Journal of Electronic Research and Application

Research and Implementation of High-Precision Multi-Axis Collaborative Control Algorithm for Stage Machinery Based on Fuzzy PID Optimization

Xuefeng Guan — 2026-06-29

Precise control of stage machinery directly affects the presentation effect and safety of performances. At present, many problems still exist in the multi-axis control of stage machinery, and the independent and controllable development of control technology for performing arts equipment has become an urgent demand for industrial development. Based on domestic chips as the hardware foundation, this study designs a fuzzy PID optimization algorithm adapted to stage machinery, builds a corresponding high-precision multi-axis collaborative control system, optimizes parameter tuning logic, and designs a distributed control architecture, high-precision trajectory planning, and error compensation strategies. Simulation verification and practical tests show that the algorithm can achieve motion control accuracy of ±0.5 mm, command response time ≤5 ms, and stable axis linkage deviation within ±0.5 mm in stage machinery control. The mean time between failures of the system exceeds 50,000 hours. Compared with imported controllers, this scheme reduces the cost of core controllers by 40% and shortens the delivery cycle by 60%, providing a feasible technical solution for the localization and intelligent control of stage machinery.

Research on the Technical Path for Efficient Integration of Power Electronic Control Systems in New Energy Vehicles

Chuanbo Wang — 2026-06-29

Fast growth in the NEV sector has pushed EV to a more effective level of integration, because it has a direct effect on the energy use, performance, and operating stability of the car. In this article, we will study the design of a novel EV electric control system, which will address the present problems of structure redundant, power dissipation, and insufficient adaptive capability of the system. It discusses the technological principles of effective integration by means of hardware, circuitry, and software compatibility, as well as an approach to implement mass production. The results can be used to improve the EV’s overall performance and to promote the development of EV.

A Preprocessing Algorithm based on Wavelength Adaptive White Balance and Enhanced Dark Channel Prior to Processing Underwater Images

Chai Wang — 2026-06-29

In order to overcome the problems of the bluish-green tone of color, bad contrast, and bad texture of underwater pictures, we introduced a two-step lightweight enhancement algorithm called WAWB-IDCP. The algorithm uses the wavelength-based white balance and an enhanced dark channel post-module, which contribute to the correct color correction and optimization of image quality, respectively. It solves the problem of color distortion and artifacts in blocks seen in traditional DCP algorithms. Experiments with multi-dimensional references were performed on three classic algorithms (Gray-world, CLAHE, DCP). The experimental results on the UIEB dataset prove that our algorithm has the highest subjective visual performance and also performs well on other quantitative measures, like the standard deviation of the algorithm of 44.71 and color cast control of 11.07. Furthermore, the SIFT feature experiment proves that it has a great capacity for recovering details and is also noise-resilient. The algorithm is highly performing and can be used in preprocessing underwater images.

Construction of an E-commerce Sales Prediction Model Based on Deep Learning

Changjun Lv — 2026-06-29

Due to the characteristics of e-commerce sales data—high dimensionality, diverse types, and nonlinear relationships—traditional models exhibit insufficient adaptability. This study aims to construct a deep ensemble prediction model to provide scientific support for enterprise inventory and marketing decisions. The model integrates convolutional neural networks (CNN) with an improved weighted deep forest (WDF). A multi-granularity scanning mechanism enhances perception of local spatiotemporal features; a binary adaptive differential evolution algorithm dynamically selects key features; and Bayesian optimization precisely tunes model hyperparameters. Experiments are conducted using large-scale data from mainstream e-commerce platforms. Results show that the model achieves a mean absolute percentage error (MAPE) of only 6.8% on the test set, outperforming baseline models such as ARIMA, random forest, and XGBoost in capturing nonlinear trends and long-term dependencies. This adaptive ensemble model effectively addresses feature redundancy and model mismatch issues, significantly improving the stability of sales prediction and providing reliable technical support for precise e-commerce operations.

