In high-concurrency scenarios, network and disk I/O-intensive operations often compete for shared resources, resulting in a decline in the server's load capacity. To address this challenge, this paper proposes a sophisticated high-concurrency server optimization solution. It utilizes various Reactor models in the Linux system, combined with the powerful Epoll mechanism and thread pool, to conduct research and optimization on the server's load capacity.Firstly, the event-driven and other modules required by the Web server are implemented and integrated. Secondly, the number of Reactors, the number of threads, and the business processing time under the Linux system are designed and controlled, and the design and implementation scheme of the high-concurrency server based on the Reactor model with the Epoll mechanism and thread pool are determined. Finally, the performance differences and the best usage scenarios of Web servers with different Reactor models in high-concurrency environments are analyzed through stress tests. The comparison results show that the QPS (Queries Per Second) indicator of the Web server based on the multi-Reactor multi-thread model is three times higher than that of the single-Reactor single-thread Web server, verifying its overall advantages in high-concurrency and long-term business processing. The research results demonstrate the applicable scenarios of different Reactor models, providing theoretical basis, implementation examples, and data support for choosing the appropriate Reactor model in actual server development, helping developers select the most suitable Reactor model according to specific server requirements to ensure higher efficiency in high-concurrency scenarios.

This study investigates dictionary-based word segmentation algorithms, which are essential in Natural Language Processing (NLP). Chinese word segmentation poses significant challenges due to the lack of clear word delimiters in the language. This paper explores the advantages and limitations of dictionary-based segmentation algorithms, focusing on how data structures such as Trie and Double-Array Trie (DAT) can enhance segmentation efficiency. An analysis of Trie and DAT structures leads to an optimization achieving constant-time state transitions. This paper evaluates and compares various segmentation algorithms, including full segmentation, forward maximum matching, backward maximum matching, and bidirectional maximum matching. The inherent limitations of dictionary-based segmentation, particularly its dependence on dictionaries and poor disambiguation capability, are also discussed.

Compared with SC decoding, BP decoding with the parallel mechanism has higher throughput and lower latency, which is more suitable for the demand of 5G scene. To further improve its FER performance and reduce the memory overhead, a recurrent neural network-aided bit-flipping BP decoding of polar codes is proposed. Firstly, it uses bit flip to correct the wrong decoded bits during the decoding iteration. And then, the offset min-sum approximation is used to replace multiplication operation. Lastly the improved recurrent neural network architecture is adopted to realize parameter sharing. The simulation shows that the proposed scheme has a better error correction ability with fewer flipping times, and can effectively reduce the computational resource consumption and extra memory overhead of BP decoding.

To address the motor synchronization errors in the take-up system, an improved deviation-coupled synchronous control strategy based on fuzzy adaptive PID control is proposed. Traditional dual-motor synchronization methods struggle under conditions of high speed, highly dynamic operation, and significant disturbances, particularly in mitigating synchronization errors and system oscillations caused by parameter variations and sensor delays. This paper establishes a pendulum-angle feedback model, introduces a dynamic velocity compensation mechanism and adaptive synchronization gains, and develops a fuzzy adaptive PID controller. The controller dynamically adjusts PID parameters in real-time using pendulum-angle errors and their rate of change as inputs, achieving precise control of the motors' speed difference. Simulink simulation results indicate that, compared to traditional PID control, the proposed method reduces motor speed overshoot by 75% and decreases disturbance recovery time by over 50%. Experimental validations confirm that the method significantly reduces pendulum-angle fluctuations, enhances dynamic response speed, and improves system stability, thus fulfilling practical engineering requirements.

To enhance the safety and efficiency of coal mine gas monitoring, this study develops an intelligent Internet of Things (IoT) monitoring system incorporating a Fuzzy-PID control algorithm. The system is structured into four layers—sensing, network transmission, control service, and mobile application—ensuring real-time data acquisition, stable transmission, intelligent processing, and remote monitoring. The Fuzzy-PID algorithm dynamically adjusts control parameters to improve response time and accuracy under nonlinear and uncertain conditions. Simulation experiments validate the system's performance, comparing traditional PID, Fuzzy, and Fuzzy-PID control strategies. Results indicate that the traditional PID algorithm achieves a response time of 2.0 s but exhibits oscillations of ±0.1 concentration units. The Fuzzy control algorithm stabilizes gas concentration within 4.0 s with deviations below ±0.05 units. The proposed Fuzzy-PID algorithm achieves an optimal balance, stabilizing gas concentration within 2.5 s with deviations reduced to less than ±0.03 units. These improvements enhance mine safety by reducing gas concentration fluctuations and providing real-time risk alerts. Practical deployment in a coal mining enterprise confirms the system’s capability in reducing manual intervention by 30% and improving early warning accuracy by 25%, demonstrating its potential for intelligent mine development.

