Aiming at the problems of large randomness,slow convergence,and poor path quality of traditional RRT-Connect algorithm in path planning,a RRT-Connect robotic arm path planning algorithm based on improved APF guidance was proposed. Firstly,a dynamic bias strategy was introduced to reduce the problem of large randomness of the original RRT-Connect algorithm; secondly,when expanding a new node,the node randomness and the weight of the total potential field were adaptively adjusted according to the distance from the obstacle; then,a robotic arm collision detection model was incorporated and a two-way direct-connected linear interpolation detection was carried out until the initial path was generated; finally,three times of B-spline smoothing was carried out on the generated path to produce feasible paths that conform to the actual motion constraints of the robotic arm. 3 different environments were constructed in 2D and 3D to simulate and to compare with the improved algorithm,the results showed that the improved algorithm could effectively improve the relevant performance. The application of the improved algorithm to the physical platform further proves the effectiveness and feasibility of the algorithm.

The study of the artificial intelligence algorithms for job shop scheduling has gained attention due to the advancements in intelligent manufacturing technologies like smart factories. Dynamic events in the job shop are crucial factors affecting scheduling effectiveness. To this end,it proposes a novel approach employing the deep reinforcement learning to solve the dynamic flexible job shop scheduling problem with random job arrival. Initially,a mathematical model is formulated for the dynamic job shop scheduling problem with the objective of minimizing the total tardiness. Subsequently,eight job shop state features are extracted,and six composite scheduling rules are designed. An ε-greedy action selection strategy is adopted,and the reward function is designed. Finally,the advanced D3QN algorithm is introduced to solve the problem and the effectiveness of this method is verified on different scale of instances. The results show that the D3QN algorithm effectively solves the dynamic flexible job shop scheduling problem with random job arrival,and the winning rate in all instances is 58.3 %.Compared with traditional DQN and DDQN algorithm,the total tardiness is reduced by 11.0 % and 15.4 % respectively,which proves that this method further enhances the production efficiency of the job shop.

In order to improve the operation efficiency of multi-AGVs in the automatic packing system, a two-stage online cooperative scheduling algorithm is proposed to solve the multi-AGV online scheduling problem, which is based on the optimization objective of minimizing the AGV operation time combined with the constraints of the actual working conditions. The algorithm is developed based on the simulation model of the automatic packing system. Firstly, the handling task assignment algorithm based on the AGV running time is used to solve the handling task assignment problem; secondly, the path planning algorithm with AGV priority rules and conflict resolution strategies is designed to solve the path planning problem; finally, the Spatio-Temporal Blocking Table (STBT) is used to record the Spatio-Temporal Blocking Degree (STBD) of the paths and the estimated waiting time, and the information in the table is incorporated into the two-stage collaborative scheduling algorithm as the constraints. The effectiveness of the proposed algorithm is verified by simulation cases with different scales, and the superiority of the proposed algorithm is verified by comparison experiments with related research results.

To address the inefficiencies in process route design and the difficulty in reusing knowledge in the machining of precision transmission components,a knowledge graph-based process route recommendation model was proposed for precision transmission components. Firstly,an ontology-based method was used to construct the process knowledge schema layer for precision transmission components.Secondly,entity recognition and relation extraction were achieved using the Electra+BiLSTM+CRF model and the BiLSTM+Self-Attention model,respectively,and Damerau-Levenshtein distance was used for knowledge fusion,completing the data layer construction. Then,based on the constructed process knowledge graph,process route recommendations were achieved by combining the similarity of component process units and process route structures. Finally,a process route recommendation system was developed and demonstrated with an example of a specific type of ball screw. Experimental results show that the recommendation accuracy reaches 89.5 %,proving the feasibility of the model,which can enhance knowledge reuse and improve process route design efficiency,providing a more scientific and reasonable reference for decision-making.

Addressing issues such as low planning efficiency,long search time,and complex paths in improving intelligent optimization algorithms,the egret swarm optimization algorithm was first applied to mobile robot path planning,and a new method for mobile robot path planning based on the improved egret swarm optimization algorithm was proposed. In the exploration stage,the algorithm utilized adversarial learning for population initialization to reduce the cost of path search;the sine cosine algorithm and greedy strategy were used to improve the updating of individual positions of egrets,in order to balance the local development and global search capabilities of the algorithm;coordinate fine-tuning strategy was used to obtain a safe and reliable planning path. In the optimization stage,the vertical distance limit method and segmented Bessel curve were used to optimize the path,in order to obtain the final motion path of the mobile robot. The simulation results show that this algorithm significantly improves the efficiency of path planning compared to the comparative algorithm,with shorter overall time consumption and better paths. It can reduce the number of path turns and thereby improve the overall work efficiency of mobile robots.

