Advances in Smart Vehicular Technology, Transportation, Communication and Applications

The deadline for goods delivery represents a critical milestone in railway freight transportation, reflecting the railway’s unwavering commitment to shippers. Prompt delivery is not only a matter of punctuality but also a testament to the industry’s dedication to maintaining high-quality service and competitiveness. Especially in the complex realm of through train operations, where trains seamlessly transition between stations and operational sections, timely delivery becomes an even more critical factor. This study, centered on the concept of streamlined integration, delves into the nuances of two pivotal scenarios: the intricate connection of through trains at a single technical station and the complex interlinking across multiple operational sections. By formulating 0–1 programming models that incorporate practical constraints like operating time standards and delivery time limits, we aim to optimize these scenarios and enhance the efficiency of railway freight transportation. Through validation using data from existing literature examples, we demonstrate the practical significance of our approach in improving the overall performance of the railway industry.

It brings a challenge to solve the passenger service-like railway container transportation optimization, because it is NP-hard. A mixed-integer linear programming (MILP) model as well as an improved column generation algorithm with comprehensive using column and row generation techniques is proposed to optimize the train route, frequency, stop schedule simultaneously. A real-world instance of the backhaul railway network between China and Europe for China Railway Express is used to test our model and algorithm. Computational results show that our approach can solve these problems within reasonable solution time and small optimality gaps.

Due to the high frequency of departures, Urban Rail Transit (URT) has emerged as the top choice for transportation among numerous individuals. However, when there is a disruption, the shorter departure interval poses a huge challenge to train rescheduling. Disruption may cause large-scale train delays or even paralysis of the entire line operation, seriously affecting the travel of passengers. This paper conducts a thorough study on the train rescheduling problem during disruption in URT systems. We implement holding and short-turning strategies and propose a NIP model aimed at minimizing passenger travel costs and deviations from the original train timetable, with the objective of obtaining a better train rescheduling timetable in a short time. We propose an iterative optimization algorithm utilizing Adaptive Large Neighborhood Search (ALNS), and we perform a case study using operational data from the Chengdu metro to validate the efficacy of the train rescheduling timetable. The result indicates that our approach decreases passenger travel costs by 15.50% and reduces timetable deviations by 16.47% when compared to the initial train timetable.

As urbanization accelerates, the subway, with its unique advantages of high speed and high capacity, plays an increasingly important role in the process of urban development. However, with the accelerating pace of subway construction in the country, subway accidents have occurred frequently in the recent years, exposing the vulnerability of subway operations. This paper analyzes relevant research at home and abroad to establish the concept of subway station operation vulnerability based on the attributes of exposure, susceptibility, and adaptability. Based on this concept and operational accident statistical analysis, an evaluation index system is established. Combining the theories of rough sets and extension sets, a multi-level extension evaluation model is constructed. Finally, the multi-level extension evaluation model is applied to conduct case evaluations and provide recommendations.

In order to align with the varying demands of market over time and the developmental needs of the railway, there has been a significant increase in the frequency and scale of adjustments to the train diagram. The old and new alternate passenger train diagram represents a dynamic operational plan that synergizes the operation of trains with the utilization of rolling stock during alternate periods. This paper introduces an integrated cooperative optimization time–space network for the new and old alternate train diagrams. It treats the utilization of rolling stock during alternate periods and resolution of conflicts between the old and new routes as a unified description of the optimization problem for the operation paths of passenger train stock in the spatio-temporal network of alternate periods. Building upon this, we set the objective as the cost of rolling stock utilization during alternate periods, with rolling stock utilization rules and train running intervals as constraints. And then we establish a comprehensive cooperative optimization 0–1 integer planning model and design a Lagrangian heuristic algorithm.

With the gradual formation of the development pattern of multi-system and network of regional rail transit, the operation mode between each system of regional rail transit is gradually moving from single to collaborative. In this paper, we focus on the dynamic optimization of train operation diagram for the cooperative development of multi-system regional rail transit, take the transfer process between railways and urban rail transit as the research object, take the minimum waiting time of passengers as the optimization objective, construct an optimization model with the constraints of the departure time and the interval of urban rail transit trains, solve the model by using the DQN optimization algorithm, and design the case study to verify the model. The results of the case study show that the total waiting time of passengers is 156.48 h, which is better than the total waiting time of 174.65 h under the equal interval, and the optimization rate is 10.4%, which proves the validity of the model, by optimizing the departure interval of the train schedule under the conditions of dynamic passenger flow through real-time decision-making.