Practical Design of Coin Measurement Equipment Based on Virtual Simulation Technology

Siyu Liu — 2026-06-29

Coin circulation plays a fundamental role in daily consumption and commercial settlements across various civilian scenarios. Traditional coin weighing equipment development has heavily relied on physical prototypes for repeated calibration, suffering from chronic issues such as slow R&D progress, excessive testing costs, and limited adaptability to diverse operational environments, thereby failing to meet practical standards for multiple application scenarios. With practicalization of coin weighing devices as the core research focus, virtual simulation technology has been integrated throughout the development process. This approach establishes a comprehensive lifecycle design and validation framework, clearly defining the design logic, technical implementation methods, and deployment plans. Through virtual modeling combined with simulation testing and continuous iterative improvements, common defects in legacy equipment, including material jamming, measurement accuracy errors, and insufficient anti-counterfeiting recognition precision, have been effectively addressed. The overall structural design, operational functions, and performance of the equipment have undergone holistic enhancement and upgrading.

Passenger Escalator Fall Detection Algorithm Based on SCGD-Yolo11m-Pose

Huidi Zhang — 2026-06-29

Escalator fall detection algorithms in subway stations are a crucial means of preventing passenger accidents. However, in a real scene, issues such as the loss of small objects due to scale changes and interference caused by complex backgrounds can lead to false positives and false negatives. This paper proposes a passenger escalator fall detection algorithm based on SCGD-Yolo11m-pose network. First, RFD module was introduced during the downsampling stage to improve the robustness of feature extraction. Second, in C2PSA, deformable attention was used and named C2DA. This enables the model to enhance its ability to perceive various falling poses in complex backgrounds. Finally, in neck network, Gold-Yolo structure replaces the PANet network to strengthen the recombination ability of multi-scale features and improves the accuracy of the model in complex background. Additionally, only four keypoints defined by COCO, both shoulders and both hips, are retained, which improves computational efficiency. Experimental results on our self-built subway escalator fall dataset show that the improved model is improved by 2.3% on AP50 and 2.5% on AP50:95. This validates the effectiveness and practicality of the proposed algorithm in ensuring the safety monitoring of subway passengers.

Progress on Probabilistic Shaping Techniques for Optical Fiber Communication Systems

Wenwen Bian — 2026-06-29

The continuous growth of global data transmission demands significantly higher channel capacity. According to Shannon’s theorem, an upper bound exists for the capacity of additive white Gaussian noise (AWGN) channels. This limit can be closely approached by optimizing conventional modulation schemes. Probabilistic shaping (PS) represents a critical technique to achieve this goal. By employing PS, the signal-to-noise ratio (SNR) gap between practical modulation formats and the Shannon limit can be reduced by up to 1.53dB. PS methods are generally categorized into direct and indirect schemes. Direct PS features low hardware complexity and high processing speed, making it suitable for long-blocklength and linear systems. In contrast, indirect PS can approach the Shannon limit more closely and is better adapted to medium-to-short blocklength and nonlinear scenarios. Nevertheless, it suffers from high hardware complexity and low computational efficiency. Given that direct PS has been well developed and widely deployed, while indirect PS still exhibits considerable room for improvement, future research will concentrate on the enhancement and optimization of indirect PS for nonlinear channel environments.

Research on the Application of Electromagnetic Compatibility Standards for Electronic and Electrical Products in Testing and Certification

Juan Tao — 2026-06-29

This study focuses on the electromagnetic compatibility related standards of electronic and electrical products, introduces its theoretical basis, main international standards, test plans and certification systems, studies the specific adaptability of products, the differences in standard interpretation, and the compliance of consumer electronic products and industrial equipment. It also conducts research on emerging situations such as 5G, sustainable electromagnetic compatibility, and artificial intelligence-driven tools, as well as future standardization trends, so as to promote industrial development.

Research on Speed Sensorless Hybrid Control Strategy of Permanent Magnet Synchronous Motor

Shengli Wang — 2026-06-29

In order to obtain the real-time information of the rotor of permanent magnet synchronous motor (PMSM) and improve the accuracy of speed and position estimation of the traditional speed sensorless control strategy of PMSM, a speed sensorless control scheme of PMSM in full speed range was proposed. A PMSM control system based on improved high-frequency injection algorithm and optimized sliding-mode observer is designed. In the low-speed region, the motor position error signal is extracted by the second-order generalized integrator. In the high-speed region, an improved reaching law sliding mode observer combined with a normalized phase-locked loop is used to eliminate the phase delay. In the full speed domain, the smooth weighting function based on the power function is used to dynamically adjust the weighting coefficient to realize the smooth switching of the algorithm. A system simulation model is built in MATLAB/Simulink to verify the effectiveness of the theoretical algorithm. The simulation results show that the hybrid control strategy of PMSM designed in this paper can realize the full-speed operation of PMSM under the conditions of no-load and load operation.

