In the discrete control system of the robotic arm， the traditional sliding mode method often encounters high-frequency chattering problems and control failure caused by input saturation. To break through this bottleneck， this paper proposes a trajectory tracking control method for the robotic arm based on anti-saturation sliding surface and adaptive reaching rate， aiming to address the influence of input saturation and external disturbances on control performance. By directly integrating the anti-saturation suppression factor into the design of the discrete sliding surface， dual constraints control over both input amplitude and input variation rate are achieved， and an adaptive reaching rate is constructed to suppress the chattering in the traditional discrete sliding reaching law， improving the stability and tracking accuracy of the system under disturbances. The hyperbolic tangent function is used instead of the sign function to further eliminate high-frequency chattering phenomena and improve control smoothness. Simulation results show that the proposed control method achieves smaller joint position and velocity errors on a two-degree-of-freedom robotic arm， converging within 0.35 seconds， with steady-state error controlled within 0.01 rad， and the maximum steady-state error is reduced by 74.16% and 68.71% compared with the traditional method. The root mean square error is reduced by 57.47% and 19.07%， respectively， verifying its superiority in anti-saturation and chattering suppression. The research in this paper provides an effective control strategy for the control application of robotic arms in complex environments with input saturation and disturbances， with high application value and engineering significance.

In order to maintain the bending ability of the pneumatic soft gripper while improving the fingertip contact force of the fingers， a dual module composite pneumatic soft gripper with rope-traction and collaborative grasping was designed. The finger combines two different pneumatic network structures， a slow pneumatic network and a fast pneumatic network. The front end of the finger uses a fast pneumatic network to ensure the bending angle， while the back end uses a slow pneumatic network to provide greater fingertip contact force. At the same time， a rope traction drive is added to further improve the fingertip output force during finger grasping. A mathematical model was established based on the Yeoh model to describe the relationship between finger bending angle and input air pressure. ANSYS was used to perform finite element simulations of the bending angle and fingertip contact force of the designed dual module composite pneumatic soft finger under different air pressures. A physical object of soft fingers was manufactured using 3D printing and deposition casting. The bending angle and fingertip contact force of the fingers were measured， and grasping experiments were conducted using a gripper. Experiments have shown that the maximum bending angle of a single soft finger can reach 164°， and the maximum fingertip contact force is 2.75 N. The grasping results show that this claw has strong adaptability to objects of different shapes and good load capacity， and has good grasping effects on objects of different diameters.

Aiming at the difficulty in hooking flexible wire loops due to their dimensional and pose uncertainties during the continuous looping process of thick carbon fiber puncture fabrics， this paper proposes a machine vision-based algorithm for estimating the dimensions and pose of flexible bodies. First， to address the challenge of binocular matching caused by the lack of distinct texture on the surface of flexible wire loops under a dual-view system， an improved Space-Carving method is proposed to obtain sparse point clouds of the target. Then， to overcome the issue that fixed models cannot be used for point cloud registration due to the dimensional uncertainty of flexible wire loops， a parametric model of spatial curves for flexible bodies is introduced. Using the sparse point clouds as input， the Levenberg-Marquardt optimization algorithm is iteratively applied to solve the parameter vector of the spatial curve， thereby estimating the dimensions and pose of the flexible body. Experimental results demonstrate that the improved space carving algorithm can still achieve relatively accurate sparse point clouds even with a limited number of viewpoints. By fitting the designed spatial curve model to the sparse point clouds， stable and accurate estimation of the dimensions and pose of carbon fiber wire loops is achieved， with precision meeting the requirements for subsequent automated hooking processes.

To enhance the compliance of human-robot interaction and the responsiveness to operational intent in collaborative robots operating within free space， this paper proposes an adaptive impedance control strategy based on operational intent. Current research on adaptive impedance/admittance control suffers from insufficient system stability analysis， reliance on additional biosensors， or excessive algorithmic complexity， thereby limiting its application in practical dynamic interaction scenarios. This paper aims to minimise energy consumption during human-robot interaction. By analysing the impact of operational forces and end-effector velocity on system power， a dynamic mapping relationship between damping parameters and operational intent is derived. An adaptive impedance model is established that adjusts its own damping coefficient based on load force and motion velocity， achieving low damping， high velocity， and high responsiveness during dragging operations， while maintaining high damping for precision during manipulation of delicate components. Trailing experiments demonstrate Pearson correlation coefficients exceeding 0.9789 for force-displacement responses across all directions at the robotic end-effector. Under irregular external forces， motion remains smooth with rapid adaptive impedance parameter adjustments， validating this strategy's effectiveness and robustness within dynamic， uncertain human-robot interaction scenarios.