The development of the collaborative operation mode of multi-standard rail transit system formed by urban rail transit, municipal railway, intercity railway, and trunk railway has become the key to promote the integrated development of regional transportation. The core task of multi-standard regional rail transit coordination is to realize the dynamic preparation of train operation scheme under the background of different standard cooperative operations in the context of dynamic passenger flow demand. This paper with the system rail transit operation enterprise operation benefit maximization and passenger travel cost minimization as the target function, collaborative optimization of departure frequency, stop scheme elements, considering the capacity matching degree, train stops, minimum departure frequency, passenger flow demand and transportation capacity limit constraints, build multi-system coupling train operation scheme dynamic optimization model, and using the principle of genetic algorithm, the design algorithm to solve. The calculation results demonstrate that after optimizing the train operation scheme of the multi-standard rail transit system, the operational capacity provided by the train operation plans for lines L and S aligns perfectly with the transfer passenger flow at stations S1 and S3 and other rail transit systems, falling within the optimal operational capacity matching range. This validates the effectiveness of the model and algorithm. The research findings offer a decision-making approach for dynamically preparing train operation schemes in multi-system coupling scenarios.

With the continuous development of rail transit in China, the coordinated development of different standard rail transit systems has become an irreversible trend. The core of multi-standard rail transit transportation organization lies in devising train operation plans that achieve mutual coordination and capacity matching. Existing optimization research on train operation plans mainly focuses on optimizing from the perspectives of enterprises or passengers, with the research direction targeting single-standard train operation plans. Therefore, this study coordinates and optimizes multi-standard train operation plans based on demand and capacity matching, aiming to maximize comprehensive capacity matching, enhance economic benefits for operating departments, and minimize passenger travel costs. A coordinated optimization model for multi-standard rail transit train operation plans is established and solved using the ideal point method. Finally, using a certain region's multi-standard rail transit as an example, the optimal operation plan is determined. The results indicate that the deviation between the comprehensive capacity matching and optimal matching of the optimized train operation plan has been reduced by 28.1%, leading to improvements in economic benefits for operating departments and passenger travel costs.

The traditional optimization and adjustment process of train operation plan is relatively lagging behind changes in passenger flow. When the daily passenger flow changes, the traditional process of optimizing and adjusting train operation plan cannot timely meet passenger flow demands while having minimal changes in passenger transportation and service organization. Therefore, this study adjusts the train operation plan by minimizing adjustments while meeting passenger flow demands. It constructs a two-layer adjustment model for high-speed train operation plan based on passenger flow demands: The upper layer aims to adaptability and optimal adjustment weighting, and the lower layer focuses on minimizing the total impedance of the train service network. An annealing-evolution algorithm is designed to solve the train operation plan constructed in this study. Finally, by building a simple high-speed railway line and using initial OD passenger flow data and initial train operation plan, the model and algorithm's effectiveness are validated by adjusting the train operation plan based on two fluctuations of OD passenger flow using the adjustment model.

The new and old alternate train diagrams, serving as a pivotal transitional operational scheme bridging current and forthcoming schedules, are of paramount importance in ensuring a seamless transition of the railway transportation system from its current state to an updated one. This paper extensively examines the characteristics of transitional train diagrams, selecting indicators from multiple dimensions such as transport efficiency, orderliness, stability, and passenger service quality to construct a comprehensive evaluation framework for these transitional schedules. Subsequently, the entropy method is employed to precisely reflect the discriminative capability of each indicator, followed by the assignment of appropriate weights. Furthermore, through the integration of the Technique for Order of Preference by Similarity to Ideal Solution approach, the resemblance between these indicators and the ideal solution is assessed, thereby computing the comprehensive scores of each diagram plan and evaluating the quality of the transitional train diagrams based on these scores. Finally, the feasibility of the evaluation methodology is validated through small-scale numerical illustrations.

This article explores the widespread and practical application of computer operating systems in the field of transportation and communication security, emphasizing the key role of operating systems in resource management, process management, memory management, file system management, device management, interrupt handling, and security. Computer operating systems play an important role in the field of communication security, effectively resisting various network threats and ensuring the security and stability of information. In transportation and communication security, operating systems provide important guarantees to ensure the safety and reliability of communication systems. By implementing a series of security measures, such as user authentication, multi-factor authentication, access control, strict security auditing, and patch management, malicious attacks and data leaks can be effectively prevented, ensuring data security in the transportation and communication industry. The article also introduces the application of UOS system in the field of transportation and communication. The system adopts an autonomous operating system, ticketing terminals, servers, middleware, databases, etc., realizing the independent software and hardware platform of the entire ticketing application. It has successfully passed the pilot verification of the entire ticketing process, breaking the long-term monopoly of foreign systems such as Windows and CentOS.

