bonding and thus improve the PLQY of 0D OIMHs.

Multidimensional characterization of particle morphology and mineralogical composition using CT data and R-vine copulas

Computed tomography (CT) can capture volumes large enough to measure a statistically meaningful number of micron-sized particles with a sufficiently good resolution to allow for the analysis of individual particles. However, the development of methods to efficiently investigate such image data and interpretably model the observed particle features is still an active field of research. When image data of particles exhibiting a wide range of shapes and sizes is considered, traditional image segmentation methods, such as the classic watershed algorithm, struggle to recognize particles with satisfying accuracy. Thus, more advanced methods of machine learning must be utilized for image segmentation to improve the validity of subsequent analyses. Moreover, CT data does not include information about the mineralogical composition of particles and, therefore, additional SEM-EDS image data has to be acquired. In this paper, micro-CT image data of a particle system mostly consisting of zinnwaldite–quartz composites is considered. First, an image segmentation method is applied which uses deep convolutional neural networks, in particular an adaptation of the U-net architecture. This has the advantage of requiring less hand-labeling than other machine learning methods, while also being more flexible with the possibility of transfer learning. Then, fully parameterized models based on vine copulas are designed to determine multivariate probability distributions of descriptor vectors for the size, shape, texture and composition of particles—allowing for the estimation and interpretable characterization of interdependencies between particle descriptors. For model fitting, the segmented three-dimensional CT data and co-registered two-dimensional SEM-EDS data are used. The models are applied to predict the mineralogical composition of particles, solely on the basis of particle descriptors observed in CT data.

Multi-criteria refrigerant screening and performance optimization comparison of a basic and composition-adjustable dual cooling source system

Climate variability and staff mobility have exacerbated the vulnerability of the indoor environment, which calls for air-conditioning units capable of adapting to the active regulation demand of the changing heat and humidity loads. The identification of the refrigerant constitutes an essential sector of the air-conditioning system design. A composition-adjustable dual cooling source (CADCS) chiller has been proposed utilizing the characteristics of inefficient condensation of the non-azeotropic mixture. Separated flows of the working fluid with flexible mass fraction and flow rate respectively treat the waste heat and moisture. Multi-criteria screening mechanism for the working fluid is established involving: i) application field. The critical temperature is better restricted within 300 K to 400 K for application in air-conditioning. ii) environmental impact. The GWP of the candidate cannot exceed 800. iii) cycle architecture adaptability. The boiling point difference preferably exceeds 20 K. iv) liquid separation potential. The obvious temperature glide and substantial separation range allow for the integration of liquid separation technology. Thermodynamic performance optimization indicates that propane/isobutane outperforms all the candidates at a mass fraction of 0.5:0.5 characterized by obvious temperature glide and close to temperature change of the heat transfer medium. The maximum coefficient of performance (COP) is found to be up to 5.2 for the BDCS cycle and 4.42 during separation at vapor quality 0.436 for the CADCS cycle. The exergy loss of the evaporator is reduced by 17.1 %-73.7 % while the exergy efficiency is improved by 3 %-59.7 % due to temperature match refinement in the CADCS cycle. The refrigerant charge in the CADCS cycle can be saved by 5.54 %-15 %. The accumulating waste heat demands a lower vapor quality during separation, while the moisture load barely exerts an effect. Moreover, the effect of the subcooling degree on the cycle performance is analyzed and suggested at 1 °C. The CADCS cycle constitutes a promising prospect of dynamic treatment for thermal and humidity load in the air-conditioning room due to its flexible function and improved heat transfer match.

Safety Needs Management for People with Disabilities in Contemporary Cities

Abstract Adapting city space to the needs of disabled people is an increasingly important area of activities of local government authorities. Well-thought-out architecture of buildings, road construction and adaptation of broadly understood urban transport ensure free use of city life for all users. The aim of this study is to analyze chosen urban planning solutions tailored to the needs of disabled people. Research methods of this paper are problem-oriented theoretical studies, case studies.