Research on Crop Leaf Disease Identification and Severity Assessment Based on Lightweight Multitask Deep Networks

Jiali Du — 2026-06-29

In response to model 1, we first cleaned and standardized 61 crop disease image categories by removing duplicates through comparing image filenames with label files using provided path information. Valid samples were resized and augmented to construct a multi-disease classification model based on the lightweight MobileNetV3-Large, with category IDs mapped to disease names. The model was trained and validated with cross-entropy loss, AdamW optimizer, and cosine annealing learning rate, with epoch-dependent loss and accuracy curves recorded. For model 2, a few-shot recognition solution was developed based on model 1, retaining 10 training samples per category. Using pre-trained MobileNetV3-Large as the feature backbone (parameters < 20M), only upper convolutional and classification layers were fine-tuned. Enhanced augmentation, label smoothing, and cosine annealing mitigated overfitting and class imbalance, achieving ~73% validation accuracy for 61 categories; Grad-CAM confirmed the model focuses on leaf lesions. Regarding model 3, severity-graded prediction was implemented by mapping 61 diseases to 3 severity levels via appendix JSON annotations and disease description tables. Images were regrouped to build a three-classification dataset, and a severity prediction model with MobileNetV3-Large (transfer learning, augmentation) was trained, outputting overall accuracy, macro-F1, recall, and a confusion matrix; Grad-CAM visualized key lesions for high-confidence correct predictions. For model 4, a lightweight integrated multi-task model was developed for simultaneous disease identification and severity assessment, using MobileNetV3-Large as the shared feature backbone with 61-category disease and 3-category severity classification heads. Joint optimization via multi-task loss enabled feature sharing and fine-grained assessment, with joint accuracy, confusion matrices, and Grad-CAM analyzing synergy and lesion focus, supporting interpretable diagnosis reports.

Two-Dimensional Material Multifunctional Integrated Devices for Intelligent Sensing Systems: From Basic Sensing to Neuromorphic Computing

Xin Chang — 2026-06-29

Intelligent sensing systems are the core of the Internet of Things and human-computer interaction, and there is an urgent need for multi-functional integration, low power consumption, and miniaturization of devices. Two-dimensional materials provide an ideal platform for the integration of sensing, energy storage, and computing. This article reviews the latest progress in multifunctional integrated devices based on two-dimensional materials in sensing, micro energy storage and neuromorphic computing, analyzes the integrated applications driven by the intrinsic properties of materials, explores device co-design strategies, refines the “structure-material-algorithm” innovation paradigm, and looks forward to challenges and directions.

Encoding Necessity for Standing Government Access to Platform-Held Data: A Three-Axis Model

Linglan Xia — 2026-06-29

Government access to platform-held data is increasingly implemented not through isolated requests but through durable interfaces: dashboards, periodic reporting pipelines and application programming interfaces. This shift changes the object of legality. A single request can be assessed by asking whether a stated purpose justifies the particular disclosure. A standing interface, by contrast, creates an ongoing access capability whose intrusiveness accumulates through repetition, aggregation and time. This article develops a three-axis model for encoding the requirement of minimum necessity at the configuration layer of such interfaces. The model treats data fields, extraction frequency and temporal persistence as the minimal auditable surface of standing access, while requiring purpose, recipients, selectors and onward sharing to be recorded in the same authorisation schedule. Using Chinese administrative interface governance as a stress test, especially health-code pipelines and ride-hailing supervisory feeds, the article shows how lawful or plausible access channels can drift through field accretion, cadence escalation and retention extension. EU and US materials are used as design exemplars rather than as complete solutions, illustrating the importance of reasoned requests, strict-necessity limits, duration sensitivity and auditable safeguards. The article concludes by proposing a procedural toolkit: reasoned authorisations, versioned parameter sheets, access logs, renewal triggers, independent audit and remedies for drift. Where feasible, a policy-as-code counterpart can make authorised limits executable at the gateway.