Aiming at the problem of insufficient accuracy in datum plane fitting for annular end faces of industrial pipes and vessels during automated assembly， welding， and online inspection， caused by boundary noise， local deformation， and non-uniform sampling， a curvature-topology dual-driven datum plane fitting method is proposed. The method extracts principal curvatures based on spectral decomposition of the local covariance tensor， and constructs planarity and degeneracy indices to achieve non-uniform sparsification of the point cloud. Furthermore， Topological Data Analysis （TDA） is introduced， where a Vietoris-Rips complex is constructed and persistent homology analysis is performed to extract a skeletal point set with closed topological features， thereby embedding the structural prior of the annular end face. The plane normal vector is determined from the eigenvector associated with the smallest eigenvalue of the covariance matrix computed on the skeletal point set， achieving highly robust pose estimation. Experiments show that the proposed method achieves a mean absolute angular error of less than 0.03° and a single fitting time of less than 0.03 s. It is approximately 300 times faster than the Least Median of Squares （LMedS） method and improves accuracy by about 10 times compared with traditional methods such as Principal Component Analysis （PCA） and Random Sample Consensus （RANSAC）. This method requires no manual intervention， offers high accuracy， strong robustness， and real-time processing capability， making it suitable for automated positioning and datum establishment of pipe end faces in complex industrial environments. It can provide reliable technical support for welding path planning and online inspection in intelligent manufacturing.

This study addresses the challenges of recognition and pose estimation in the automatic water washing of railway catenary insulators by proposing a multi-stage fusion algorithm based on Lidar point clouds. At the target recognition level， the Part-A2 detection algorithm architecture is employed， incorporating a foreground point relative position encoding mechanism to significantly enhance the deep learning capability of local geometric features of insulators， thereby improving recognition accuracy. For pose estimation， an innovative RANSAC-OLS cascaded fitting strategy is proposed： first， the insulator ROI point cloud is extracted based on the spatial coordinates output by the Part-A2 algorithm； then， the RANSAC algorithm is used for coarse axis fitting to eliminate initial value sensitivity； finally， the spatial pose of the insulator is precisely estimated using the least squares method in conjunction with the insulator's external shape mathematical model. Experimental results demonstrate that the proposed solution achieves an mAP recognition accuracy of 82.25% within the operating range of 7~20 meters， representing an improvement of 8~25 percentage points over traditional point cloud models. In terms of pose estimation， the yaw and pitch angle errors are controlled within 8°， fully meeting the accuracy requirements for automatic water washing equipment alignment.

Aiming at the collaborative optimization problem of flexible job shop scheduling and AGV material handling in intelligent factory under multi-variety and small-batch production mode， a joint scheduling method of production and handling based on Dueling Double Deep Q-Network （D3QN） is designed to minimize the maximum completion time， and the conflict-free path planning of AGV is realized by combining the A * algorithm with time window. The design algorithm is compared with a variety of rule scheduling methods， and a variety of different scale examples are designed for experimental verification. The results show that the scheduling performance of D3QN algorithm is better， and it has good optimization effect and generalization ability. At the same time， the influence of the number of AGVs on the scheduling performance is analyzed， and finally it is concluded that it conforms to the law of diminishing marginal benefit.

To address the problems of long training time， slow convergence， and low success rate of reinforcement learning （RL） in mobile robot navigation tasks under dynamic environments， this paper proposes a reinforcement learning navigation performance optimization method based on Convolutional Long Short-Term Memory （ConvLSTM） networks. The proposed approach employs ConvLSTM to perform spatio-temporal modeling of historical occupancy grid sequences， predicting future environmental changes and generating a risk cost map. A risk-guided reward function is then designed to enable the agent to identify potential collision areas in advance and make foresighted decisions. This method preserves the adaptability of reinforcement learning in complex environments while significantly reducing ineffective exploration during training. Comparative experiments conducted in the Gazebo simulation environment demonstrate that the proposed method improves convergence speed by approximately 40%. The navigation success rate improves by approximately 9%~15% during the 8~10 hour training phase. The results verify the effectiveness of dynamic risk prediction based on ConvLSTM in accelerating reinforcement learning navigation， providing a practical and efficient solution for safe navigation of mobile robots in complex dynamic scenarios.