The operation plan of high-speed rail directly affects the competitiveness of passenger transport products, necessitating a sound evaluation scheme for timely adjustments. The evaluation of the existing operation plan mainly focuses on a single train, which cannot reflect the impact of high-speed rail network operation. From the perspective of complex network, a new concept of route matching is proposed. Taking a single train as the object, indexes are selected from three aspects: path matching, economic benefit, and transportation efficiency, and an evaluation index system for the benefit efficiency of the high-speed railway operation plan oriented to the road network is constructed. An evaluation model based on improved TOPSIS-factor analysis is proposed. Taking the high-speed rail network with the Beijing-Shanghai high-speed railway as the backbone for example, this paper optimizes the operation plan from the perspective of adjusting the train running path and provides some reference for the operation department.

This article primarily investigates the impact of using voice-controlled smartphones, main screen video play, and co-pilot screen video play on drivers’ driving time and collision probability in the era of artificial intelligence. Specifically, through Analysis of Variance (ANOVA), it was found that “driver's main screen play” significantly affects driving safety. Nonetheless, “driving while making a phone call” and “driving with the co-pilot screen play” do not have a significant impact on driving safety. Moreover, the impact of the former is markedly less than the latter.

Estimating the duration of lithium-ion battery testing is crucial for devising battery testing plans and optimizing resource allocation. However, there is currently a lack of mature and practical estimation models. In this chapter, we propose a method for estimating the remaining charging time of lithium-ion batteries by integrating a Convolutional Neural Networks-Gated Recurrent Unit (CNN-GRU). This method first extracts features from the raw data of the battery testing process and then combines the advantages of convolutional neural networks in extracting spatial features and gated recurrent units in capturing temporal features. It constructs a non-linear duration estimation model considering multi-source heterogeneous features, ultimately forming a precise and efficient solution for duration estimation. Experimental results demonstrate that the CNN-GRU model constructed in this study exhibits lower estimation errors compared to other algorithms, with an average absolute percentage error of less than 2%. This model can accurately predict the remaining charging time of batteries, providing robust support for optimizing and scheduling battery testing plans.

Standard curves play an important role in the accurate operation of instruments. Today, there are various methods to fit standard curves for instruments. In this paper, the lower-order and higher-order fittings of the microarray scanner standard curve are discussed separately, and we used the maximum relative error as the evaluation index. In the low-order fitting experiments, the Least Squares Algorithm (LSA) could not form an effective constraint on the relative error, which led to the high relative error in its fitting curve in the region of small values, but the Neural Dynamics Optimization Algorithm (NDOA) fitting curve reflected greater advantages in relative error. With the increasing computational power of the hardware, we cannot avoid discussing the higher-order fitting of the data, the NDOA appears to be a poor fit in the higher-order fitting, so we use the LSA (N = 7) with the constraint of minimizing the sum of squares of the errors (SSE) and the Multi-Layer Perceptron (MLP) with the relative error minimization to fit the data at higher order, respectively. Through the higher-order fitting experiments, we find that MLP outperforms LSA (N = 7) in terms of the maximum relative error, and MLP reduces the maximum relative error of the fitted curve to 0.89%. These results will provide some reference value for subsequent standard curve fitting studies of measuring instruments.

As one of the key technologies for ADAS, traffic sign detection and recognition can effectively obtain traffic sign information on the road such as warnings, instructions, and prohibitions, effectively improving driving safety while ensuring road safety. However, the characteristic expression of traffic signs changes in foggy and night scenes, making the task more challenging in complex scenes. In this paper, we propose a multi-source traffic sign dataset for complex situations, named ITT100K. The construction of this dataset effectively improved the generalization of the model. A SE-YOLOX specific integration model with attention mechanism is proposed. By using the attention mechanism to allocate weights to the output of three complex scenes, the model can focus on feature learning of specific scenes and improve the recognition accuracy of the model in complex scenes. Experiments on ITT100K and foggy driving dataset show that the proposed method has achieved significant improvement.

The Internet of Vehicles (IoV) establishes a comprehensive network that connects vehicles, pedestrians, and urban infrastructure elements. In the IoV paradigm, vehicles function as intelligent entities equipped with sensing platforms and computing facilities, facilitating connectivity with other vehicles, roadside units (RSUs), and cloud servers. The advancements in autonomous driving present challenges, with a primary emphasis on addressing security threats associated with the sharing of data among various IoV entities. Within the realm of IoV, a thorough analysis of the substantial big data generated by vehicles is imperative for making informed decisions and inferences. This paper proposes a secure framework for big data analytics (BDA), wherein the data collected from vehicles undergoes a secure analysis at the BDA center. The effectiveness of the proposed methodology is showcased through a comprehensive analysis of security, resilience against attacks, and big data analytics performance.