Remote Sensing Object Tracking refers to the process of detecting, recognizing, and tracking targets on the ground or at sea using remote sensing technology, particularly sensors mounted on satellite or aerial platforms to obtain high-resolution remote sensing image sequences. Current methods for remote sensing object tracking face challenges such as low tracking success rates and inefficiencies. This paper proposes a neural network for remote sensing object tracking based on SiamRPN++, which introduces an improved network structure incorporating the C3Minus module and a coordinate attention mechanism within the backbone extraction network. Furthermore, we design a feature extraction module, ResSwinT, that combines ResNet and Swin Transformer architectures to integrate local and global information obtained from feature maps as foundational features. This approach effectively addresses the aforementioned issues, and quantitative experiments demonstrate an increase in accuracy and success rates by 1.9% and 4.7%, respectively, indicating that our method effectively handles object tracking in remote sensing images.

To address the problems of data sparsity and cold start in collaborative filtering algorithms, this paper proposes an improved course recommendation method that integrates knowledge graphs and collaborative filtering. First, the RippleNet model is used to construct a knowledge graph based on course-attribute-relation triples and generate a recommendation list. Then, an item-based collaborative filtering algorithm utilizes users’ historical interaction behavior to produce another recommendation list. Finally, a weighted linear method is employed to fuse the recommendation list generated by the RippleNet-based course knowledge graph and the one generated by collaborative filtering, resulting in the final course recommendation list. Experiments conducted on the public dataset MOOCCube demonstrate that the RippleNet-CF method improves precision, recall, and F1-score, while also effectively mitigating the issue of data sparsity.

This paper proposes a vehicle and pedestrian detection model based on an improved RT-DETR to address the issues of high redundancy in feature extraction and insufficient accuracy for small targets in existing real-time detection models, especially in complicated traffic scenarios. The core of this improved model is to embed a parameter free SimAM (Simple Attention Module) attention mechanism in the backbone network. The SimAM mechanism dynamically generates three-dimensional attention weights through energy functions, effectively enhancing the expression ability of fine-grained features of pedestrians and vehicles. This improvement not only reduces redundant information in the feature extraction process, but also improves the detection accuracy of the model for small targets, enabling the model to more accurately identify and locate small targets when dealing with complex traffic scenes. The experimental results show that on the BDD100K dataset, the improved model achieved an average precision of 73.6%, which is 3.7 percentage points higher than the original RT-DETR, effectively enhancing the model's capability to detect vehicles and pedestrians in intricate environments.

Road surface disease detection is a vital component of road maintenance. Traditional deep learning-based detection methods face challenges such as low detection accuracy, high false alarm rates in complex scenarios, and significant missed detection rates for small targets like potholes. To address these limitations, this paper proposes an improved pavement disease detection algorithm based on RT-DETR. First, a lightweight backbone network named LMBANet is constructed by integrating DRB and ADown modules. This network enhances feature extraction capabilities without increasing computational overhead during inference, preserving local details of low-level features while expanding the receptive field to capture long-range semantic information and reduce false detection of diverse defects in complex scenes. Second, an small-target enhanced feature pyramid network is designed using SPDConv and OmniKernel. By feeding large-scale feature maps extracted by the backbone into a feature fusion layer and enhancing multi-scale feature representation through EFKM, this network resolves the high missed detection rate of small targets in the original model. Experimental results demonstrate that on the RDD2020 dataset, the improved network achieves an mAP of 69.2%, representing a 2.1 percentage point improvement over the original network, while simultaneously reducing parameters and computational costs.

Deep learning has emerged as a vital approach for identifying and addressing vulnerabilities in software systems. A key challenge in this process lies in effectively representing code and leveraging AI techniques to capture and interpret its semantics and other intrinsic information. This paper employs bidirectional slicing techniques to extract code slices containing control and data dependencies from program dependency graphs, targeting key points of different vulnerabilities. To represent the node features within the slices, code tokens are mapped to integers and transformed into fixed-length vectors, leveraging Word2vec and BERT models to embed the code nodes and extract structural graph features. The embedded feature matrix is then fed into a Gated Graph Neural Network (GGNN), which aggregates information from nodes and their neighbors to enhance long-term memory of graph-structured data. By iterating through several time steps within GRU units, the final node features are generated. Additionally, edge relationships are used to propagate and aggregate information, further improving the accuracy of vulnerability detection. Experimental results demonstrate that the proposed model achieves an F1-score of 93.25% on the BigVul dataset, showcasing strong detection performance.