To address the issues of inefficient resource allocation， prolonged design cycles， and high costs resulting from static task-personnel matching in the collaborative design of complex heavy equipment， this paper proposes a proficiency-based task-personnel matching method： First， a structured information model incorporating multi-dimensional task characteristics and dynamic personnel attributes is constructed. Next， proficiency calculation equations for both general personnel and each senior individual are fitted by considering task similarity， historical completion frequency， personnel grade， and other factors， with validated effectiveness. Constraints such as skill requirement matching and workload values are introduced to ensure every assigned personnel meets task demands. Finally， an optimized task-personnel matching model is constructed with the objective of minimizing normalized total time and total cost. This model is solved using an elite retention genetic algorithm， yielding optimal matching solutions that guide subsequent personnel allocation. Case studies demonstrate that this method achieves synergistic optimization of efficiency， cost， and constraint satisfaction， validating its superiority in complex heavy equipment collaborative design scenarios.

To address the issues of low efficiency and susceptibility to local optima in the layout planning and optimization of casing production workshops， a mathematical model for the layout planning and optimization of aero-engine casing workshop was developed with the objective of balancing space utilization and logistics costs， incorporating constraints from the casing production process and workshop logistics pathways. Considering the complexity of both production and logistics in casing manufacturing， as well as the characteristics of genetic algorithms and harmony search algorithms， an integrated approach combining Systematic Layout Planning （SLP） with an improved genetic-harmony search algorithm （SLP-GA-HS） was proposed. The procedural flow and specific implementation of the algorithm were designed in detail. Using the layout planning and optimization of a specific casing production workshop as a case study， a workshop layout optimization model was constructed. Analyses of logistics and non-logistics were conducted， followed by optimization algorithm solving and comparative algorithm analysis， demonstrating the effectiveness of the proposed method.

In the context of intensified international competition， accelerated technological iteration， the pursuit of "dual carbon" strategic objectives， and the green and low-carbon transformation of energy systems， large-scale energy group enterprises are confronted with practical challenges such as insufficient integration of digital architecture and inefficient management of architectural assets. To address persistent pain points in the application of TOGAF within large enterprises—such as inconsistent granularity across architecture domains and inadequate alignment between digital architecture development and business operations—this paper proposes a construction methodology based on an enhanced meta-model. This approach enables unified technical specifications， standardized business rules， and model-driven asset representation， thereby establishing a cohesive enterprise digital architecture. The proposed framework facilitates seamless integration from architectural design to digital implementation of business processes， validates the effectiveness of the method in enhancing the efficiency and collaborative capabilities of enterprise architecture asset management， and offers both theoretical foundations and practical pathways to support the digital transformation of large energy group enterprises.

Simultaneous Localization and Mapping （SLAM） tasks are typically based on the assumption of a static environment and often suffer from severe pose estimation accuracy degradation when disturbed by dynamic objects. This paper proposes a visual SLAM algorithm that integrates dynamic feature values and mask distribution values， with the following key innovations： 1） An environmental assessment mechanism based on the Exponential Weighted Moving Average （EWMA） of dynamic feature values is proposed， which achieves accurate and robust perception of environmental dynamics by integrating historical reprojection errors. 2） A lightweight improved YOLO-seg semantic segmentation method is employed for rapid dynamic object identification and pixel-level mask generation， significantly improving efficiency while maintaining accuracy. 3） A local Bundle Adjustment （BA） optimization strategy based on mask distribution values is designed， which constructs a weighting coefficient by introducing mask compactness and area ratio， effectively mitigating the optimization imbalance caused by feature point removal. 4） An adaptive multi-threaded collaborative system architecture is constructed， enabling on-demand triggering of the dynamic judgment and semantic segmentation threads and optimized resource scheduling. Experiments on the TUM RGB-D dataset demonstrate that in highly dynamic scenarios， the proposed algorithm reduces the absolute trajectory error by up to 94.1%， 38.8%， and 77.3%， compared with ORB-SLAM3， Dyna-SLAM， and YOLO-SLAM， respectively， significantly enhancing the accuracy， robustness， and real-time performance of SLAM in dynamic environments.