Study on the Handling Stability of Intelligent Driving Vehicles under High-Speed Conditions Based on Nonlinear Models

Ziang Wu — 2026-06-29

Nonlinear Model Predictive Control (NMPC) has attracted growing attention in vehicle motion control. This paper proposes an NMPC-based trajectory tracking controller for autonomous vehicles that incorporates body roll dynamics to mitigate the accuracy loss and instability caused by increased nonlinearities during high-speed cornering. Using Newton’s second law and vehicle kinematics, and considering body roll geometry together with load transfer effects, a nonlinear vehicle dynamics model is established, comprising vehicle body dynamics and an improved “Magic Formula” tyre model. On this basis, an NMPC predictive model with appropriate linear and nonlinear constraints is developed to keep the vehicle’s operating states within a feasible region. Co-simulations on a CarSim-Simulink platform under high-speed cornering and double lane-change scenarios show that the proposed controller effectively improves trajectory tracking accuracy and vehicle state stability.

PSR-DETR: An Improved RT-DETR for Rail Surface Defect Detection

Guangzhuo Li — 2026-06-29

This paper addresses low detection accuracy in rail surface defect detection. The problem comes from many defect types, large scale changes, and small dense targets. Hence, an improved model based on RT-DETR is proposed namely PSR-DETR. The PR_BasicBlock module first simplifies the model structure. It reduces parameters and computation cost. Meanwhile, it maintains satisfactory detection performance. Consequently, the network becomes more lightweight. After that, the RetC3 module adds a new attention mechanism. It enhances feature integration. It also strengthens the model’s capability to represent and distinguish targets of different scales. Finally, the SSFF module adds extra feature fusion paths. It helps the model emphasize critical regions. As a result, the detection performance is further improved. Experimental results show clear improvements, where the model does not greatly increase parameters or computation. The mAP@0.5 achieves 68.0%. The mAP@0.5:0.95 attains 44.7%, which are improvements of 6.3% and 2.7% over the original model. These findings show that the proposed method is effective and practical for enhancing detection performance.

Nondestructive Detection of Moldy In-shell Walnuts Using Visible Hyperspectral Imaging Combined with 1D-CNN

Ming Liu — 2026-06-29

To achieve rapid and nondestructive detection of moldy in-shell walnuts, this study proposes a walnut mold detection method based on visible hyperspectral imaging combined with a one-dimensional convolutional neural network (1D-CNN). Complete in-shell walnuts were used as the research objects, and hyperspectral data of normal and moldy walnuts were collected in the wavelength range of 400–900 nm. The raw hyperspectral images were processed through reflectance correction, region of interest extraction, average spectral curve construction, Savitzky-Golay (SG) smoothing, and standard normal variate (SNV) transformation. These preprocessing steps were used to reduce the influence of system noise, background interference, and scattering effects caused by differences in walnut shell morphology. On this basis, the processed spectral data were used as the input of a 1D-CNN model to classify normal and moldy walnuts. The experimental results showed that the proposed model achieved an Accuracy of 0.867, Precision of 0.923, Recall of 0.800, and F1-score of 0.857 on the test set. The results indicate that visible hyperspectral imaging can capture spectral variations caused by mold development in in-shell walnuts, and the combination of visible hyperspectral imaging and 1D-CNN can effectively identify moldy walnuts. This study provides a feasible method for nondestructive quality detection of walnuts.

A Study on the Influence of Stress State at the End of Dynamic Gate Biasing and Voltage Pre-Treatment on Threshold Voltage Shift in Silicon Carbide

Ruguan Li — 2026-06-29

This study focuses on the stress state of silicon carbide (SiC) power devices after the end of dynamic gate bias stress (DGS), and the influence mechanism of different pretreatments on the device threshold voltage (Vth) shift. Through a combination of experimental design and theoretical analysis, the impact of the instantaneous electric field distribution and charge trapping/release behavior at the end of dynamic gate bias on subsequent Vth stability was systematically explored. Research results show that the stress state (positive voltage/negative voltage) at the end of dynamic gate bias and whether pretreatment is performed will significantly affect the Vth shift of SiC devices under dynamic operating conditions. This research provides important theoretical basis and experimental guidance for the stability design and reliability improvement of SiC power devices.