Adaptasi Nuansa Lokal pada Proses Perancangan Bangunan Komersial di Bali

Title: The Adaptation of Local Architecture in the Commercial Building Design Process in Bali
 The local Balinese architecture is still being developed and maintained simultaneously in Bali, incorporating elements of history and space. Over time, modernization has impacted the existence of local Balinese architecture in its manifestation. The local government responded to the act of preserving culture by establishing a Bali Regional Regulation, which regulates the design of architectural buildings with local Balinese nuances. The existence of this culture has become the main attraction in Bali, and it requires support from the spaces or commercial buildings. Therefore, the image of commercial buildings is regarded as a manifestation of the existence of local Balinese architectural culture. This paper focuses on the building's appearance, with two object location points as the case study to determine how to adapt local nuances to different typologies, namely Pahdi Coffee Kintamani and Benoa Beach Club, as objects that can illustrate the process of design acculturation. The research evaluated the design against existing benchmarks to obtain the adaptation process used in commercial building design. The method used was qualitative descriptive. As a result, the design of Pahdi Coffee Kintamani and Benoa Beach Club can fit with the function needed and adapt the local Balinese architecture through the appropriate process.

The Contribution of Non-Malay Architects to the Development of National Architecture Identity in Malaysia

The idea of adapting Malay traditional architectural values has been gazetted as one of the resolutions for National Architecture Identity as presented in ‘Seminar Ke Arah Identiti Kebangsaan Dalam Seni Bina’ in 1981. The resolution heavily influenced the adaptation of traditional Malay architecture, motives, and artefact onto modern architecture as a form to inculcate a singular and united national identity. A lot of research and documentation on good and meaningful architecture embracing the ideals of National architecture identity through the worldview and perspectives of Malay scholars and architects have been documented through books, journals, magazines and extensively discussed during seminars, symposiums, and discourses. Ironically, the contributions of Non-Malay architects adapting Malay symbolism to enunciate or articulate national pride and identity, either due to policies or interest, are minimally discussed and highlighted. This publication attempted to document the contributions of non-Malay architects to the national architecture identity construct. It is crucial to shape future philosophies, theories, criticisms, and thoughts on the overall built environment in developing a comprehensive construct of national architectural identity that does not necessarily rely on ethnocentric criteria based on a single race dominance. The research allows us to expand and progress the local architecture narrative by including diverse voices and thoughts, especially of those who have contributed extensively in the eras of nation-building. This inclusive and diverse input will, in turn, inform and expand the discourse of national identity and unity.

Computational Segmentation of Timber Slabs with Free Column Placement

Modular floor slabs must be subdivided into prefabricable, transportable segments. This slab segmentation process conventionally uses a rectangular pattern, particularly for timber buildings. Regular segmentation patterns and strict column grids are ideal for rectangular building shapes, but restrict timber buildings to only some architectural uses, and are unideal for urban infill. Unfortunately, planning and constructing multi-storey wood buildings without a strict grid is still challenging. There is therefore a conflict between the desired column placement and the constraints imposed by building systems. This article investigates novel methods for segmenting timber floors supported by irregular column layouts. It proposes six different segmentation methods that are informed through Co-Design by structural, material waste, and transportation requirements. Co-Design allows for the direct integration and automated feedback of such diverse criteria into the early building design phase. These methods are based on three well-known computational approaches: Single-Objective Optimisation, Parametric Modelling, and Agent-Based Modelling. They could also be applied to other non-timber prefabricated floor systems. The segmentation methods are demonstrated on two example floor slabs with irregular column layouts, one with a rectilinear and the other with an irregular outline. The methods are compared using quantitative proxies for cost, fabrication time, architectural adaptability, and assembly complexity. More benchmark testing is needed, but initial results showed that the most efficient segmentations cannot adapt to irregular layouts, emphasising the need for a more adaptable approach to modular timber construction.

Managing Student Mobility in Cameroon’s University Ecosystem: A FORM/BCS Approach

Nowadays, student mobility cannot be avoided in Cameroon university ecosystem. This phenomenon has many causes. Certain students live with their parents who are civil servants and they have to move with them when they are sending to a different region; another situation is link to universities newly created in cities where the cost of the live is better than in the home university city and students prefer move to these new universities. Since student circuit is not the same in each university, it is difficult to find his level in the new university and which courses he has to follow in order to complete his training. This problem is crucial in Cameroon university ecosystem and we tackle it in this paper. The Feature Oriented Reuse Method with Business Component Semantics (FORM/BCS) is a software domain engineering method that has been proposed to design an adaptable architecture for systems belonging in a same business domain. In this work, we apply the FORM/BCS method to manage student mobility in Cameroon university ecosystem. The result of this work is a management model that allows once a student gets an enrolment in a new university, to transfer credit from the student home university to the host one.

