To address the common challenges in industrial equipment bearing fault diagnosis， including cross-operational condition transfer difficulties and target domain label scarcity， this study proposes an intelligent diagnostic method integrating feature enhancement and domain-adversarial learning. By constructing a Symmetric Dot Pattern feature map based on wavelet transform （WT-SDP）， the original one-dimensional vibration signals are mapped into two-dimensional geometric-semantic feature representations， effectively addressing the limitations of traditional methods in translation invariance and inefficient modeling of long-range dependencies. This approach significantly improves feature separability and noise robustness. Furthermore， a domain-adversarial neural network （DANN） framework is designed， incorporating a gradient reversal layer to achieve multi-scale alignment of feature distributions between source and target domains. This eliminates reliance on target domain labels while mitigating domain shift issues， thereby enhancing model generalization under heterogeneous operating conditions. Experimental validation on cross-domain transfer tasks from the Case Western Reserve University bearing dataset （CWRU） and the Dynamic Diagnostic System testbench （DDS） demonstrates that the proposed method achieves an average diagnostic accuracy of 95% on target domains， representing a 20% improvement over baseline models. This research provides a novel solution for cross-domain fault diagnosis in industrial equipment under complex operating conditions， with the proposed method showcasing significant advantages in domain adaptation efficiency.

Part machining accuracy inspection is a critical aspect of production， and efficient deviation analysis methods can save costs while ensuring part quality. For three-dimensional point clouds obtained through laser 3D scanning， a 3D part deviation analysis system has been designed. By comparing the scanned data with the CAD design model of the part， the deviation of the part is calculated and visually displayed. The system is divided into three parts： data preprocessing， alignment， and deviation calculation. In the data preprocessing stage， the CAD 3D model is discretized into a point cloud， while points from the scanned data are extracted， and key points from both are identified to facilitate alignment. During the alignment stage， an OBB segmentation box coarse registration algorithm is designed for coarse alignment， followed by the ICP algorithm for fine alignment， effectively preventing the ICP algorithm from falling into local optima. In the deviation calculation stage， a point-surface substitution method is used to compute the initial deviation values， and finally， an outlier processing system is designed to handle anomalies. Compared with professional software， the measurement error is at the micrometer level.

Audio-Visual Question Answering （AVQA） is a core task in multi-modal reasoning， requiring models to effectively align and understand both visual and auditory signals. Existing methods often struggle with three major challenges： （1） insufficient modality alignment due to decoupled pretraining and task supervision； （2） difficulty in identifying key segments from long and redundant video streams； and （3） limited adaptability of static fusion strategies. To address these issues， this paper proposes a unified end-to-end framework that introduces a question-guided pooling module with an Optimal Transport loss for fine-grained semantic alignment and a dynamic expert fusion module （FlexFuseMoE） based on a Mixture-of-Experts （MoE） architecture that enables early， middle， and late fusion through routing mechanisms. Extensive experiments on MUSIC-AVQA demonstrate that the proposed method achieves state-of-the-art performance with enhanced interpretability， validating the effectiveness of question-guided alignment and dynamic fusion. This approach provides a generalizable solution for complex multi-modal tasks.

As product markets increasingly trend toward dynamic variability and personalized customization， the underlying execution logic of production systems must possess greater flexibility and adaptability. However， this requirement inevitably leads to a sharp increase in the combinations of physical control logic， thereby complicating debugging and reducing system development efficiency. To address these challenges， this paper proposes a process-driven modeling method for production control logic， aiming to achieve decoupled representation of the production execution logic and automatic generation of the control logic. The method constructs a four-element model of control logic and further introduces encapsulated action unit and action chain models. On this basis， this study establishes a process-driven framework and its implementation workflow for generating production control logic， and develops a low-code modeling and automatic generation prototype system to support the proposed method. Experimental results demonstrate the effectiveness and flexibility of the proposed method in modeling complex control logic.

The refrigerant filling line， as a typical discrete manufacturing system， faces challenges such as complex multi-station signal coordination， long control logic verification cycles， and high on-site commissioning costs. Traditional simulation verification methods are inadequate for achieving a closed-loop verification of both the entire line's control logic and physical behavior. To address these issues， this paper proposes a digital twin-driven virtual commissioning system for flexible refrigerant production. By constructing a three-layer collaborative model comprising "Device-Signal-Control"， high consistency between the virtual model and the physical production line is ensured. Additionally， a real-time closed-loop verification mechanism based on the “PLC-OPC-Twin Model” is designed， forming a dynamic loop between control logic， signal interaction， and device behavior， thus enhancing the observability and diagnostic capability of the debugging process. Experiments show that the digital twin model established based on this virtual commissioning system can expose key control logic defects， signal conflicts， and robot path interference in the design and commissioning process， enabling rapid iteration and early verification of the filling line's control strategy， and providing a reusable technical framework for its flexible deployment.