Dynamic Finite Element Simulation Analysis of Table Tennis Hitting

Jiahao Yang — 2026-06-29

The process of hitting a table tennis ball involves complex dynamic behaviors, in which the ball’s velocity, spin, and deformation characteristics directly affect the shot outcome. Based on the finite element simulation method, this study establishes a simplified model of the interaction between the table tennis ball and the racket. Two typical hitting patterns are simulated in the research: one dominated by impact namely the fast attack technique, while the other dominated by friction, referring to the loop drive stroke. The results indicates that in the impact mode with a normal speed of 20 m/s, the table tennis ball attains a maximum speed of 45 m/s, along with a spin of about 450 rpm. By comparison, in the friction mode with a tangential speed of 20 m/s, the ball spins at about 900 rpm but has a lower outgoing ball velocity at about 9 m/s. The simulation clearly reveals the dynamic patterns of ball deformation, energy transfer, and the evolution of the ball’s motion state. It also verifies the regulatory effect of hitting angle and speed on the ball release. This research provides quantitative evidence for the scientific training and equipment optimization in table tennis, while also demonstrating the practical value of finite element simulation in the field of sports engineering.

Optimization of Welding Sequence for Bridge Crane Girder Structure Based on Numerical Simulation

Mingyuan Li — 2026-06-29

Bridge cranes are key equipment in modern industry and logistics. Their manufacturing quality directly affects the safety and service life of the entire machine. The main girders and end beams of such cranes are typically fabricated by welding thick plates. Residual stress and deformation generated during welding are major factors that influence the structural service life and load-bearing capacity. The paper adopts the inherent strain method as the core theory. A mid-surface model of the structure is created using SolidWorks. Two-dimensional mesh discretization and geometry cleanup are performed in Hypermesh. Finally, numerical simulations of different welding sequence schemes are carried out using Sysweld software. During the research, the double ellipsoid heat source model is first calibrated and validated for T-welded joints to ensure the accuracy and applicability of the heat source parameters. Subsequently, simulations are performed according to different welding sequence schemes. Residual stress distribution contours and deformation contours under each scheme are extracted, and a quantitative comparative analysis is conducted on the maximum residual stress, deformations in each direction, and total deformation, leading to the selection of an optimal welding sequence scheme. The simulation results show that different welding sequences have a significant effect on the final residual stress and deformation distribution. Improving the welding sequence can effectively control welding-induced residual stress and deformation, providing guidance for optimizing the welding process of bridge cranes.

The Development and Current Situation of Flexible DC Distribution Technology

Shengquan He — 2026-06-29

Against the strategic background of the “dual carbon” goals, flexible DC distribution technology has become a core pillar of the new power system with its high-efficiency, flexible and controllable technical advantages. It can not only coordinate the relationship between distributed generation and the power grid, but also adapt to high-proportion new energy access and promote the low-carbon transformation of the power grid. This paper analyzes the advantages and core technologies of flexible DC distribution technology, expounds its development status, and prospects its development path from three aspects: R&D of power electronic equipment, optimization of flexible DC distribution protection schemes, and coordinated development, aiming to provide a reference for the stable development of the new power system.

Development and Application of an AI Popular Science Digital Human System Based on Local Private Large Model Technology

Yongqiang Wang — 2026-06-29

With the continuous development of artificial intelligence technology, digital human technology offers extensive applications in science popularization education. This paper designs an AI popular science digital human system using local private large model technology, which integrates key technologies including speech recognition, natural language processing, speech synthesis, and digital human driving to enable intelligent interactive Q&A with users. The system adopts a locally deployed architecture, fine-tuned based on the Qwen large language model, and combines SenseVoice speech recognition, CosyVoice speech synthesis, and the LiveTalking digital human driving engine to build a complete popular science interaction process. The system has been put into practical use in scenarios such as science and technology festivals in primary and secondary schools and science and technology exhibition halls, which effectively improves the fun and interactivity of science popularization education and provides a new solution for cultivating scientific literacy among teenagers.