In order to address the issues of poor solution accuracy, convergence, and local optimality often observed in the firefly algorithm when solving multi-objective optimization problems, this study proposes a multi-objective firefly algorithm with dynamic reflection guidance and bi-dimensional variation (MOFA-DGM). MOFA-DGM incorporates Tent mapping during the population initialization stage to ensure uniform coverage of the entire feasible domain, thus improving the quality of the initial population. During the firefly position updating stage, a dynamic reflection guidance model is introduced, which uses both the elite solution and the current optimal solution from the external archive, along with an enhanced simplex method, to calculate the reflection solution that guides the movement of the fireflies. This approach enhances the convergence of the algorithm. Additionally, a two-dimensional variation mechanism is proposed to help the population escape local optima and prevent the formation of firefly clusters during late iterations, thereby enhancing the exploration capability of the algorithm. Experimental results comparing MOFA-DGM with recent multi-objective evolutionary algorithms demonstrate that MOFA-DGM effectively improves solution accuracy, enhances algorithm convergence, and improves the algorithm’s ability to search for optimal solutions compared to the other algorithms.

Fuzzy least squares projection twin support vector machine (FLPTSVM) fails to address the difficulty of parameter selection and the limitation of a single kernel function. In view of this, this paper proposes a hybrid kernel function fuzzy least squares projection twin support vector machine by wolf pack algorithm (WPA-HFLPTSVM). This paper designs a novel method to construct a hybrid kernel function by combining polynomial and Gaussian kernel functions. The wolf pack algorithm is selected to use the classification accuracy as the fitness value for comprehensive optimization of the kernel parameters and penalty parameters of the hybrid kernel function in order to obtain the optimal combination of parameters to improve the classification performance. Experimental results demonstrate that the overall performance of the hybrid kernel function outperforms other kernel functions; Compared to the classical SVM algorithm, HFLPTSVM exhibits superior classification performance and generalizability; by utilizing the WPA, we are able to search for optimal parameter combinations for HFLPTSVM.

In this paper, a hybrid DE-GOA algorithm is proposed. In the exploration and exploitation stages of the gannet optimization algorithm (GOA), the idea of differential evolution (DE) algorithm is introduced. To test the availability of the proposed hybrid algorithm, this paper makes use of the CEC2013 benchmark functions to compare 30D, 50D, and 100D dimensions with five swarm intelligent optimization algorithms. The experimental results of the proposed hybrid algorithm in this paper are more excellent than the compared optimization algorithms. Therefore, the proposed algorithm has stronger optimization ability and competitiveness.

To establish the optimal predictive model for water inflow in mines, this study is based on the data of water inflow over 650 days at the 1303 working face of the Li Lou Coal Mine. Descriptive statistical graphs are generated to examine the central tendency, dispersion, and distribution of water inflow, along with interpretation based on foundational geological data. The results indicate that the water inflow in the study area exhibits a continuous decrease trend within fluctuations. This feature reflects that the water source for the working face primarily derives from aquifers dominated by static storage and is unrelated to precipitation. Reconstruction parameters were determined based on the C–C method, and phase space reconstruction was conducted. A model was constructed utilizing the weighted first-order local method, resulting in a maximum Lyapunov exponent calculation of 0.24839, with an effective prediction duration of 4 days. Within this effective prediction duration, the average prediction error was 1.74%, indicating good predictive performance; after exceeding the effective prediction duration, the accuracy rapidly decreased, with an average error of 11.12% and a maximum error of 16.13%. These results indicate that applying chaotic methods to short-term prediction of mine water inflow demonstrates high precision, and the predicted outcomes hold practical application value.

Jincang embroidery, a traditional skill in Quanzhou, Fujian, China, has Litchi Jump, Jincang Convex embroidery, Dragon Scales Overlapping Armor, and gilt embroidery stitches as its main forms of artistic expression. These embroideries are unique in style and visually exquisite, but the technological process is complex, time-consuming, and needs to be done by hand existing research on the transfer of embroidery styles is relatively scarce and does not cover the unique artistic expression of Jincang embroidery. In view of this, a convolutional neural network algorithm suitable for the transfer of Jincang embroidery styles is proposed. First, the convolutional neural network (CNN) VGG-19 style transfer algorithm model was utilized, regarding the Jincang embroidery texture image as a style image. Second, image style features are captured by constructing a Gram matrix. Then the L-BFGS algorithm is applied for iterative optimization to finally generate images with the texture image style of Jincang embroidery. The quality of the migration results was evaluated by PSNR and SSIM. The results show that the images of irregular dot patterns and hand-painted flower patterns fused with Jincang embroidery are of higher quality. Therefore, applying computer vision technology can innovate Jincang embroidery design, so that the surface texture characteristics of Jincang embroidery are transferred to the design of textile fabric patterns, not only for the Jincang embroidery to inject new vitality but also for the creation of textile fabric pattern design to provide a new way of thinking.