Additive manufacturing technology has found broad applications in aerospace and high-end equipment manufacturing. However， fragmented process data， insufficient process knowledge accumulation， and limited knowledge transfer still hinder its practical application. To address these challenges， this study designs and implements a digital thread system for metal additive manufacturing based on a low-code platform. The system adopts the digital thread as its core logical framework and integrates three functional modules： a process parameter database， a part parameter database， and a knowledge base. These modules enable end-to-end data integration and closed-loop management across the entire workflow， from design and printing to verification. By leveraging low-code rapid development and a microservice architecture， the system achieves standardized storage of multi-source data， traceable control of the printing process， and dynamic accumulation of knowledge. The results demonstrate that the proposed system significantly enhances data transparency and reusability throughout the metal additive manufacturing process. It provides scalable technical support for rapid iteration， process optimization， and knowledge inheritance of complex components， offering a feasible pathway toward the digital transformation of aerospace maintenance and high-end manufacturing sectors.

Due to the fact that laser displacement sensors can only obtain single point distance information along the direction of the laser beam， it is necessary to accurately determine the unit direction vector d （ l， m， n ） of the laser beam in the tool center point （TCP） coordinate system of the robot through system calibration， in order to achieve the conversion from one-dimensional distance values to three-dimensional spatial coordinates. In response to this demand， an innovative calibration method based on standard sphere targets has been proposed. Firstly， control the robot to drive the laser sensor to uniformly sample the surface of the standard ceramic ball in different poses， and synchronously record the laser sensor data and the robot's end pose information； Based on spherical geometric constraints， establish a overdetermined nonlinear equation system containing unknown parameters such as laser beam direction vector， and solve it using a nonlinear least squares optimization method. To verify the calibration accuracy， the calibrated measurement system was used to conduct measurement experiments on the diameter of standard ceramic balls. From the 50 collected points， 25 points were randomly selected five times for sphere fitting. The experimental results show that the error between the measurement results and the standard ceramic ball diameter is less than 0.03 millimeters， which verifies the effectiveness and practicality of the proposed laser beam unit direction vector calibration method.

With the widespread adoption of ultra-high voltage switchgear， the requirements for assembly accuracy and efficiency continue to increase. Traditional manual assembly suffers from high labor intensity， high centering errors， and insufficient consistency， necessitating the development of intelligent and automated assembly methods. To address the challenges of ultra-high voltage disconnectors， such as the complex spatial structure and the fact that the assembly plane and the centering axis are not in the same coordinate system， this paper proposes a robotic assembly method based on 3D point clouds and vision guidance. First， the moving end intermediate conductor is scanned using a 3D point cloud to obtain its spatial pose in the basin coordinate system. Hand-eye calibration is then used to achieve coordinate system transformation. Subsequently， the pose estimated from the point cloud is error-corrected using vision guidance technology to obtain a high-precision assembly alignment reference. Furthermore， inverse kinematics and path planning are used to generate the assembly trajectory. Force/position hybrid control is then introduced to achieve smooth contact and stable insertion during the assembly process. Experimental results demonstrate that this method achieves an overall assembly accuracy of 97%， meeting the efficiency and precision requirements of the assembly process and enabling automated assembly of high-voltage disconnectors.

To improve the success rate of robotic arms in grasping target objects under different environmental conditions， this study proposes an efficient grasp pose detection model， PaTMGNet， based on a parallel Transformer architecture. The model innovatively incorporates a multi-layer perceptron （MLP） and a Shuffled Attention module to reduce the amount of training data needed and enhance pixel-level relationships and inter-channel dependencies among feature information. A multi-scale feature fusion mechanism is employed to improve grasping detection accuracy. To evaluate the model's performance， it was tested on the Cornell and Jacquard benchmark datasets， achieving accuracies of 98.9% and 94.7%， respectively； on the multi-object dataset， it achieved an accuracy of 94.9%. To further validate its superior performance， the model was deployed and tested both in simulation environments and on a physical robotic arm， reaching accuracies of 97.3% and 94.2%， respectively. Experiments demonstrate that this algorithm has significant advantages in both single-object and multi-object grasping tasks.

Aiming at the problem that collaborative robots lack force sensors when performing contact force tracking under unknown environmental stiffness changes， this study proposes a variable impedance control method that integrates virtual force sensing and model predictive control to enhance the robot's force tracking accuracy and adaptability in uncertain environments. Firstly， based on motor current， joint angle and angular velocity signals， a virtual force sensor is constructed via a generalized momentum observer， and the end-effector contact force of the robot is estimated online. Secondly， a model predictive controller is designed， which takes the estimated force and internal robot states as inputs， dynamically adjusts impedance parameters and calculates trajectory corrections in real-time， thereby achieving adaptive impedance regulation during the force tracking process. Experimental results show that the generalized momentum-based disturbance observer can effectively estimate the end-effector contact force， where the force estimation error in key degrees of freedom is less than 1 N， and rapid convergence within 4 seconds is achieved， providing a reliable feedback basis for sensorless force tracking. Effective adaptive force tracking control for different environmental stiffnesses is realized by the proposed method without external force sensors， providing a feasible control solution for compliant operations of robots in uncertain environments.

