The incremental hole drilling technique that combines moiré interferometry with the hole drilling method is recognized as an effective method for measuring internal residual stress.The bonding agent used in this technique plays a crucial role in transferring strain from the structure′s surface to the grating layer during the experimental process.However,the bonding agent′s influence is often not considered when solving for the calibration coefficient.Utilizing finite element software,the effects of the bonding agent on the calibration coefficient were investigated by considering five factors:material properties of the bonding agent,the shape and size of the bonding agent,the thickness of the bonding agent,the increment of hole drilling,and the strain measurement method.The impact of the bonding agent was demonstrated by comparing the calibration coefficient error with and without considering the bonding agent.Based on the results,it can be concluded that neglecting the bonding agent′s influence can significantly impact the calibration coefficient′s accuracy.Excluding the bonding agent from the finite element model will lead to underestimating the calibration coefficient.The error in the calibration coefficient exceeds 2.5 % and is highly dependent on the drilling increment and the measurement method.Furthermore,the effect of bonding agents on thin-walled components is more significant than on thicker samples.

With the increasing application of SiCp/Al, higher requirements are put forward for their processing efficiency and surface quality. In response to the above questions, based on the orthogonal test of L₂₅(5⁶), 20 % of SiCp/Al composites are processed using a new and efficient EDM processing method—ejecting-explosion EDM. The main optimization indicators are MRR, TWR and Ra, and the process parameters are high-voltage wave current, high-voltage wave voltage, low-voltage wave current, low-voltage wave voltage, pulse width and phase difference. Based on the orthogonal test results, the influence of each process parameter on different optimization indicators and the combination of single-objective optimal processing parameters are studied by using the Signal to Noise Ratio (SNR). Based on the Grey Relational Analysis (GRA), the weights of each optimization index are determined according to the Analytic Hierarchy Process (AHP), and the multi-objective optimal processing parameter combination of the three optimization indicators is comprehensively considered and verified by experiment. The results show that the optimal process parameter combination for multi-objective optimization is high-voltage wave current 6 A, high-voltage wave voltage 120 V, low-voltage wave current 2 A, low-voltage wave voltage 100 V, pulse width 10 μs, and phase difference 50 %, and processing by using this processing parameter combination can effectively improve the processing efficiency and surface quality.

Aiming at the problem of Automated Guided Vehicle (AGV) scheduling in flexible job shop,based on the sustainable perspective,considering the problem of workshop energy consumption,under the condition of the number of machines and AVG,a sustainable flexible job shop scheduling model is constructed with the optimization objectives of minimizing the maximum completion time,workshop energy consumption and the number of AGV used.Firstly,an Improved Whale Optimization Algorithm (IWOA) was designed,which introduced a nonlinear convergence factor and adaptive inertia weight on the basis of the standard whale optimization algorithm to improve the search ability and convergence speed of the algorithm.Secondly,the fuzzy membership theory was used to construct the loss function to obtain the optimal compromise solution of the multi-objective model. Finally,the performance of the algorithm was verified based on examples. The experimental results show that IWOA shows good effects in solving instances of different sizes,which provides an effective practical way for solving the sustainable flexible job shop optimization scheduling with AGV transportation.

In order to solve the step effect and deposition problems in the production of rotating thin-walled parts by additive manufacturing,improve the quality and accuracy of parts manufacturing,and improve the efficiency of additive manufacturing,a multi-axis robot additive manufacturing trajectory optimization based on layer-by-layer generation was proposed. Firstly,the direction vector was constructed based on the inverse geometric model of the six-axis robot,and the direction parameters were extracted from it to define the objective function and constraint function of different layers,and a layer-by-layer trajectory optimization strategy was proposed.Secondly,considering the axis redundancy generated by the coaxial deposition system,the axis redundancy was used to improve the motion trajectory of the robot,and the trajectory optimization was applied to the actual manufacturing of rotating thin-walled parts.Finally,to verify the effectiveness of the proposed method,point-by-point and non-optimized trajectory strategies were selected,and the hollow hemisphere,Laval nozzle and intake funnel configurations were selected for experimental verification.The experimental results show that the proposed method can provide a smoother joint motion trajectory,improve the manufacturing quality and manufacturing efficiency of parts,and compared with the non-optimized trajectory strategy,the proposed method has higher geometric accuracy,which can effectively improve additive manufacturing.

