Irregularly shaped objects often appear inside the pipelines of nuclear power systems， and soft grippers with passive deformation ability are hence needed for grabbing operations. While the load capacity of the fixed stiffness soft gripper is usually low， the variable stiffness is an important way to improve the load capacity of the soft gripper. On the other hand， the force perception is very important for grasping fragile objects， but rigid force perception is difficult to integrate with soft grippers. Therefore， it is crucial to develop a flexible force perception function module that can be integrated with the soft gripper. Based on the characteristics of the tendon-air hybrid drive structure and the coordinated working relationship of the muscles on the upper and lower sides of the bionic human fingers， this paper proposes a design idea of a soft gripper that uses antagonistic action to change the stiffness of the soft fingers and integrates a force perception function. The designed soft gripper has three pneumatic chamber fingers， each of which is prepared by the distributed casting silicone rubber process， and the deformation process of a single pneumatic chamber finger is simulated using simulation software. To achieve contact force perception during grasping， a force sensing system based on Hall chip magnetic field intensity sensing is integrated at the tip of the soft finger， and the performance of the sensor is calibrated and tested. Through experiments， the adaptive grasping ability of the soft gripper for different target objects and the load capacity of the tendon-air antagonistic drive are tested. The experiments show that the tendon-air antagonistic drive can effectively improve the load capacity of the soft finger， and the perception module can achieve real-time measurement of the contact force of the grasping target.

Most amphibious biomimetic robots encounter problems such as insufficient motion ability， poor environmental adaptability and low simulation rate. This article adopts a novel central pattern generator （CPG） with dual neuron mutual inhibition as its main controller based on the basic rhythmic gait of salamanders， and ensures the phase coupling relationship between adjacent CPG units by adjusting the excitation suppression parameters between each neuron. Based on this， a salamander robot spinal cord like control neural network is established. The neural network consists of two layers：Interneuron and Motor neuron. The Interneuron layer generates rhythmic signals， which are then integrated by the Motor neuron layer before outputing to the joint muscle model to drive the robotic movement The performance of spinal cord control network was simulated and analyzed by combining Simulink and Webots. The simulation results show that the amphibious salamander biomimetic robot can effectively achieve rhythmic gait such as swimming and land crawling. The neural network for motion control of the salamander robot designed in this paper is feasible and effective.

High-precision and high-frequency positioning technologies are crucial for ensuring the efficiency and safety of aerial construction robots. In response to the feature-sparse aerial construction environments and the requirement to maintain operations at night， a novel localization algorithm for construction robots' exterior walls is proposed. This algorithm integrates the Extended Kalman Filter （EKF） with an Inertial Measurement Unit （IMU） and a single-line laser radar. The IMU facilitates state prediction for the filter， and a single-line laser radar positioning method adapted to building wall features is proposed for the observation update of filters. The experimental results show that the integrated positioning algorithm can improve the frequency of position and attitude estimation with the positioning accuracy being relatively high. Among them， the average absolute error is less than 4 mm when the maximum swaying of X-axis is 800 mm and the maximum swaying of Y-axis is 350 mm， while the average absolute error is less than 0.1 ° when the maximum twisting of yaw angle is 11 °， meeting the needs of actual engineering.

Traditional design optimization methods for gear always pursues the optimal results without considering the influence of uncertain factors such as manufacturing errors and usage wear， which can cause significant fluctuations in the stability of gear transmission. This paper takes the lightweight and transmission safety of the pulley of the tracked robot as the optimization goal， and uses the global sensitivity to analyze the influence among the parameters to determine the parameters to be optimized. The response surface method is used to establish an approximate model， Monte Carlo sampling is applied to make statistics on the optimization results， and the multi-objective genetic optimization algorithm is employed to solve the model， and finally， the robust optimization design of the pulley is carried out by using 6-Sigma method. Compared with the response surface deterministic method， the method proposed in this paper not only reduces the mass of the pulley mechanism but also ensures the reliability of the pulley transmission.