Aiming at the current situation that the absolute positioning accuracy of industrial robots is not high,a method of Kriging prediction of positioning error with comprehensive consideration of geometrical and non-geometrical error influencing factors was proposed. Firstly,a 3D dynamic tracking measurement system was used to measure the position errors of a set of points in the workspace of the industrial robot,and the experimental variational function was constructed from the position error; secondly,the particle swarm algorithm combined with the Kriging method was used for parameter optimization of the experimental variational function to obtain the optimal model and parameters; then,the weight coefficients and Lagrange multipliers were solved by Kriging equation combined with the optimal experimental variational function model,and the Kriging prediction equation was utilized to predict the positioning error;finally,the positioning errors were compensated and the results were veried. Experimental results show that the average absdute positioning error decreased from 1.048 9 mm to 0.178 6 mm,with an increase in accuracy of 82.97 %; the root mean square error decreased from 0.393 7 mm to 0.058 5 mm,with an increase in accuracy of 85.14 %,verifying the efficiency and practicality of this method.

To ensure the smooth operation of the car, a method for controlling the instability of the car′s lateral angular velocity based on the onboard angular velocity sensor was proposed. Firstly, the yaw rate data of the vehicle was obtained using the in-vehicle angular velocity sensors, and based on this, observational signal descriptions and analytical equations were establi-shed. Then, under the theory of yaw dynamics, evaluation indices for vehicle yaw rate instability were set, and the actual yaw rate of the vehicle was compared with its ideal value. The difference between the actual observed yaw rate and the ideal value was converted into fuzzy variables through fuzzy rules, and the evaluation indices were fused using fuzzy sets. Finally, based on the fusion results, the control torque for the vehicle's yaw rate was determined, achieving control over vehicle yaw rate instability. The results show that using the snake shaped working condition as the test case, the vehicle's yaw rate controlled by the studied method is less than 15 rad/s, which does not exceed the limit value, and can ensure that the vehicle wheels change under the same standard, effectively controlling the stability of the yaw rate and ensuring the smooth operation of the vehicle.

With the extensive operation and rapid increase of high-speed trains running mileage in China,issues such as wheel-rail wear,changes in operating environments,and mismatches in suspension parameters have led to an increase in abnormal vibrations in train bodies. These issues significantly reduce the ride comfort of passengers,with insufficient bogie hunting stability being particularly prominent. To address these problems,research on adaptive control strategies for hunting stability of high-speed train from the perspective of active control of yaw dampers is conducted. Firstly,yaw dampers are replaced with actively controlled suspension components,and a nonlinear dynamic model of the vehicle system is established. Then,an adaptive fuzzy PID controller is designed by integrating PID control principles with fuzzy control principles,and the parameters are optimized using a non-dominated sorting genetic algorithm with an elite strategy. Finally,based on a co-simulation model using SIMPACK and Simulink,the control effects of the designed active control system are compared and analyzed under different operating conditions. The results show that,with the active yaw dampers,the peak amplitude of the lateral acceleration at the end of the bogie frame is reduced by up to 54 %,and the Root Mean Square (RMS) value is reduced by up to 33.24 %,significantly improving the lateral vibration performance of the bogie frame and providing technical support for bogie hunting stability control strategies.

Radial plunger hydraulic motor is widely used in medium and large mechanical equipment.However,due to its internal impact and fatigue wear problems,the life and performance of radial plunger hydraulic motor are affected to some extent.To solve these problems,an improved particle swarm optimization for optimization radial plunger hydraulic motor internal curves was proposed. The constant acceleration curve was reconstructed into the constant acceleration curve with compensation region to reduce the impact and the sudden change of the constant contact stress. Based on the Particle Swarm Optimization (PSO),an improved particle swarm optimization was constructed by adding adaptive nonlinear dynamic weight and multi-subpopulation competitive optimization strategy. The angle of each zone was reassigned and the internal curve of radial plunger hydraulic motor with compensation zone was reconstructed. The results before and after optimization show that the maximum contact stress decreases by 2.54 % and the abrupt value at the maximum contact stress decreases to 0. The contact stress no longer rises step by step,there is a rising process,and the impact is small. This research can provide reference for the design of radial plunger hydraulic motor,effectively reduce fatigue and wear,reduce impact,and extend the service life of hydraulic motor.

In complex component assembly scenarios based on augmented reality assembly guidance,the occlusion of the parts by the hands leads to significant errors and even failure in the pose calculation of the parts. At present,pose estimation algorithms for manually assembled parts do not consider the use of hand information when solving part occlusion problems,making it difficult for pose estimation accuracy to meet the requirements of augmented assembly. In response to the above issues,the article proposes a 6D pose estimation algorithm for parts that integrates hand posture,namely the HandICG algorithm. This algorithm integrates the pose information of the hand with the Iterative Corresponding Geometry (ICG) algorithm. When hand occlusion occurs,the pose information of the hand is applied to the solution of part pose,significantly improving the accuracy of object pose estimation under hand occlusion. Experiments show that the Average Distance of Model points (ADD) index reaches 74.73 %,which is 2.61 times that of ICG. This algorithm significantly improves the accuracy and robustness of part pose calculation in augmented assembly scenarios.