Order to improve the accuracy and efficiency of the prediction of the strengthening effect of Abrasive Water Jet (AWJ) strengthening process on the surface properties of 3D printed AlSi10Mg materials, firstly,the surface strengthening experiment of AlSi10Mg material strengthened by abrasive water jet was carried out.Then,based on the GA-GRNN neural network,the experimental data samples were trained with the surface hardness and surface residual stress as the target respectively,and the surface performance prediction model of 3D printed AlSi10Mg was established.Finally,the main parameters of AWJ strengthening in the established neural network model were optimized by genetic algorithm.The results show that the surface hardness and surface residual stress of AlSi10Mg material are effectively improved after abrasive water jet strengthening.The error of the established GA-GRNN prediction model is within 2.3 %,which has high accuracy.After optimization by genetic algorithm,the best parameter combination of surface hardness is obtained jet pressure 33 MPa,abrasive particle size 0.15 mm,target distance 12.4 mm,and the surface hardness is 159.25HV. The optimal parameter combination of surface residual stress is jet pressure 40 MPa,abrasive particle size 0.13 mm,target distance 15 mm,and the surface residual stress is -137.4 MPa. It provides data support for the parameter selection of the surface of the subsequent abrasive water jet strengthening parts.

In response to the challenges posed by complex technology and high cost in micro-surface printing circuit,a method for printing micro-curved circuit is proposed using three Degrees of Freedom (3DoF) printer.This method focuses on achieving conformal circuit on micro-surface with the utilization of a 3DoF printer.Additionally,a self-adaption 3D printing control algorithm is also developed.The algorithm takes DXF format files as input,from which it extracts the line data.Subsequently,the data undergoes discrete processing,and each data point is systematically measured.Based on the feedback data,the algorithm then reconstructs the circuit.The final output of the algorithm is in the form of G-code,directly suitable for 3DoF printer.Finally,circuit printing experiments on micro-curved surface prepared from different substrates and lighting experiments are conducted to confirm the feasibility of this approach.Furthermore,the process parameters are also investigated for 3DoF printing micro-curved circuit.By employing a design orthogonal test,the impact of various printing parameters is analyzed,leading to the identification of the optimal parameter combination.The results demonstrated that the best circuit quality is achieved when using a needle diameter of 0.16 mm,a printing air pressure of 0.3 MPa,a printing speed of 100 mm/min,and a printing height of 0.4 mm.Under these conditions,the printed circuit is continuous and uniform,showcasing its superior quality.

Aiming at the problems of single target,poor flexibility of information model and low universality in the collection of production data in traditional workshops during digital twin,a workshop production data collection method for digital twin is proposed.Firstly,by analyzing the production data acquisition and application requirements of digital twin workshop,a multi-level production data acquisition architecture is proposed. Secondly,focusing on the processing layer of the architecture,the key technologies of the construction of the digital representation model of production resources based on OPC Unified Architecture (OPC UA) are elaborated. Finally,based on the workshop application example,combined with the key technology and architecture,the digital twin system is developed,which verifies the feasibility of the acquisition method and effectively supports the operation of the digital twin system in the assembly workshop.

Aimming at the issue of control accuracy affected by malfunctions such as jamming, saturation and damage that are prone to occur in robot multi-joint manipulator, an optimal fault-tolerant control law was proposed using the designed fault observer. Firstly, a fault model of the robot multi-joint manipulator was established based on the relationship between the spatial position of the end of the robot multi-joint manipulator and the rotation angle of each joint. Then, the tracking errors of the multi-joint manipulator rotation angle and contact force were defined, and the fault observer was designed. The model parameters were estimated by the introduced RBF neural network. Finally, performance index matrix of robot multi-joint manipulator with the tracking error of position/contact force and faults was established according to the idea of dynamic programming, and the fault-tolerant control law including Hamiltonian equation was designed. The RBF neural network was used to solve the optimal performance index, so as to obtain the optimal fault-tolerant control law. The simulation results show that the designed optimal fault-tolerant control law can overcome the influence of various faults, the maximum error of fault estimation is only 0.09 N·m, the maximum error of position tracking is only 0.14 cm, and the maximum error of contact force control is only 0.18 N, which verifies the feasibility of the designed method. The measured results of the six degrees of freedom multi-joint manipulator show that the maximum error of positioning on 8 spatial coordinates is only 0.17 cm, and the maximum error of the contact force is only 0.22 N, which verifies that the designed optimal fault-tolerant control law has superior engineering practicability.

Multi-sensor data fusion technology can better solve the problem of data incompatibility generated by smart devices,improve device operational efficiency and productivity. Existing methods lack high-performance performance and do not consider data privacy protection issues. It innovatively introduced federated learning into multi-sensor data fusion. The federated learning local model used the Gated Recurrent Unit (GRU) algorithm to solve the multi-sensor data fitting problem. A novel parallel stereoscopic multi-sensor data fusion method was designed for the first time,which has excellent fusion performance and ensures the privacy of each client′s data. The experimental results demonstrate the correctness and rationality of this method,as well as its advantages in robustness.