Aiming at the complex interaction mechanism between the polishing tool and the workpiece， and the low accuracy of the prediction model established by the regression model or the empirical formula， the ASO-BP neural network modeling method is used in the robot polishing process to predict the roughness and material removal depth of the workpiece surface after polishing， thereby solving the complex nonlinear problem between the polishing process parameters and the roughness and material removal depth. In order to quantitatively control the material removal depth while reducing the surface roughness of the workpiece， a process parameter optimization method combining genetic algorithm and ASO-BP prediction model is proposed. This method solves the dual-objective optimization problem of minimizing surface roughness and quantitative material removal depth and outputs the optimal process parameter combination. The effectiveness of the ASO-BP multi-objective prediction model and the feasibility of the process parameter optimization method combined with genetic algorithm are proved by simulation and experiment.

While the natural frequency is widely used in the fatigue damage detection of structures， it exhibits poor sensitivity to the early initiation of microcracks. This paper proposes a method for detecting and quantifying early fatigue damage by exploiting the cyclostationary properties of vibration signals induced by fatigue. Firstly， the cyclostationary properties of vibration signals under cyclic loading are investigated， and the cyclic spectral density function is utilized for qualitative analysis of the fatigue damage evolution process. Then， by demodulating the cyclostationary signals using the square envelope spectrum， a cyclostationarity index is constructed to quantitatively characterize the degree of fatigue damage. Furthermore， the changing trend of the damage index is explored in relation to different stages of crack propagation. Finally， the effectiveness of this method for early fatigue damage detection is validated through the comparison with damage detection based on natural frequency.

In response to the low accuracy and poor real-time performance of scratch defect identification in cigarette box quality inspection， a scratch feature enhancement method based on Stripe statistics and particle swarm optimization （PSO） of Gabor filter parameters is proposed. On the basis of using the gray level co-occurrence matrix （GLCM） to extract the features of Gabor filtered images， the Stripe texture statistic is designed to quantify the “scratch” feature under complex backgrounds. The image variance and Stripe together constitute the fitness function of the particle swarm optimization algorithm， and the fitness value is strongly coupled with the effect of the filter extracting scratches. A filter group composed of Gabor filters with different angles is used， the image template is obtained by calculating the average of each pixel in the multi-dimensional image output， and the image mask is used to remove background interference， thereby judging whether there are scratch defects in the image. The verification results of the image dataset collected by the optical platform show that even if the scratches are subject to different degrees of background interference， the Stripe_PSO method can still guarantee an accuracy and recall rate of over 90%， and effectively detect scratches with unobvious color differences on the cigarette box.

Cylindrical battery surface defects have a serious impact on its operational safety， and an automatic detection algorithm is proposed accordingly to address the difficulties in detecting defects on the cylindrical battery column surface. Firstly， an imaging device for the high reflective surface of the battery column is designed to avoid the influence of highly reflective metal； Then a two-stage data augmentation method is established based on image transformation and Deep Convolutional Generative Adversarial Network （DCGAN） to improve the model training accuracy； Finally， the detection algorithm is improved for the defects in both the edge parts of the battery and the tiny defects that are difficult to recognize， and the extraction ability of the edge defects is enhanced by designing the edge region feature enhancement module （ERFEM） of the battery， while the bidirectional feature pyramid network （BiFPN） is improved to merge the shallow and small target defects， and the computational amount of the detection module is reduced by using the spatial and channel reconstruction convolution （ScConv） to improve the detection speed. The experimental results show that the improved YOLOv8-EBS algorithm achieves an average detection accuracy of 93.7%， and the detection speed reaches 105 frames per second， which meets the demand for high-speed and high-precision detection.

Electronic air suspension can improve the ride comfort， the handling stability and the fuel economy of vehicles in actual driving by actively adjusting the height and stiffness of the vehicle body. In the collaborative control of the entire vehicle air suspension， four-wheel deviations occur during the height adjustment process of the air suspension due to differences in system structural parameters at each wheel and uneven distribution of the vehicle load， thereby making the vehicle's posture unstable and affecting the actual performance of suspension control. To address this issue， this article optimizes traditional control strategies and proposes a control strategy based on the integration of PI control optimized for four-wheel deviation and duty cycle correction， effectively achieving collaborative control of the entire vehicle height. This approach effectively achieves coordinated control of the vehicle's height by fully accounting for the inconsistencies in the four-wheel suspension system， resolving dynamic deviations during the collaborative control process， and improving the accuracy and stability of the air suspension height control.