To address the issue in Distributed Flexible Job Shop Scheduling where inter-factory transport time is often simplified as a fixed value， failing to reflect the impact of individual workpiece differences， this paper constructs a dynamic transport time model based on workpiece characteristics. This model comprehensively considers multiple dimensions such as workpiece complexity， machining time characteristics， process route variability， and operation stages， thereby providing a more accurate characterization of the logistics process in actual production. To solve this model， a hybrid framework combining Genetic Algorithm and Tabu Search is adopted for optimization. Experiments conducted on extended standard MK and La series benchmark instances show that the proposed algorithm yields significant improvements in both makespan and total transport time compared with GA， TS， GA-OP， and GA-JS algorithms， validating the effectiveness and superiority of both the model and the algorithm. Thus， the dynamic transport time model proves effectiveness in enhancing scheduling accuracy， offering a new approach for refined scheduling in distributed manufacturing.

Facing the critical strategic demand for enhancing the resilience and security of industrial and supply chains at the national level， the paper focuses on digital twin technology as a key enabler for driving the digital and intelligent transformation of warehousing and logistics systems. A three-stage evolutionary trajectory is systematically outlined， progressing from static modeling to dynamic synchronization and ultimately to intelligent decision-making. In view of the structural challenges in traditional warehousing and logistics systems such as data silos， lagging equipment maintenance， rigid processes， insufficient flexibility， and lack of holistic optimization， this study constructs a systematic empowerment pathway encompassing five dimensions： “omni-domain data integration—intelligent operations and maintenance reconstruction—process simulation optimization—flexible collaborative scheduling—global decision simulation.” The construction of a unified data foundation enables standardized access and high-quality governance of multi-source heterogeneous data； Deployment of a predictive maintenance platform significantly enhances equipment reliability and system continuity； Application of virtual simulation and dynamic optimization technologies achieves intelligent restructuring of warehouse operations and efficiency multiplication； The construction of an elastic resource scheduling mechanism enhances the adaptability of the system to dynamic demands； By building a full-chain simulation decision-making sandbox， the system is boosted from empirically driven local decisions to data- and model-driven global autonomous optimization. In the practical application within the chemical fiber industry， this technology system has increased production efficiency by 5%， reduced operational costs by 15%， improved equipment utilization by 10%， and shortened fault-handling time by 30%. Looking ahead， digital twin technology will evolve toward “holistic coordination and intelligent symbiosis，” providing critical support for constructing autonomous as well as controllable modern warehousing and logistics systems and cultivating new productive forces.

In goods-to-person picking systems for production logistics， multiple SKUs are often mixed together on the same inventory carrier， and different picking orders may be fulfilled by accessing the same carrier. Conventional order batching methods based on SKU similarity focus only on SKU overlap among orders， making it difficult to group orders that involve different SKUs but can be picked using the same inventory carrier into the same batch. As a result， unnecessary carrier movements are often incurred. To address this issue， an order batching method incorporating a carrier association network is proposed. Building upon the traditional SKU-based association network， a carrier-level association degree calculation method is proposed to characterize the relationships among orders at the carrier level. The resulting association degree is then integrated into the order batching decision process， enabling a more accurate representation of order correlations induced by shared inventory carriers. Experimental results demonstrate that the proposed method exhibits good convergence performance. Compared with the optimal solutions obtained by a general-purpose solver， the average optimality gap is 11.9%. Moreover， the proposed method significantly reduces computational time and improves overall picking efficiency， thereby verifying its effectiveness and practical feasibility.