In semi-supervised deep learning models, multiple deep neural networks are often utilized for optimization, significantly increasing the complexity of the system. Thus, the demand for lightweight networks has become vital for practical applications. Therefore, we propose a lightweight semi-supervised semantic segmentation method by using pixel-level contrastive learning. First, a lightweight segmentation network is proposed as the base architecture. Here, both standard convolution and atrous convolution are replaced by the depth-wise separable convolution and concentrated comprehensive convolution to reduce the channel number of the encoder and decoder, as well as the network load. Additionally, a residual link structure is incorporated to prevent information loss. Finally, to improve the utilization of available data, we explore a supervised pixel-level contrastive loss function to maximize the limited label information under a semi-supervised context. Comparison experiments on the Pascal VOC 2012 and Cityscapes datasets demonstrate that the proposed method outperforms many state-of-the-art semi-supervised semantic segmentation methods.

Genetic algorithms (GAs) are a class of optimization algorithms inspired by natural selection and genetics principles. They have gained significant popularity in various fields due to their ability to solve and optimize complex problems efficiently. This paper provides an overview of GAs, their working principles, applications in problem-solving, and optimization. The paper also discusses the advantages and limitations of GAs and provides insights into future research directions in this field.

In the realm of software development, effective collaboration among team members is crucial for success. Jira Software, a widely used project management tool, is highly regarded for its ability to facilitate team collaboration in planning, tracking, and release management. This chapter presents a detailed survey-based exploration of Jira’s collaboration potential and its impact on optimizing team workflows in these critical areas. Analyzing survey responses from software development teams across various industries reveals insights into how Jira is utilized, its perceived benefits, encountered challenges, and strategies to enhance its effectiveness in team collaboration. The findings from this survey provide valuable information on best practices, common pitfalls, and opportunities for improving the utilization of Jira for seamless team planning, tracking, and release management.

Multi-objective feature selection includes two objectives, which choose the relevant features for minimal number and obtain the better performance of classification accuracy. Binary Particle Swarm Optimization is intelligent optimizing algorithm with binary bits position and suitable for feature selection. A BPSO algorithm is proposed with a new velocity updating strategy with learning from neighbors in this chapter. Meanwhile, the novel algorithm utilizes a new transfer function for the updating of the bits position of particles. The multi-objective feature selection is done with the proposed Binary PSO, which employs the Pareto front to assist decision-makers for selection of the feature subset. The experiments are conducted with 10 datasets of the UCI repository with machine learning, and the experimental results demonstrate that the proposed BPSO outperforms three other BPSO according to the hypervolume indicator.

Urban water supply is a fundamental part of every city’s construction, and one of the pressing issues in water distribution networks(WDNs) is leak location. Leak location is important for reducing water supply losses and achieving sustainable water supply. The main means of leakage detection is the placement of pressure sensors in the network, and the location of possible leaks is analyzed through pressure changes. The problem of placing pressure sensors can be described as a constrained nonlinear integer programming problem, which is often computationally overloaded in real pipe networks due to their large size and high complexity using traditional optimization methods. So using meta-heuristic algorithms to solve this problem is a suitable approach. In this paper, the recently proposed gaining-sharing-knowledge-based (GSK) algorithm is used to solve the problem of optimizing the placement of pressure sensors with the aim of improving the success rate of leak location in pipe networks.

The Rafflesia Optimization Algorithm is a recently developed swarm intelligence optimization approach, drawing inspiration from Rafflesia’s natural biological principles. The three stages of the algorithm are the fruiting, planting, and pollination phases. To discover the best answer, the ROA algorithm searches locally in the first stage. By cutting down on the number of individuals, it increases execution efficiency and solution correctness in the second stage. In order to exit the local optimum, it conducts a global search in the third step. A major obstacle to the overall effectiveness of Wireless Sensor Networks (WSNs) is the battery energy constraints of sensor nodes that are dispersed throughout a given region. An appropriate cluster head set can boost message transmission, prolong the lifespan of the sensor network, and sensibly regulate energy usage. This study uses the ROA algorithm to solve the optimal cluster head selection technique, employing the energy consumption of each round as an adaptation function. When compared to the LEACH, PSO, and HFPSO algorithms, ROA can enhance message transmission, extend the lifetime of the WSN, and accelerate the convergence of identifying the ideal set of cluster heads.

In recent years, with the widespread application of machine learning technology in the field of education, this study is dedicated to assessing the impact of multidimensional data on students’ academic performance in Business English courses. By collecting six key types of data, including teaching evaluations and classroom performance, this paper utilizes the Support Vector Machine (SVM) model and takes students’ final grades as the prediction label for in-depth analysis. In addition, we conducted comparative experiments with models such as Random Forest and K-Nearest Neighbors (KNN) to validate the superiority of the SVM model. Specifically, through ablation studies, this research individually examined the contribution of different features to the model's prediction accuracy, finding that classroom performance and learning outcomes have the most significant impact on accurately predicting students’ final grades, while the influence of study habits and learning preferences is relatively minor. The SVM model performed best across all testing metrics, achieving an accuracy rate of 96.64%, which fully demonstrates its application value in the evaluation of Business English courses. These findings are significant for optimizing teaching methods and advancing personalized teaching, offering valuable references for future related research.