Aiming at the problems of traditional ant colony algorithm in global path planning of indoor mobile robot,such as low search efficiency,unsmooth path,easy to fall into local optimum and deadlock,an ant colony optimization algorithm for bi-directional search with improved A^* algorithm was designed. Firstly,the improved A^* algorithm was used to quickly converge and obtain the initial path in the grid environment,the initial pheromone matrix was constructed,and the obstacle factor was introduced to reduce the occurrence of ant deadlock. Secondly,the rules of ant colony optimization algorithm for bi-directional search were set,the heuristic function model in bi-directional search was improved,and elite ant search strategy and adaptive pheromone volatilization factor strategy were introduced. Finally,the third-order Bezier curve was used to smooth the path. The simulation results on Pycharm platform show that this algorithm combines the strong global search ability of A^* algorithm and the positive feedback characteristics of ant colony algorithm,which makes the improved algorithm optimize the path length by 12.85 % and 7.76 %,the search time by 38.17 % and 23.46 %,and the iteration times by 67.71 % and 54.41 % compared with the traditional ant colony algorithm and the sparrow search algorithm,and the global path optimization effect is obvious.

For the configuration of dual-mode power split type hybrid electric vehicle transmission system,first the structural characteristics and application range of single and dual mode hybrid electric systems are compared and analyzed. Secondly,based on the characteristics of composite allocation,three types of composite allocation four node rod models are selected,and all feasible composite allocation configurations are obtained through reverse splitting.The configuration scheme is determined based on the constraint conditions of rod length and characteristic parameters.On this basis,referring to the parameters of a certain vehicle model,by analyzing the input distribution mode,mode switching point,and mechanical point characteristics,the position and structural parameters of the third row planetary gear and clutch are determined.The final design of the transmission system configuration provides a theoretical basis for studying the dual-mode power split hybrid power vehicle transmission system.

Under actual working conditions,the shaft system of micro gas turbines is often affected by coupling loads due to various factors.The closed interference connected diaphragm coupling is a key flexible component for connecting double span rotors on gas turbines,and its coupling stiffness has become an important parameter affecting the static and dynamic performance of the system. Currently,there are many studies on the single load stiffness of diaphragm couplings,and the coupling stiffness characteristics are unclear,which is not conducive to the improvement of their design technology and shaft system stability.Based on this problem,by simplifying the solution model of the diaphragm coupling,the overall torsional stiffness characteristics of the coupling under the combined action of axial and torsional loads were studied through simulation analysis,and comparing the torsional stiffness characteristics under single torsional load.The results showed that under the action of axial torsional coupling,the overall torsional stiffness of the diaphragm coupling exhibits nonlinear characteristics. When the ratio of torsional load to axial load is less than 0.25,the axial torsional coupling effect is obvious,and the torsional stiffness of the diaphragm coupling decreases. When the ratio of torsional load to axial load is greater than or egual to 0.25,the coupling effect can be ignored and can be considered as a pure torsional state. The characteristic lays the foundation for further in-depth research on the influence of diaphragm coupling on the dynamic performance of the shaft system and its optimization design.

The response relationship between process parameters and machining quality of inner raceway precision grinding of angular contact ball bearings was explored, and the optimal process parameter set to improve raceway machining quality was identitying. Response Surface Methodology (RSM) and Multi-Objective Particle Swarm Optimization (MOPSO) algorithm were used to optimize the grinding depth, grinding wheel speed, and workpiece speed that affect raceway machining quality. First, the RSM was used to establish a significant non-degenerate model with the surface roughness and roundness error as responses. Then, variance analysis and surface plots were used to study the interactive effects of process parameters on the responses. Finally, the MOPSO algorithm was applied for optimizing the model, and K-means clustering method was used to solve the compromised solution of the optimal solution set, and experimental validation was conducted. The results showed that grinding depth and grinding wheel linear speed had highly significant effect on the surface roughness and roundness error of the raceways, and workpiece speed had highly significant effect on the roundness error and a significant effect on the surface roughness. The interaction between grinding depth and workpiece speed had a significant effect on the surface roughness, and the interaction of grinding wheel linear speed with workpiece speed and grinding depth had a significant effect on the roundness error. The optimized parameter set was verified by experiments, and the surface roughness and roundness error were reduced by 8.14 % and 16.03 %, respectively, compared to preoptimization. The regression model based on RSM and MOPSO algorithm is significant for the whole and individual variables, and has high prediction accuracy, and the optimized process parameter set can obtain good optimization results.