The performance of the locality preserving projections algorithm mainly depends on the construction of the nearest neighbor map. The construction of the nearest neighbor map was easily affected by the interference of redundant information in the original data and the impact of not having a good basis for selecting the appropriate heat kernel parameters.As a result,the local structural information of high-dimensional data cannot be fully explored,and it was also easy to be more sensitive to noise and outliers in the low-dimensional embedding process.And its feature extraction ability in fault diagnosis applications was affected. To address the above problems,a Locality Preserving Projections algorithm based on Contractive Auto-Encoder and Manifold Ranking (CAE-MRLPP) was proposed and used for mechanical equipment fault diagnosis. Firstly,it combined the label information and Spearman correlation coefficient to pre-adjust the sample spacing. Secondly,the idea of manifold ranking was introduced to construct the weights based on the information of the sorting position of the sample points and the neighboring points in each other′s neighborhood sets and the information of the number of mutual neighbors of the two. Lastly,the contractive auto-encoder was fused with the locality preserving projections based on the manifold ranking,and the optimal projection matrix was solved by iterative optimization of the gradient descent method.Then the low-dimensional representation of the fault data was obtained. A number of verifications were carried out on the rolling bearing dataset and the pumping unit dataset,and the fault identification accuracy was more than 98 %. This indicated that the algorithm has good feature extraction ability,can effectively improve the fault identification accuracy,and has good robustness and generalization ability.

To solve the problems of large error,low efficiency and poor stability in the existing detection methods of roundness error of shaft workpieces,a sub-pixel detection and evaluation method of roundness error of shaft workpieces is proposed based on machine vision technology. This method uses polynomial interpolation-segmented curve fitting algorithm to accurately locate the sub-pixel edge of image contour.Based on obtaining the sub-pixel coordinates of image edge points,the least square method is used to detect the center coordinates,and then the roundness error of shaft workpieces is detected and evaluated. The experimental results show that the average absolute error between the measured value of this method and the measured value of the coordinate measuring machine is 8.75 μm,and the roundness error measured by this method is more uniform,the fluctuation is the smallest,and the detection accuracy is higher. It provides a feasible way for the evaluation of the roundness error of shaft workpieces.

Industrial defect image samples serve as fundamental data for industrial product defect detection,classification,and grading.To address the current challenges in industrial defect inspection,which include difficulties in target detection under complex environments and insufficient sample quantities resulting in poor feature extraction, a pre-trained autoencoder generative adversarial network was proposed.Pre-trained autoencoder was used to replace the generator network of the basic Generative Adversarial Network (GAN),facilitating better integration of data features by guiding the generator network.An encoder-decoder loss function was redesigned to replace the adversarial loss function of GAN by incorporating target image features.Experimental validation was conducted using a dataset of steel coil end-face defects.Experimental results indicate that after the improved GAN data augmentation,the mean Average Precision mAP_0.5 increased by a maximum of 0.118,while the precision for single-class defect detection increased by a maximum of 0.138.

The operational status of rotor systems is a critical factor in ensuring the stable operation of equipment.To evaluate the operational condition of the rotor system during equipment operation and to determine the nature and location of faults when rolling bearings fail,a rotor system status online monitoring system based on digital twin technology was proposed. Firstly,according to the operational characteristics of the rotor system,a digital twin framework for the rotor system was proposed. Then,the key technologies in the rotor system state monitoring method were analyzed. Finally,using a multifunctional rotor test bench as the digital twin entity,a rotor system status online monitoring system based on digital twin technology was built,verifying the feasibility of the rotor system digital twin framework,as well as the real-time performance and accuracy of the twin model. This provides a new approach for equipment fault prediction and health management.

To enhance the efficiency and accuracy of the lower rocker arm optimization in vertical roller mills,an advanced optimization approach based on the Improved Northern Goshawk Optimization (INGO) algorithm and the BP Neural Network (BPNN) model was proposed. Initially,a finite element analysis of the lower rocker arm was performed.Subsequently,to overcome the issues of local extrema and slow convergence with the Northern Goshawk Optimization (NGO) algorithm,improvements were introduced through the Latin hypercube sampling method,the sine cosine algorithm,and Cauchy mutation for NGO. Following this,INGO-BP prediction models were constructed to capture the relationships between different structural parameters and the mass,maximum equivalent stress,and maximum deformation of the lower rocker arm.Ultimately,a mathematical optimization model for the lower rocker arm was established and solved,yielding a set of optimal design variables. Experimental results revealed that the mass reduction in the optimized lower rocker arm reached 16.4 %,while the maximum equivalent stress and maximum deformation remained within safe limits.In comparison to other algorithms,the INGO has the advantages of fast convergence and strong optimization ability. The INGO-BP model demonstrated high accuracy and stability in predicting the mechanical properties of the lower rocker arm,providing reference for applying optimization algorithms in structural optimization.