Research on Surface Functionalization Modification and Photoelectric Properties of Single-Walled Carbon Nanotubes

Jingchun Zhang — 2026-06-29

Single-walled carbon nanotubes (SWCNTs) are regarded as the primary candidate materials for optoelectronic devices in the post-Moore era due to their unique two-dimensional quantum confinement effect, high carrier mobility, and tunable bandgap structure. However, the intrinsic SWCNTs feature strong chemical inertness, easy agglomeration, and fixed band structure, which greatly restrict their performance in complex optoelectronic systems. Therefore, this paper focuses on bandgap renormalization induced by sp³ hybridization, Fermi level shift caused by charge-transfer doping, exciton binding energy modification via dielectric environment screening, and the influence of chiral-selective modification on circular dichroism. Based on the established Hamiltonian perturbation model and many-body Green’s function theory framework, the physical picture of macroscopic reconstruction of the photoelectric response of SWCNTs by functional groups as artificial defects or dielectric coating layers is revealed from a microscopic perspective, providing a theoretical basis for the design of high-performance and multifunctional carbon nanotube optoelectronic devices.

Analysis of Metro Car Ride Comfort Based on FBG Measurement and Validation

Zhaoqing Guan — 2026-06-29

To accurately evaluate the ride comfort of metro cars, this paper collects and analyzes carbody vibration acceleration on an urban metro line using Fiber Bragg Grating (FBG) measurement technology. Through trend term removal and bandpass filtering preprocessing, effective vibration signals are extracted, and characteristic parameters such as the Sperling ride comfort index and dominant frequencies are calculated. Simultaneously, a SIMPACK vehicle dynamics model is established with American Level 5 track spectrum excitation. Simulation results are compared with measured data to validate model reliability, and the model is then used to provide simulation comparisons for ride comfort analysis. Results show that the vertical ride comfort indices at two measurement points are 1.939 and 1.956, and lateral indices are 2.015 and 2.045, all reaching the “Excellent” level. Relative errors of ride comfort indices between simulation and measurement range from 5.2% to 6.7%, indicating high model credibility. Frequency domain analysis shows that the vertical dominant frequency of the carbody is 1.50 Hz and the lateral dominant frequency is 1.75 Hz, with good consistency between measurement and simulation. This work provides an effective method for metro car ride comfort evaluation based on FBG measurements, and offers quantitative basis for parameter calibration of metro train dynamics models.

Answer Distribution Bias in OmniBench: How Answer-Position Skew Affects Multimodal Large Language Model Evaluation

Sai Wan — 2026-06-29

OmniBench is a widely used tri-modal (image–audio–text) benchmark containing 1,142 four-choice multiple-choice questions. We discover a severe answer-position skew in OmniBench: option D is correct 48.6% of the time (χ² = 384.34, p = 5.46×10⁻⁸³), nearly twice the expected 25%. To test whether this skew distorts evaluation outcomes, we design an option-shuffling experiment: keeping all question content unchanged, we randomly reassign letter labels so that the correct answer is uniformly distributed (D ≈ 25%), then re-evaluate the same models. Results show that accuracy changes significantly in two of three tested models after shuffling (up to 4.20%, p < 0.01), demonstrating that unequal answer distribution can significantly bias model evaluation outcomes. Furthermore, we propose a label-free content-scoring evaluation method based on conditional log-probability, which achieves distribution-invariant evaluation (accuracy difference ≤ 0.18%, p > 0.4).

Application Research on a Large Language Model-Based Auxiliary Design System for Nuclear Power Engineering Modification

Lihong Tang — 2026-06-29

Aiming at the core pain points in nuclear power engineering modification, such as low efficiency of knowledge retrieval, cumbersome document preparation processes, insufficient standardization, and difficulties in expert experience inheritance, this paper proposes and constructs an intelligent auxiliary design system based on Large Language Models (LLMs). The system adopts a hybrid storage strategy combining vector databases and knowledge graphs to build a high-quality vertical knowledge base for engineering modification, successfully transforming unstructured tacit knowledge scattered in historical documents into computable and retrievable structured intelligence. On this basis, the system establishes a dynamic multi-dimensional tag system and an intelligent evolution mechanism, and innovatively creates an intelligent auxiliary design engine based on dynamic questionnaire interaction. Integrating template guidance, questionnaire interaction, and Retrieval-Augmented Generation (RAG) technology, the engine realizes the automatic preparation and compliance verification of design documents. Application results indicate that the system effectively guides the sorting of design inputs, significantly improves design efficiency and quality, and reduces the risk of human error. Finally, the paper outlines the future integration with external drawing platforms, 3D design platforms, and computational analysis platforms, aiming to build an online collaborative design platform covering the full life cycle and achieve an intelligent closed-loop of the entire engineering modification process.