Due to the problems of long research and design cycle of stacker,and low rate of information collection and working efficiency,a visual design method of stacker model based on digital twin was proposed. Combined with the technology of digital twin,a virtual debugging platform of stacker model was built by OPC unified architecture technology based on the five-dimensional model of digital twin,which realizes real-time communication and data interaction between the physical model and the digital model of stacker.The visual monitoring and optimization of the working state of the stacker were realized through the data collection of the working state of the stacker physical model,which verifies the feasibility of the visual design method of stacker model based on digital twin.This study can provide a reference for the design and transparency operation of stacker.

An improved YOLOv7 object detection algorithm was proposed to address the issues of complex texture and small defects on the end face of steel coils,as well as slow recognition speed and low detection rate of small targets using YOLOv7 algorithm. The ELAN structure in the YOLOv7 algorithm backbone network was improved by adding a PConv convolutional layer to design a new structure to reduce the model complexity and improve the model detection speed. Due to the tendency to miss detection in small object detection,a new attention mechanism CSCA was designed to improve the network′s sensitivity to small-scale objects. On this basis,WIoUv2 loss function was used to replace the CIoU loss function in the original YOLOv7 algorithm network to optimize the loss function and improve the robustness of the network.Experiments were conducted on a self-made dataset of steel coil end face defects,and the results showed that the improved YOLOv7 algorithm improved mAP@0.5 by 4.1 % and FPS by 10.84 f/s,with better detection performance than the original algorithm.

In order to solve the problem of low detection accuracy and long time when robot arms grasp unknown objects in dense scenes,a new grasping network structure called Cascade Fusion Grasp Detection (CFGD) was proposed,which was used to predict the grasping pose of objects in dense scenes. Based on the proposed backbone and several blocks for extracting initial high-resolution features,several cascade stages were also introduced to generate multi-scale features in CFGD network. A sub-backbone for feature extraction and an extremely lightweight transition block for feature integration were included at each stage.Such a design allowed for a larger proportion of parameters throughout the backbone and a deeper and more efficient feature fusion. Compared with the existing algorithms,the proposed algorithm has significantly improved the detection accuracy and speed on Cornell grasp data set,Jacquard grasp data set and custom data set. In the real grasping scene,the grasping success rate is 98.6 % in the single target scene and 94.6 % in the dense scene. Experimental results showed that the proposed algorithm can predict and grasp unknown objects in dense scenes with high accuracy.

A monitoring system for predicting the force and stability of a hydraulic gantry crane has been proposed.A finite element model of the crane structure lifting unit was established based on actual parameters to analyze the stress distribution of the crane′s relevant structures under various operating conditions.The data obtained from the finite element analysis was used to train a BP neural network prediction model,which was then combined with measurement data to quickly assess the stress distribution of the crane′s relevant structures in any given state.The feasibility of the system was verified through experiments,and the safety monitoring capability of the hydraulic gantry crane was improved.

In order to extract the bearing faults characteristic frequency under strong noise background,a vari-scale stochastic resonance system based useful signal enhancement method was designed. The structure and fault characteristics of bearing were introduced. The shortcoming of stochastic resonance theory to low noise and low frequency was analyzed,and a vari-scale stochastic resonance system was designed specifically,expanding the application scope of the theory. Then a multi-behavior particle swarm optimization algorithm for stochastic resonance system parameters optimization was proposed. Verified by simulation,under the strong noise background with a signal-to-noise ratio of -20 dB,the variable scale stochastic resonance system can still effectively extract feature frequencies from useful signals. According to the experimental data of bearings published by Xi′an Jiaotong University,under strong noise background,the fault frequency extracted by sparse reconstruction method is still submerged in the nearby frequency domain,while the feature frequency extracted by the variable scale stochastic resonance system is significantly prominent,and the feature frequency extracted by the method is closer to the true value. The experimental results indicate that the vari-scale stochastic resonance theory can effectively extract the characteristic frequencies in vibration signals under strong noise backgrounds.

The constant torque control of electric impact wrench was studied.Firstly,the collision model and torque accumulation model of electric impact wrench were established,and the factors affecting the torque output of wrench were analyzed. A constant torque control scheme based on impact times and impact block speed was proposed. At the same time,the collection scheme of impact times and the speed control scheme of double closed-loop DC brushless motor based on fuzzy PID were designed.Verified on the torque test platform of electric impact wrench,the results show that the torque control accuracy of the wrench can reach within ±4 % when the impact times are 200 times. The constant torque control method has the characteristics of high control accuracy,low cost and convenient operation. The feasibility of this method was verified by practical application.