Tool groove geometry is the core geometric element regulating cutting performance. The traditional development model relying on empirical trial and error faces challenges such as long development cycles （over 45 days）， lack of multi-physics simulation， and a performance optimization potential of 15%~20%. This paper centers on cutting simulation technology， establishing its theoretical framework and standardized application process. At the theoretical level， it elucidates the finite element method （FEM） stress-strain simulation， computational fluid dynamics （CFD） cutting fluid heat transfer analysis， and parameter calibration and application of the Johnson-Cook dynamic constitutive model. At the modeling level， based on DEFORM-3D and ANSYS Workbench， it constructs three-dimensional models of straight grooves， conventional helical grooves， and optimized helical grooves， quantitatively comparing the effects of groove types on cutting force， temperature， and chip morphology. At the parameter optimization level， it conducts sensitivity analysis on key parameters of helical grooves， clarifying the influence weights and synergistic mechanisms. The results show that，compared with straight groove， the optimized helical groove （with a helix angle of 35°， groove width of 1.6 mm， and groove depth of 2.2 mm） reduces the main cutting force by 21.9% （from 356 N to 278 N） and the maximum temperature by 25.8% （from 752 ℃ to 558 ℃）， with a chip curling radius of ≤3.2 mm and a chip evacuation smoothness score of 92. The non-equidistant helical design reduces the temperature gradient from 480 ℃/mm to 250 ℃/mm， and the thermal stress concentration factor from 2.8 to 1.9. This research provides support for precise design of tool groove geometry， promoting the transformation of cutting simulation technology from "auxiliary verification" to "core of design decision-making".

Aiming at the problems of cumbersome deployment and poor adjustment flexibility associated with QR code navigation， which is widely used in current warehouse robotics， this paper designs a global path planning method based on predefined paths and a secondary positioning and pose adjustment method based on the PL-ICP （Point-to-Line Iterative Closest Point） algorithm. The system utilizes ROS2 （Robot Operating System 2） with 2D LiDAR， IMU， and wheel odometry sensors. The methodology employs the Cartographer laser SLAM algorithm to construct a 2D grid obstacle map and robot localization data files. A self-developed RViz2 plugin is utilized to create navigation map files， upon which the A* algorithm generates global paths. The Navigation2 framework is then implemented for local path planning and motion control. Upon arrival at a designated target point， the robot calculates its pose deviation using laser feature data and corrects its position via a PID control algorithm. Simulation tests conducted on the Gazebo platform demonstrate that， after secondary positioning and pose adjustment， the robot achieves a positional accuracy of ±3 mm in the x and y directions， and a heading angle （θ） accuracy of ±0.1°.

Aiming at the failure characteristics of C-Flex bearing spring bending fatigue， a non-contact electromagnetic drive bearing life test platform based on PLC control is designed. The platform consists of mechanical system and control system. The mechanical system consists of four parts： mounting support assembly， tested assembly， driving assembly and counting assembly. The control system is developed using a model-driven design approach： a mechanical system dynamics model incorporating an experimental data-driven electromagnetic force-angle model is established， alongside a control module where the PLC utilizes a photoelectric switch to detect the bearing position signal and output DO signals. After simulation verification， this control module is automatically converted into structured text （ST） code conforming to the IEC 61131-3 standard through code auto-generation technology and deployed to an industrial controller for operation. The platform effectively solves the problems of complex structure of traditional mechanical drive and difficult realization of bearing high-frequency motion test， and provides a simple and programmable C-Flex bearing life test system.

With the rapid development of fast and high-power charging technologies， there is an increasingly urgent demand for highly reliable DC/DC converters in DC charging equipment. Traditional multi-module parallel power expansion schemes face challenges with inter-module current sharing， which restricts improvements in overall system efficiency and reliability. Particularly in low-voltage， high-current applications， the current stress and turn-off losses of switching devices increase significantly， creating an urgent need for novel topological solutions featuring zero-current soft-switching capabilities and high voltage gain. To address these issues， this paper proposes a three-phase current-fed resonant DC/DC converter based on zero-current switching （ZCS） technology. This converter achieves zero-current turn-off for the low-voltage side switches， effectively reducing switching losses and current stress. Furthermore， it ensures zero-voltage turn-on and zero-current turn-off for the high-voltage side rectifier diodes， eliminating reverse recovery losses.Research results indicate that the converter achieves full-range ZCS turn-off for the low-voltage side switches at a 2 kW power level， significantly mitigating turn-off losses and current stress. Experimental tests verify that when the load fluctuates drastically from 10 Ω to 4 Ω， the output current fluctuation rate is maintained within 5%， demonstrating excellent constant-current output performance. Compared with conventional voltage-fed topologies， this converter utilizes the inherent step-up capability of the current-fed design to achieve high voltage gain， substantially decreasing the dependence on the transformer's turns ratio.

To enhance the informational dimension of existing monocular detection systems， we designed a high-resolution image-depth information acquisition system based on RealSense. This system enables online upgrading of any current monocular camera detecting system without altering the original system’s configuration， allowing relatively simple acquisition of depth information from the original monocular camera and enabling the transition from 2D to 3D detection. To validate the method’s effectiveness， we built a hardware system using a MicroVision RS-A14K-GC8 industrial camera and an Intel RealSense D415 depth camera， and evaluated it with the proposed depth map fitting algorithm. The results indicate that， except for a small region missing depth data， the regions with depth information perform well and can essentially describe the distance differences between the target and the camera. Additionally， to assess its scalability， we constructed new hardware systems with different cameras， and they still generated satisfactory depth maps.