This study proposes a new discrete element parameter calibration method to solve the problem of calibrating discrete element simulation parameters for metal ore transfer chute. Firstly， the geometric shape of ore particles is classified and described using the Zingg classification method， and a multi-ball cluster model of particles is constructed. Then， the static angle of repose of the ore was measured using the funnel feeding method and the L-shaped box feeding method， respectively. Two second-order response surface models （RSMs） of the static angle of repose discrete element models were constructed through Latin square experimental design. Finally， the genetic algorithm is used to obtain the optimal combination of particle contact mechanics parameters. Based on the calibrated ore parameters， the discrete element analysis was conducted on an iron ore transfer chute in Pangang Group Mining Co.， Ltd. The simulation results is basically consistent with the actual working condition. The research results indicate that the discrete element parameter error calibrated by this method is small， and it has good calculation accuracy in the simulation calculation of metal ore transfer hoppers， which can meet the engineering design calculation requirements and verify the effectiveness and practicality of this calibration method.

Addressing the industrial challenges of geometric-semantic fragmentation， low dynamic updating efficiency， and the trade-off between high precision and lightweight representation in 3D reconstruction of power grid equipment， this paper proposes a novel method integrating Neural Radiance Fields （NeRF） with lightweight Building Information Modeling （BIM）. Firstly， a cross-modal alignment mechanism is designed： NeRF parses multi-source data to generate a high-fidelity density field， while an innovatively introduced Attribute-Aware Attention Module dynamically embeds BIM semantic parameters into the NeRF feature space. This achieves joint representation of geometric structures and equipment attributes， resolving semantic absence. Secondly， an incremental updating algorithm is proposed： local change regions are identified via structural similarity index （SSIM） difference analysis and BIM change logs. The adaptive learning rate pruning optimizer AdaBound fine-tunes local NeRF parameters， and a Git-style incremental storage strategy reduces single-device update latency to 3.2 minutes while cutting storage overhead by 98%. Lastly， a cloud-edge-end collaborative architecture is constructed： The model is compressed to 0.3 M parameters via knowledge distillation. Coupled with a static-dynamic hierarchical rendering strategy， mobile-end loading latency of 180 ms and an interactive frame rate of 30 FPS are achieved. Experiments in a 220 kV substation scenario demonstrate superiority over traditional BIM and standard NeRF： Chamfer Distance drops to 2.03 cm， semantic recall reaches 98.7%， and transmission success rate under weak network conditions rises to 98%. This method provides key technical support for smart grid inspection and dynamic operational analysis.

Aiming at the problem that current riveting quality detection systems typically only monitor the maximum riveting force and effective riveting displacement during the riveting process， which fails to effectively identify potential defects， this paper designs a real-time quality detection system for the electric riveting gun process， achieving real-time quality monitoring of the riveting process. Based on an analysis of the three stages of the riveting process， key characteristic points are extracted， and a multi-feature zoned quality judgment model is constructed. A BP neural network is utilized to establish a nonlinear mapping model between the riveting force and current， enabling the prediction of the riveting force. An upper computer integrated with the neural network prediction model and multi-feature quality judgment logic is developed to realize real-time quality detection of the riveting process. Experimental results show that by judging the three key points in the riveting process， this system enhances the automation level and accuracy of riveting quality detection， providing an effective solution for quality monitoring of connection processes in smart manufacturing.

To address measurement accuracy limitations in ultrasonic oil level detection systems caused by probe installation tilt （0°~5°）， mechanical vibrations （0-100 Hz）， temperature drift （-30-80 ℃）， and oil quality degradation， this study proposes an intelligent compensation algorithm utilizing high-precision angle sensor fusion. The system employs the ADIS16470 MEMS high-precision angle sensor to monitor ultrasonic probe tilt in real-time， combined with a dynamic tilt compensation model that eliminates beam pointing errors from installation deviations and mechanical vibrations. Adaptive Kalman filtering integrates multi-modal sensor data to achieve noise suppression and error correction for dynamic oil level measurement. A deep learning-based nonlinear mapping model between oil level and acoustic propagation time is developed to overcome the limitations of traditional linear calibration methods. Experimental results demonstrate that under multiple disturbances including 10-200 Hz broadband vibration， temperature fluctuations （-30-80 ℃）， and oil quality aging， the system achieves breakthrough performance with 85 ms step response time， 0.3 mm steady-state error， and 150 Hz vibration suppression bandwidth. This significantly enhances the reliability and automation level of transformer oil level monitoring.