With the rapid development of FDM 3D printing technology， the demand for printing efficiency and molding accuracy is increasing cross various industries. This paper proposes a motion system control strategy based on a combination of BP neural network and PID control algorithm to address the issue of insufficient forming accuracy in Delta structured FDM 3D printers. By introducing BP neural network into the motion control system of the printing device， dynamic adaptive adjustment of control parameters has been achieved， effectively improving the stability and accuracy of the printing process. The experimental results show that this control strategy significantly reduces printing errors and improves the surface quality and dimensional accuracy of the molded parts. This study provides a new solution for improving the accuracy of Delta structured FDM 3D printers， which is of important theoretical significance and practical application value for promoting the development of 3D printing technology.

This paper proposes a semi-active suspension control strategy based on the Twin Delayed Deep Deterministic Policy Gradient （DDPG） for the intelligent control problem of semi-active suspension with Continuous Damping Control （CDC） dampers. Firstly， a simulation model of a four-degree-of-freedom half active suspension system was constructed. Then， the forward and inverse models of the CDC damper were constructed. By creating a reinforcement learning training environment based on the double delay DDPG algorithm， two typical working conditions were carried out in MATLAB/Simulink environment， namely， semi-active suspension system control effect simulation experiments under typical random road surfaces and deceleration belt road surfaces， and compared with passive suspension， The root mean square values of the vertical acceleration of the spring mass in the semi-active suspension based on the double delay DDPG reinforcement learning control algorithm were reduced by 17.69% and 33.42%， respectively. The root mean square values of the vehicle pitch angle acceleration were reduced by 8.67% and 8.27%， respectively. The double delay DDPG control strategy enables the semi-active suspension system to achieve better smoothness.

The status monitoring of the electrolytic copper foil manufacturing process is of great significance to ensure the safe operation of the production system and the quality stability of copper foil. Aiming at the problems of the variation of relevant variables in electrolytic copper foil manufacturing process being nonlinear， the distribution characteristic being non-Gaussian and noise interference and the lack of consideration of product quality status characteristics in the existing multivariate statistical analysis and deep learning monitoring methods， a status monitoring method for copper foil manufacturing process based on denoising quality-supervised autoencoder is proposed. Firstly， the quality-supervised mechanism is introduced into the denoising autoencoder to extract the nonlinear and product target quality feature of the process status. On this basis， the process status monitoring statistic is constructed based on k-nearest neighbor to further extract the non-Gaussian feature of the process data， and the control limit is determined by the kernel density estimation， so as to realize the status monitoring of the copper foil manufacturing process. Taking the production and manufacturing process and the TE industrial process of an electrolytic copper foil enterprise as an example， the results show that compared with the traditional methods， the proposed method has a better monitoring effect on the status monitoring of the copper foil manufacturing process， and strong engineering value and application prospect in the actual industrial production scenarios.

To address the issues of low spatial utilization and Inbound and outbound efficiency in cargo location allocation for mechanical processing warehousing scenarios， a hybrid model （GA-XGBoost） integrating Genetic Algorithm （GA） and XGBoost parameter optimization is proposed. By constructing a two-layer encoding mechanism for feature selection and hyperparameter collaborative optimization and combining an improved greedy algorithm with dynamic priority adjustment， a multi-constraint decision model is established with optimization objectives of spatial utilization， inventory time， and prediction accuracy. The experiments based on warehousing data of 500 cargo locations and 1，200 types of goods show that the average in/out time is reduced to 17.9 minutes， representing an 18.7% efficiency improvement； the mean squared error of prediction is reduced to 0.012， and the number of convergence generations is decreased by 19.4%. This method effectively balances multi-objective constraint relationships and provides a dynamic cargo location allocation scheme for intelligent warehousing systems that coordinates high-density storage and efficient operations.