During the production of solar wafers on automated production lines, wafers frequently break. To meet silicon wafer manufacturers’ demands for high precision, reliability, and real-time capabilities in online defect detection, this chapter proposes a deep neural network for silicon wafer defect detection based on prior knowledge. Initially, following the prior knowledge that silicon wafer damage often occurs at the edges, an additional prior mask branch is added to the YOLOv5 model, making the network more focused on the edges of the wafers. Secondly, we introduce a parameter-free attention module, SimAm, into YOLOv5, which enhances the model’s ability to extract features of defects on silicon wafers without increasing the model’s complexity. Compared to YOLOv5s, the experimental results show that the performance of our network improved by 2.1%, meeting the need for online defect detection of silicon wafers.

The gannet optimization algorithm (GOA) is a newly proposed swarm intelligence optimization algorithm, and the algorithm is inspired by the predatory habit of the natural creature gannet, which has a strong exploration-exploitation performance. In this paper, we optimize the GOA by using the core concept of an evolutionary matrix in the QUasi-affine TRansformation Evolutionary (QUATRE) algorithm to form the Gannet Optimization Algorithm with QUasi-affine TRansformation Evolutionary (QTGOA). The performance of the QTGOA is compared with five classical optimization algorithms using the CEC2013 test set, and the performance of the QTGOA is proved to be competitive.

The act of combining image data with additional information from multiple remote sensors is known as remote sensing image data fusion. It focuses on processing multi-source data that are redundant or complementary in space or time according to specific rules to obtain better results than any other. More accurate and richer information from a single data, generating a composite image with new spatial, spectral, and temporal characteristics. This paper proposes an image fusion method using the Rafflesia Optimization Algorithm to adjust parameters adaptively. Compared with the traditional method, it has achieved good results.

On the network structure optimization problem of hyperspectral data depth mapping network, combined with the insufficient application of existing network structure to the learning and expression of hyperspectral line features, a depth kernel mapping network optimization method based on structure adaptation is proposed from the perspective of adaptive optimization of network structure. A hyperspectral data classification method based on multiple optimized depth kernel mapping network is presented. Firstly, the kernel structural and learning parameters of the depth kernel mapping network are optimized. This method can adaptively adjust the network structure according to the distribution characteristics of hyperspectral data and improve the performance of hyperspectral image classification. Secondly, the accounting sub and network node optimization methods proposed above are applied to the network to improve deep kernel learning mapping network in hyperspectral image classification. The experimental results show that improving the depth kernel mapping network from the perspectives of kernel operator, mapping network node, and network structure can effectively improve the feature extraction and classification performance of hyperspectral data.

In response to the shortcomings of the current Harris hawks algorithm in terms of convergence, local development ability, and global search ability, this paper proposes an improved Harris hawks algorithm (PREHHO) to address the aforementioned issues. In the population initialization stage, the algorithm uses a piecewise mapping function to make the population distribution more uniform; in the exploration stage, use elite selection strategy to increase the convergence speed of the algorithm; during the exploration phase, PREHHO uses a restart strategy to jump out of local optima. In order to verify the effectiveness of the new algorithm in solving problems, this paper tested it using the CEC2017 function set and compared it with five swarm intelligence optimization algorithms in the 10D, 30D, and 50D dimensions. The experimental results showed that the PREHHO algorithm proposed in this paper has strong optimization ability compared to other intelligent optimization algorithms.

With the burgeoning development of IoT technology, the frequency of network attacks targeting IoT devices is escalating, amplifying the prominence of security concerns. This study proposes a traffic identification methodology based on Long Short-Term Memory (LSTM), adept at discerning diverse attack patterns such as flood attacks, intrusion attacks, and deception attacks. Leveraging the CIC IoT dataset 2023 from the University of New Brunswick (UNB), this dataset comprehensively encompasses attack categories, effectively mirroring the spectrum of attacks faced by IoT devices in real-world scenarios. The model proposed in this study yielded exceptional results, achieving an accuracy rate of 96%. Experimental findings substantiate the efficacy of the proposed approach in detecting and identifying various attacks on IoT devices. This study presents a viable strategy for recognizing diverse attack patterns encountered by IoT devices, thereby fortifying the security of IoT ecosystems.