Design and Implementation of a Self Adaptive Architecture for QoS (SAAQ) in IoT based Wireless Networks

Design and Implementation of a Self Adaptive Architecture for QoS (SAAQ) in IoT based Wireless Networks

IBD sharing patterns as intra-breed admixture indicators in small ruminants

In this study, we investigated how IBD patterns shared between individuals of the same breed could be informative of its admixture level, with the underlying assumption that the most admixed breeds, i.e. the least genetically isolated, should have a much more fragmented genome. We considered 111 goat breeds (i.e. 2501 individuals) and 156 sheep breeds (i.e. 3304 individuals) from Europe, Africa and Asia, for which beadchip SNP genotypes had been performed. We inferred the breed’s level of admixture from: (i) the proportion of the genome shared by breed’s members (i.e. “genetic integrity level” assessed from ADMIXTURE software analyses), and (ii) the “AV index” (calculated from Reynolds’ genetic distances), used as a proxy for the “genetic distinctiveness”. In both goat and sheep datasets, the statistical analyses (comparison of means, Spearman correlations, LM and GAM models) revealed that the most genetically isolated breeds, also showed IBD profiles made up of more shared IBD segments, which were also longer. These results pave the way for further research that could lead to the development of admixture indicators, based on the characterization of intra-breed shared IBD segments, particularly effective as they would be independent of the knowledge of the whole genetic landscape in which the breeds evolve. Finally, by highlighting the fragmentation experienced by the genomes subjected to crossbreeding carried out over the last few generations, the study reminds us of the need to preserve local breeds and the integrity of their adaptive architectures that have been shaped over the centuries.

A multi-constrained domain adaptation network for remaining useful life prediction of bearings

The prediction of the remaining useful life (RUL) of bearings could reduce maintenance costs and prevent serious consequences. Due to the variable working conditions of bearings, domain adaptation methods for cross-domain RUL prediction have received extensive attention. However, existing research often focuses solely on aligning marginal distributions while overlooking the alignment of conditional distributions and fails to retain domain-specific information during the alignment process. To address these issues, we propose a multi-constrained domain adaptation (MCDA) method for cross-domain RUL prediction. MCDA method is based on an adversarial domain adaptation architecture, where the improved domain discriminator aligns the data features of different degradation stages separately to achieve the alignment of marginal distribution and simultaneously aligns the conditional distribution. The state discrimination module is introduced to align the marginal and conditional distributions while retaining the target domain-specific information. To confirm the effectiveness of the proposed method, we conducted RUL prediction tasks for bearings in various cross-domain scenarios using both publicly and self-built datasets. The experimental results showed that the MCDA method could better handle cross-domain RUL prediction tasks compared to existing methods.

An image-driven micromechanical approach to characterize multiscale remodeling in infarcted myocardium

Myocardial infarction (MI) is accompanied by the formation of a fibrotic scar in the left ventricle (LV) and initiates significant alterations in the architecture and constituents of the LV free wall (LVFW). Previous studies have shown that LV adaptation is highly individual, indicating that the identification of remodeling mechanisms post-MI demands a fully subject-specific approach that can integrate a host of structural alterations at the fiber-level to changes in bulk biomechanical adaptation at the tissue-level. We present an image-driven micromechanical approach to characterize remodeling, assimilating new biaxial mechanical data, histological studies, and digital image correlation data within an in-silico framework to elucidate the fiber-level remodeling mechanisms that drive tissue-level adaptation for each subject. We found that a progressively diffused collagen fiber structure combined with similarly disorganized myofiber architecture in the healthy region leads to the loss of LVFW anisotropy post-MI, offering an important tissue-level hallmark for LV maladaptation. In contrast, our results suggest that reductions in collagen undulation are an adaptive mechanism competing against LVFW thinning. Additionally, we show that the inclusion of subject-specific geometry when modeling myocardial tissue is essential for accurate prediction of tissue kinematics. Our approach serves as an essential step toward identifying fiber-level remodeling indices that govern the transition of MI to systolic heart failure. These indices complement the traditional, organ-level measures of LV anatomy and function that often fall short of early prognostication of heart failure in MI. In addition, our approach offers an integrated methodology to advance the design of personalized interventions, such as hydrogel injection, to reinforce and suppress native adaptive and maladaptive mechanisms, respectively, to prevent the transition of MI to heart failure. Statement of significanceBiomechanical and architectural adaptation of the LVFW remains a central, yet overlooked, remodeling process post-MI. Our study indicates the biomechanical adaptation of the LVFW post-MI is highly individual and driven by altered fiber network architecture and collective changes in collagen fiber content, undulation, and stiffness. Our findings demonstrate the possibility of using cardiac strains to infer such fiber-level remodeling events through in-silico modeling, paving the way for in-vivo characterization of multiscale biomechanical indices in humans. Such indices will complement the traditional, organ-level measures of LV anatomy and function that often fall short of early prognostication of heart failure in MI.