Human operation errors and process defects in the production of large tow carbon fiber will lead to defects such as long and short filaments， hairballs， joints and stuck pulp on the surface of large tow， thereby affecting product quality and even causing security risks in the production process. The on-site vision system， however， is merely capable of detecting defects and saving pictures without realizing the classification and positioning of defects. Therefore， based on the actual application scenario of surface defect detection of large tow carbon fibers， this paper proposed a lightweight tow surface defect detection algorithm based on YOLOv5s. Firstly， MobileNetV2 was introduced into the backbone network of the original model for lightweight improvement， and ODConv dynamic convolution module was inserted to improve the model performance. Secondly， the Dyhead dynamic detection head structure was used to replace the original detection head， which could improve the model performance without excessive increase in computation. Next， CARAFE upsampling operator was introduced to replace the original nearest upsampling operation to improve the aggregation effect of model feature information. Finally， the ablation experiment and comparison experiment were carried out on the dataset of the self-made large tow carbon fiber surface defect. The experimental results show that the improved lightweight algorithm proposed in this paper has higher running speed and detection accuracy than that of the other three classical models， providing a new method and idea for solving the surface defect detection problem of large tow carbon fibers.

To realize the effective continuous target detection and tracking of the insulator flushing robot in the occlusion environment， a detection and tracking method based on computer vision in the occlusion environment is proposed. Firstly， the attention mechanism is added to the YOLOv5 detection algorithm to enhance the recognition accuracy of the detection algorithm， the scale filter in the DSST algorithm is then combined with the KCF tracking algorithm to make the KCF scale adaptive. Thereafter， the Multi-PROSAC-ORB occlusion recognition algorithm is constructed to realize the occlusion recognition. Finally， the above three algorithms are fused and an occlusion judgment condition is proposed to ensure the continuous and stable recognition and real-time performance of the target in the case of occlusion. The experimental results show that the proposed method can effectively avoid the low accuracy of target tracking in the occlusion environment while ensuring the real-time performance， and the target tracking accuracy is increased by 10.9% and the tracking success rate is increased by 13.6% compared with the target tracking accuracy in the unocclusion recognition， which has high accuracy and real-time performance.

Unmanned aerial vehicles （UAVs） can be used as signal relays， and outdoor UAVs can provide signal coverage for indoor users in emergency situations. The existing research on UAV signal coverage mostly focuses on indoor users in specific locations or outdoor users， and the proposed methods available are not suitable for the signal coverage scenarios of indoor users. In the case of the random distribution of indoor user locations， a single-hovering-plane and multi-hovering-plane particle swarm optimization （PSO） algorithm for UAV deployment is proposed to determine the optimal hovering position and transmission power of a single UAV， thereby optimizing the deployment strategy of a single UAV. Furthermore， to address the issue of limited transmission power of a single UAV， a method combining K-Means and PSO is proposed to achieve multi-UAV signal coverage for indoor users. The simulation experiments show that the proposed algorithm reduces the power required for a single UAV to cover indoor users in single-UAV scenarios and optimizes the transmission power and number of UAVs in multi-UAV scenarios.

This paper addresses the challenge of inadequate timeliness and accuracy in processing multi-source heterogeneous data within special equipment assembly workshops， which tends to hinder real-time transparent management and control during the assembly manufacturing process of specialized equipment. To tackle this， a method for fusing multi-source heterogeneous data in these workshops is proposed with the approach based on an analysis of the composition and characteristics of operational data in these workshops， and the benefits of Multi-Agent technology is applied to construct a framework for data fusion. The research on the methods involved in data layer fusion and feature layer fusion is also conducted. The feasibility and effectiveness of the proposed method are confirmed through simulation examples， thereby providing reliable， timely， and accurate data support for the intelligent operation and control of special equipment assembly workshops.