Engineering optimization is the process of selecting the optimal parameters to achieve performance improvement under given constraints. Optimization has a wide range of applications in different engineering fields, and accurate design boundaries and constraints to find the ideal design parameters are essential for the design of engineering projects, which can enhance the overall performance of systems and products. Whether it is component design for machinery, energy efficiency optimization for power systems, or performance improvement for software development, engineering optimization has a wide range of applications that focus on increasing efficiency and reducing costs. Engineering optimization problems are handled in this paper through a new meta-heuristic algorithm, the food digestion algorithm (FDA). Parameters are optimized step by step through the three-step digestion process of the food digestion algorithm, the mouth, stomach and the small intestine. The particles at the previous digestion point affect the particles at the next digestion point and influence the update of the next particle point. We selected three optimization algorithms to test and compare with the FDA algorithm chosen in this paper, and the results show that the optimization result of FDA is optimal in two engineering optimization problems.

This study utilizes the GPT-4 language model to enhance log summary generation and content-based question-answering sessions, offering administrators and non-experts a more accessible and efficient way to retrieve information. LogAI is employed as a log parsing tool, which, after preprocessing and anomaly detection, generates summarized texts that feed into the GPT model. By leveraging carefully crafted prompts, the GPT-4 model can deliver precise summaries, visual explanations of logs, and responsive answers to user queries based on the log content.

With the rapid development of 5G technology, there is a growing expectation for its applications in the communication field. One of the key performance indicators for 5G networks is handover quality, making its prediction a significant area of current research. This study aims to explore the use of deep learning models to accurately predict handover quality in 5G networks. By analyzing 5G network data, including signal strength, physical cell identity (PCI), and reference signal received power (RSRP), the study conducted power signal classification on relevant datasets. Subsequently, a GRU neural network model and a CNN model were employed to learn the complex relationships among these data, enabling accurate prediction of handover quality. The experimental results demonstrated that this approach could control the average error within 4% to improve 5G network performance and provide significant practical value in the establishment of intelligent 5G networks. Future work will focus on further exploring the practical applications of this model, with the goal of enhancing the service quality and user experience in 5G networks.

This paper proposes an auxiliary framework based on an archive-based covariance matrix and the Gaussian distribution model with shifted mean to improve the gannet optimization algorithm (GOA). A covariance matrix based on an archive is used to build the eigen coordinate system, and samples are taken in it in combination with the shifted mean. To explore the effectiveness of the proposed framework, we compare the results of the framework-assisted GOA on IEEE CEC2017 with other swarm intelligence algorithms, and experimental results demonstrate that framework-assisted GOA is efficient and competitive.

With the emergence of machine vision in defect detection system, it greatly enhanced the accuracy, reliability, and quality of production in industries by saving time and cost. In this study, the image processing and semantic segmentation analysis played vital roles in extraction of defects information on optical filter surface with the help of Halcon software. The study focused on identifying, localizing, and characterizing different types of defects on several shapes of optical filter surfaces with potential applications in optical industry. The experimental results showing overall pixel accuracy of 98.8% indicated that proposed deep learning model is highly efficient and accurate for real-time batch inspection than manual inspection in the optical industries.

In the era of big data, vehicle detection has been a highly complex and widely discussed topic. With the acceleration of urbanization and the increase in traffic volume, the management and monitoring of road traffic have become increasingly urgent, highlighting the growing importance of vehicle target detection. However, existing algorithms suffer from issues such as low accuracy, high false negative rates, and poor robustness. Therefore, this study introduces an improved vehicle detection algorithm, GPF-YOLO, based on the YOLOv8 model. This algorithm integrates the GD mechanism, polarized self-attention (PSA), and the optimized loss function Focaler-IoU, aiming to enhance target detection accuracy. The GD mechanism improves multi-scale feature fusion capabilities, enhancing the detection of targets at different scales. Polarized self-attention introduces a Polarized Self-Attention mechanism that effectively handles fine-grained information about target locations, especially in pixel-level regression tasks. Focaler-IoU optimizes the bounding box regression loss function, focusing on the impact of challenging and straightforward sample distributions on regression results, further improving target detection performance. Experimental results demonstrate that GPF-YOLO outperforms the YOLOv8 model with a 2.05% improvement in mAP@50. Our approach surpasses other target detection algorithms, significantly enhancing the overall system’s robustness and accuracy. It provides a higher level of precision in vehicle detection, offering critical technical support and references for real-time target detection tasks.

In the current era, digital art has been applied in various industries, especially in interactive art exhibitions that combine projection and installation in public spaces. These exhibitions offer memorable experiences with their fantastical and multidimensional interactions. The rapid development of new media technology has accelerated the creation and dissemination of technical art, transforming the mediums of communication, content creation, and expression. Against this backdrop, interactive installation art has emerged as a new form of artistic creation. TouchDesigner is a software suitable for sensor integration and artistic design. This study demonstrates the generation of manageable digital signals through audience interaction with light-dependent sensors and wind speed sensors connected to Arduino UNO devices. The research examines the application of digital design techniques, such as projection and sensor installation design, in conjunction with cultural and historical contexts and established prop fans, using the creation of the interactive game Whispers of the Fan as an example. This text investigates the use of innovative sensor devices and experimental artistic creation by integrating TouchDesigner into art. It aims to digitize material cultural heritage into digital art and combine installation art with spatial exhibitions, providing new avenues for future installation art expression.