A Bioinspired Neuromorphic Chip Based on Protein-Protein Interaction Network Topology and Its Application in Adaptive Gait Control for Legged Robots

Qiang Chen — 2026-06-29

Traditional footstep controller chips of the legged robot have problems such as low energy utilization rate, lack of flexibility, and poor environmental resistance ability. In order to solve this problem, this study proposes a neuromorphic chip design based on PPI network topology, which mimics the small-world property, scale-free degree distribution, and modular structure observed for the PPI networks using a set of protein network processors (PNPs).In memristor crossbar arrays, synaptic weights can be stored and computed in situ to eliminate the memory-processor bottleneck. A three-level motion hierarchy deals with control on different timescales: a reflex level for millisecondscale posture stabilization, a rhythm level to generate periodic gait patterns, and a strategy level for long-horizon motion planning and terrain adaptation. Fabricated in 28 nm CMOS, the chip contains 1024 PNPUs and draws a total power of 2.5W. Experiments with a quadruped robot show that our chip is able to achieve better than 10x energy efficiency over traditional model predictive control running at 1.5 m/s walkspeed on flat ground, with terrain transitions completed within 0.8s. With a quarter of processing units disabled, we still retain more than 80% of locomotion capability, verifying that our model generalizes to unseen situations.

Experimental Research on Multi-Source Information Fusion-Based UAV Integrated Navigation in Bridge Areas

Hui Li — 2026-06-29

Aiming at the inherent limitations of traditional bridge inspection methods, such as low efficiency, high operational risk, and limited inspection coverage, this study proposes an autonomous navigation and obstacle avoidance scheme for bridge inspection uncrewed aerial vehicles (UAVs) based on multi-source data fusion. Firstly, high-precision time synchronization is implemented. The time-stamped data is divided into three major data categories, namely visual images, LiDAR, and GNSS/IMU data. An improved Kalman filtering algorithm is then adopted to achieve spatiotemporal registration and error compensation of the multi-source data, which significantly enhances the accuracy and stability of environmental perception during UAV flight. Secondly, a hierarchical autonomous navigation strategy is designed by combining the structural characteristics of bridges. The strategy realizes global path planning based on bridge structural features and conducts real-time optimization of local obstacle avoidance paths, ensuring that UAVs maintain safe and efficient operation in complex bridge environments (e.g., narrow spaces, complex structural components, and variable weather conditions). Finally, experimental verification is conducted in a real bridge inspection scenario, and the results demonstrate that the proposed scheme outperforms traditional methods in key performance indicators, including navigation positioning accuracy, obstacle avoidance response speed, and inspection coverage rate. Specifically, the positioning error is reduced by 32%, the obstacle avoidance response time is reduced by 28%, and the inspection coverage rate is increased to over 96%. This research provides important technical support for the engineering application of UAVs in bridge inspection and holds practical significance for promoting the intelligent operation and maintenance of infrastructure. The proposed multi-source data fusion method and hierarchical navigation strategy can also serve as a reference for other UAV-based inspection tasks in complex industrial environments.

Construction and Application of a Knowledge Graph-Based Educational Question Answering System for Probability Theory and Mathematical Statistics

Yihan Huang — 2026-06-29

In recent years, with the rapid development of artificial intelligence and big data technologies, knowledge graphs have gained widespread attention and application. As a fundamental course in mathematics and statistics, Probability Theory and Mathematical Statistics contains complex and highly interconnected knowledge points, making traditional learning methods less effective for understanding its internal logic. Therefore, constructing a knowledge graph and developing a corresponding question-answering system for this subject is of great significance. This project uses the Probability Theory and Mathematical Statistics Tutorial (3rd Edition) as the data source to construct a knowledge graph based on Neo4j. Cypher language and APOC tools were used for data import and graph construction, while Neo4j Bloom was employed for visualization. In addition, a question-answering system was developed using natural language processing techniques and the Flask framework to provide intelligent query services. The system can help students better understand and learn probability theory and mathematical statistics while reducing dependence on traditional textbooks.