Towards Automation of the FORM/BCS Method

The software industry is facing more complex computer systems, with short development and sustainability issues. To deliver good software with these constraints, software reuse has become a central concept for minimizing design and realization costs. This study improves upon Feature-Oriented Reuse Method with Business Component Semantics (FORM/BCS), a software development method that produces adaptable architectures from reusable domain components. This is a promising method for reusable software assets and model creation. The objective of the FORM/BCS is to bring the industrial production chain to the software. This study proposes a model to automatically transform the FORM/BCS business subsystem component into a process business component. Two metamodels for business subsystems and process business components were developed. In addition, this study establishes correspondences between the source metamodel and target metamodel classes, transformation rules, and the instance of the source metamodel and generates the target metamodel instance. Detailed findings can help practitioners reduce software design costs and development time, and contribute to the advancement of knowledge in software engineering.

A Study of Folk Aesthetic Culture of Minnan Ancestral Hall in the Context of Big Data and Artificial Intelligence

In the past, people with the same family name gathered in a region were made into a family, and each family had its family temple for worshipping ancestors and remembering their forefathers, and such family temples with special significance were called ancestral halls. The southern Min region has experienced more hardships compared to the northern region, so it pays more attention to the family and places an important place on the transmission of blood ties. The values of people are subtly changed in this environment, and through the inculcation of generations, the people of southern Fujian internalized the values inherited from this architecture of ancestral halls into their aesthetic consciousness. The purpose of this study is to explore the impact of technological development on the evolution of regional aesthetic characteristics in the context of Big data and Artificial intelligence, starting from the folk aesthetic culture of Minnan ancestral halls. The architecture of Minnan ancestral halls is characterized by the adaptation of ancestral hall architecture to the climate of Southern Fujian, the unique architectural framework of ancestral halls, the use of vibrant colors in Minnan ancestral halls, and the Minnan ancestral halls. The art of stone carving and aesthetic literacy are two major aesthetic features of Minnan's ancestral halls. The number of Minnan ancestral halls is huge, and to analyze their aesthetic characteristics it is necessary to use the ability of Big data storage to analyze massive data and to extract some representative ancestral hall buildings from the database of Minnan ancestral halls for analysis. The various aesthetic features of the database are clustered and analyzed through Big data analysis technology to summarize the representative aesthetic features of Minnan ancestral halls. At the same time, some people in the Minnan area were selected as the research subjects, and the data related to the folk aesthetic consciousness of the people were obtained using a questionnaire survey, and then the "aesthetic elements" and "aesthetic consciousness" were input into the artificial intelligence training model by combining Big data and Artificial intelligence technology. The relationship between elements is explored through artificial intelligence decision-making. Finally, this conclusion is illustrated in four dimensions. From the dimension of color, the residents of Minnan mostly use relatively bright colors to decorate their houses, which is consistent with the use of red and black colors in Minnan's ancestral halls. In terms of shape, the architectural style of Minnan ancestral halls also influenced the development of later buildings, some of which evolved from Minnan ancestral halls as prototypes. In terms of materials, the Minnan ancestral halls were built with a combination of wood and stone, while later generations only increased the proportion of stone. In terms of culture, the development of folk aesthetic consciousness is echoed by the core values of blood and clan ties transmitted by the Minnan ancestral halls left.