To address the challenges of high human intervention risks， insufficient stiffness and poor environmental adaptability of conventional automated equipment， in the sealing operations of nuclear waste containers， this study proposes a high-stiffness modular 3-DOF robotic system for closure grasping and sealing. The mechanical system comprises a base frame， locomotion mechanism， lifting mechanism， and an integrated end-effector （combining rotary， gripping， and tightening functions）. Innovations include a dual closed-loop linkage lifting mechanism and a worm-gear-planetary compound reduction system， enabling precise and reliable power transmission under heavy-load conditions. A closed-loop vector equation for the lifting mechanism is established， and a genetic optimization algorithm is employed for multi-objective optimization design of structural dimensions， thereby determining the key structural dimension parameters. The key structural dimensions and parameters have been determined using a genetic optimization algorithm. Parametric finite element models were established to evaluate the mechanical performance of critical components： the base frame exhibits a maximum von Mises stress of 6.445 MPa （safety factor： 36.4）， the mobile chassis shows a deformation of 0.055 mm， and the tie-rod system achieves a maximum stress of 3.479 MPa （safety factor： 67.5）， all significantly below the allowable stress of Q235 carbon steel. The integrated end-effector platform demonstrates a maximum stress of 6.747 MPa （safety factor： 34.8） and a deformation of 0.0139 mm. This research provides a numerical simulation based structural design methodology and mechanical evaluation framework for automated sealing operations in radioactive environments， ensuring compliance with the stringent demands of nuclear industrial applications.

To address the engineering issue of stress concentration at the valve stem and seal ring in DN400 triple-offset metal-sealed butterfly valves under 2 MPa reverse medium pressure， leading to structural deformation and seal failure， a systematic optimization design method balancing structural strength and sealing performance is established. To this end， a multi-objective optimization framework coupling “Finite Element Analysis （FEA） – BP Neural Network - NSGA-II” is proposed. First， sample points were generated within the design space of cone apex angle， third eccentricity angle， and seal ring thickness， using Latin Hypercube sampling. Finite element simulations were then performed to calculate stress responses in the valve stem and seal ring. Second， using the finite element results as training data， a four-output BP neural network proxy model is constructed to approximate the structural stress response patterns. Subsequently， the NSGA-II algorithm performs global optimization under the constraint of a minimum sealing pressure not less than 12.48 MPa. Balanced solutions are selected for finite element verification and experimental validation. Results demonstrate that the optimized design reduces peak stresses on the valve stem and seal ring by 21.5% and 58.8%， respectively， with no measurable inelastic deformation （less than 0.02 mm）. It achieves Class A bidirectional zero leakage in 2 MPa water pressure tests， significantly enhancing the valve's safety margin and sealing reliability. The study concludes that this integrated “FE—BP—NSGA-II” framework effectively supports multi-objective optimization design for triple-offset butterfly valves under high-pressure reverse flow conditions， providing a replicable engineering methodology for reliability design and intelligent development of large valves.

Kitchen appliances represent an important export category for China， and high-pressure foreign trade policies have put higher demands on product innovation. Taking the turntable air-frying microwave oven with integrated features as the research subject， economical innovative structures were constructed by applying DFX methodology. In terms of Design for Cost （DFC）， to address the issues of high door-assembly cost and light coloring of the bottom surface of food， a new air-frying-assembly featuring rear-mounted heating tubes and a volute air duct design was proposed， overcoming the design idea limitation of symmetrical heating tubes. Meanwhile， a rotating cooking accessory was designed and the air outlet area of the cavity was optimized to address the shortcomings of the new scheme. These measures reduced the temperature rise and material cost of the door-assembly， while also increasing the French-fry cooking score from 75 to 85. In terms of Design for Manufacturing and Assembly （DFMA）， given that the cavity system had many parts and complex processes， TRIZ theory was used to simplify the cavity system structure and eliminate painting and assembly processes. A 45 mm wall bracket was added to the rear of the cavity system to address the temperature rise at the test angle. The above innovative measures have reduced overall product cost by 10%， which effectively ensures product competitiveness in the global market.