To address the real-time issue of loss calculation in Permanent Magnet Synchronous Motor (PMSM) using the finite element method,a loss prediction model for PMSM based on BP neural network optimized by Dung Beetle Optimizer (DBO) algorithm was proposed. The study focuses on a 40 kW automotive PMSM. Firsty,the electromagnetic field loss solution model of the motor was established in the Finite Element Analysis (FEA) software Maxwell. Next,600 sets of control parameter combinations (armature current,internal power factor angle,speed) were selected for the motor loss solution through the optimal space-filling experimental design method to get the data set required for training the neural network. Finally,the DBO algorithm was utilized to optimize the BP neural network and a loss prediction model for PMSM based on the DBO-BP neural network was constructed. The predictive performance was compared with traditional BP neural networks model and BP neural network model optimized by genetic algorithms. The results indicate that the DBO-BP neural network prediction model surpasses the other two neural network models in prediction accuracy,with the prediction error controlled within 5.86 %,and the computation speed was 1 267 times faster than the finite element model. This effectively replaces the time-consuming finite element model,enhancing the real-time capability and accuracy of loss prediction,thus providing an effective method for motor loss prediction.

Sliding bearing is the key component of electronic water pump which directly affects the stability of the impeller rotor system.In order to establish a connection between the design and manufacturing factors with the vibration and noise performance of the pump assembly and reduce the vibration and noise effectively,finite element analysis and experimental research were conducted based on a low-power cantilever external rotor electronic water pump to explore the influence of sliding bearing parameters on the vibration and noise characteristics of the electronic water pump. The results showed that the magnitude of the water film stiffness of the sliding bearing affected the existence of the relative swing mode between the pump rotor and the core shaft. Compared with CF/PPS composite material bearing,graphite bearing electronic water pump effectively reduced the various harmonic components in the vibration spectrum and eliminated the frequency modulation sound of blades. Excessive eccentricity of the shaft neck could easily trigger the pump rotor-core shaft relative swing mode under low-speed conditions,leading to structural resonance and noise.Reducing the clearance of the bearing could greatly reduce the frequency vibration of the water pump blades under high-speed conditions,and suppress the structural noise of the rotor-core shaft relative swing mode.

Aiming at the problems of variable wind speed,nonlinearity and strong perturbation in direct-drive permanent magnet synchronous wind power generation system,a hybrid control method based on fuzzy control and Linear Active Disturbance Rejection Control (LADRC) is proposed. Firstly,the Linear Extended State Observer (LESO) in linear self-jamming is improved to enhance the ability of disturbance observation,while the fuzzy control is utilized to dynamically adjust the LADRC control parameters to enhance the anti-jamming ability and robustness of the system. Simulation results show that compared with the traditional PI controller,this method can better estimate the torque fluctuation caused by wind speed,accurately track the upper rated speed,and maximize the wind energy utilization.

The cavitation and fuel atomization characteristics of an ultra-high pressure fuel injector in a certain diesel engine were investigated through Computational Fluid Dynamics (CFD) fluid simulation. On the basis of verifying the diesel fluid grid,the effects of nozzle diameter,nozzle inlet roundness radius,and nozzle angle on the diesel gas phase flow field were analyzed,and the average diesel gas phase fraction and atomization effect at the nozzle outlet were studied. The results show that the diameter of the nozzle,the radius of the chamfer,and the angle all significantly affect the diesel atomization characteristics. As the diameter of the nozzle increases,cavitation phenomenon shifts from the bottom to the top,and the average diesel gas phase fraction first increases and then decreases;increasing the radius of rounding reduces cavitation intensity and average diesel gas phase fraction,but improves flow performance;the increase in angle causes cavitation to transfer from the lower wall to the upper wall,resulting in an increase in average diesel gas phase fraction and improved atomization effect. Through orthogonal simulation experiments,the optimized parameter combinations were obtained as nozzle diameter of 0.26 mm,rounding radius of 0.02 mm,and angle of 90°,the diesel flow rate at the nozzle outlet was 189.3 mL/s,and the average diesel gas phase fraction was 25.6 %. Finally,the accuracy of the simulation model was verified through experiments,and the optimized fuel injector has excellent atomization effect and flow characteristics.