The assembly workshop for the nose section features pulsating assembly processes and synchronized unit production， with on-time delivery being a key condition for ensuring the smooth operation of the assembly line. The assembly of civil aircraft nose components is currently a semi-automatic assembly system and human resource allocation still relies heavily on experience， which can lead to mismatches between resources and production tasks. With the increasing production requirements for nose sections， the task quantity and complexity of personnel allocation are also continuously increasing， necessitating an effective allocation method. Addressing the human resource allocation problem for component assembly units under pulsating rhythm constraints， this study considers constraints such as assembly line numbers， start times of each line， and pulsating rhythm on the timely completion of component assembly and delivery. Additionally， spatial constraints and technological constraints of the component assembly unit are considered in constructing a mathematical model with the objective of minimizing human resource costs. Based on dynamic work hour estimation， a hybrid algorithm using simulated annealing and particle swarm optimization is designed to address integrated problems of team scheduling and task assignment， and the feasibility of the model and the superiority of the algorithm are verified through simulation.

For the issue of point cloud registration of multiple LiDAR scans in large bulk cargo piles at ports， this paper proposes an automatic point cloud registration algorithm that combines coarse and fine registration. This is achieved through an initial coarse registration using Sample Consensus Initial Alignment （SAC-IA） based on an improved Fast Point Feature Histogram （FPFH） and a fine registration using Iterative Closest Point （ICP） accelerated by k-dimensional （k-d） trees. The process starts with filtering and downsampling the 3D point cloud data to address the issues of noise and large data volumes in large bulk cargo pile point clouds. For the problem of large initial positional differences， an initial coarse registration algorithm based on the improved FPFH is proposed. Finally， to address the long duration of fine registration， an ICP algorithm accelerated by k-d trees is presented. The experimental results show that compared to the ICP algorithm， the 4PCS algorithm and the SAC-IA-ICP algorithm， the proposed registration algorithm reduces the registration time for ship cargo piles by 94.3%， 93.3%， and 66.59%， and for yard cargo piles by 81%， 90.13%， and 47.99%， respectively. The registration errors for ship cargo piles are reduced by 84.39%， 1.25%， and 28.19%， and for yard cargo piles by 90.34%， 3.15%， and 13.04%， respectively， demonstrating its effectiveness in registration.

In response to the challenges posed by the high annotation costs and the fluctuating annotation accuracy associated with the traditional manual "capture-annotate" approach used in the unordered sorting logistics process of smart factories， an automated data generation method is proposed. This paper focuses on the algorithm for automatic generation of image datasets， establishing a simulation environment based on the Pybullet open-source physics simulation engine. Through the configuration of virtual environment rendering parameters and processing of segmented images captured by virtual cameras， the goal of generating high-quality image datasets and annotation files is achieved. The dataset generated from the virtual environment is employed to train the YOLO v5m target detection network， and the trained model is subsequently applied to real scene images captured by the Kinect v2 camera. The experimental results demonstrate a 92.1% detection rate for objects， 98.2% correct classification rate for two categories of objects， and an average recognition accuracy of 97.6%. The time taken to generate a single image and label file is less than 1 second， making this approach more suitable for efficiently generating standardized image datasets compared to manual methods.

Bird activities have threatened the stable operation of electrical equipment in the substations. To address the current equipment's low detection accuracy and poor bird-repelling effect， this article proposes a method that combines an audio-visual perception module， an acoustic-optic bird repelling module and the remote control. The module-in-one intelligent bird detection and repelling robot system uses the time difference of arrival （TDOA） algorithm based on particle swarm optimization to realize the sound source positioning of the microphone array and applies the improved YOLOv5 and DeepSORT algorithms to achieve precise positioning and tracking of bird targets. Finally， the targeted bird repelling is achieved based on acoustic-optic equipment. The experimental results show that the distance estimation accuracy of the sound source reaches 96.42%， the phase estimation error is less than 1.1 °， and the visual recognition accuracy reaches 89.7%. The fusion of audio-visual modal data effectively improves bird detection accuracy. The problem of blind spots in the visual field existing in the bird detection process can be solved faster and more accurately to complete the task of detecting and driving birds.