In the rapidly evolving landscape of artificial intelligence, the demand for smart and personalized human-machine interaction has grown. Facial expression recognition has emerged as a crucial research direction, becoming a hot topic in recent years. Traditional facial expression recognition algorithms require manual feature extraction, which can be complex and sometimes inefficient. In contrast, approaches based on deep learning leverage neural networks to construct end-to-end models that automatically extract features. The static facial emotion recognition technique using deep convolutional neural networks (CNNs) is currently a mainstream research direction due to its high classification accuracy and strong generalization capabilities. The facial expression recognition algorithm proposed in this paper utilizes a DenseNet (Dense CNN) model trained to extract features and classify facial expressions. The development is carried out using PyCharm software, incorporating PyQt5 and OpenCV modules to design a graphical user interface for the facial expression recognition system. This system not only meets the requirements for facial expression recognition in image and video files but also facilitates real-time facial expression recognition through a camera.

Compared to coarse-grained image classification, fine-grained image classification tasks exhibit smaller differences between object categories. For species such as birds, which possess a rich diversity of appearance and morphological features, fine-grained bird image recognition has become an important task in the field of computer vision. Additionally, this task is of significant importance for the conservation and research of endangered bird species. In this study, we propose a novel module for fine-grained image classification issues—the Feature Enhancement Fusion and Regularizer Module (FEFR). This module includes Enhancement Fusion Module (EFM) and Feature Regularizer (FR). EFM integrates multi-scale information within the same stage, expanding the model’s capacity to capture context and enhancing the interaction between features at different scales to enrich and complexify feature expressions. FR optimizes the feature space by bringing closer the distances between features of the same category while increasing the separation between different categories, thereby enhancing the model’s robustness to intra-class differences and its ability to recognize inter-class variations. Utilizing MetaFormer as the backbone, our experiments on the CUB-200-2011, NABirds, and Ningxia-Birds datasets demonstrate the efficacy of the FEFR module in fine-grained image classification tasks.

With the vigorous development of fields such as big data, artificial intelligence, the Internet of Things, and computer vision, the demand for basic engineering capabilities of university students has undergone a qualitative change. Therefore, the market has put forward higher requirements for the design of relevant courses, especially basic courses. It is hoped that basic courses can closely revolve around emerging disciplines, particularly vibrant fields such as artificial intelligence. In this context, this study aims to explore the deep integration of university-level advanced mathematics courses and computational mathematics courses through the framework of applied research projects, and to investigate their effects on cultivating students’ computational thinking abilities. By combining practical application scenarios, the integration of advanced mathematics and computational mathematics methods is explored to enhance students’ comprehensive abilities and innovative thinking.

The unavoidable prerequisites for the superior development of China’s economy are the promotion of digital industrialization and digitization as well as the deep integration of digital technology and the actual economy. The coupling coordination degree model is used in this paper to analyze the coupling coordination between the advanced manufacturing sector in Dongguan and the digital economy. The results show that, from 2011 to 2020, the composite index of the advanced manufacturing sector and digital economy in Dongguan shows a gradual upward trend, with both sectors’ coupling coordination degrees moving from mild dysfunctions to high-quality coordination. According to the comprehensive index, Dongguan’s advanced manufacturing sector should be centered on the advancement of greening, enhance profitability and innovation capacity, and create a better environment for the real-world implementation of the digital economy. Dongguan’s digital economy also has room for improvement in the communication infrastructure, in addition to maintaining its advantages in the field of information technology.

This study takes higher education in Dongguan area as the research object to explore whether organizational commitment and organizational culture affect employees’ job satisfaction and conducts a questionnaire survey with a convenience sampling method for faculty and staff of higher education in Dongguan area. The methods used in this study are narrative statistical analysis, reliability analysis, t-test, one-way variance analysis, and regression analysis. The results of the study show that in the analysis of organizational commitment differences, younger or less senior staff are weaker than senior staff in terms of linking with the organization; staff with a monthly income of more than 60,000 yuan come to identify more strongly with the organizational commitment. In the analysis of organizational culture differences, the time experienced by older and more senior staff in the organization will make them different from younger and less senior staff; the monthly income can reflect the organizational culture and the time experienced by staff in the organization and the degree of identification with the organizational culture. In the difference analysis of job satisfaction, older teaching staff have longer seniority and should receive higher compensation than younger teaching staff, so their job satisfaction is relatively higher; when the income is higher, it means that their rank and seniority are better than other teaching staff, and the higher compensation they receive, the higher their job satisfaction is. The regression analyses show significant positive correlations and are the same as the hypotheses established in the relevant literature.