A novel method of the electric-field-driven (EFD) jet deposition micro-scale 3 D printing for fabricating high resolution circuit on paper-based surface was proposed.The basic principles and process flows of the proposed method were presented.The effects and rules of the main processing parameters (e.g.,the voltage,back pressure,printing speed) on the accuracy and quality of silver wires fabrications were revealed by a serial of experiments.Then,by using optimized parameters and the nano silver paste (75% silver content and 35 Pa·s (at 25℃) dynamic viscosity),the micro/nano-scale structure with high resolution and high aspect ratio was fabricated by multi-layer stacking on three kinds of paper substrates.For example,the line width with 15 layers stacking on RC photo paper may be kept at 10 μm,and the aspect ratio may be increased to 6.33,causing the 94.8% decrease of the resistance.Finally,flexible electromagnetic actuators (FEMAs) and complex conductive patterns were fabricated on different paper substrates to prove the printing ability.The results show that the new method of EFD 3 D printing technology combined with high viscosity and low temperature sintering nano silver paste may provide an effective way for high-performance paper-based electronics.

This paper reviewed the development routes of developed countries such as the United States， Japan， and Germany since the emergence of industrial engineering for more than 100 years， discussed the regular characteristics of enterprise management innovation， and drew the conclusion that smart manufacturing also needed IE/LM to provide management support. According to the smart manufacturing project cycle， the functions of IE/LM in smart manufacturing engineering were analyzed from the perspectives of basic preparation， scheme selection， and integration development. The framework of lean smart management system for smart manufacturing transformation of Chinese enterprises was given， and the improvement ideas and methods of smart-lean integration were explained based on a case. Finally， some key technologies of smart manufacturing management facing transformation and upgrading of Chinese enterprises were presented based on the actual needs of China. 

The theoretical system of intelligent manufacturing for HCPS confirmed the central position of human in the intelligent manufacturing system. Starting from the demand of human-machine collaboration in the intelligent manufacturing system, the emphases of human factors in HCIM were discussed from three levels such as behavior, intention, and cognition, based on the theory of gulf. Focusing on virtual-real fusion scenarios, multimodal human-machine interaction, cognitive quantification and other methods, the importance of human factor engineering in promoting the integration of human-computer intelligence was expounded. Finally, research direction and development suggestions of human-centered intelligent manufacturing from the implementation of HCPS intelligent manufacturing systems were put forward.

The welded joints were susceptible to defects and stress concentration, rendering them vulnerable areas for fatigue crack initiation and propagation under fatigue loads. In comparison to homogeneous materials, the microstructure and stress localization in each of regions for the joints further complicated the fatigue issue in welded structures. Unlike ideal experimental conditions, the actual service environments of welded structures were intricate, it was necessity to consider the coupling characteristics between environmental factors and welded structures when predicting welded structure fatigue life. Therefore, the internal factors influencing welded structures were summarized and analyzed while reviewing existing life prediction models from perspectives encompassing complex loads and extreme service environment. Combining the latest research progresses, the recommendations were proposed to enhance fatigue life assessment methods for the welded structures.

 The reliability analysis results of complex structures based on the Kriging model were highly dependent on the fitting accuracy of the Kriging model. In the constructing processes of the Kriging model, the selection of different correlation and regression functions affected the accuracy of the model. In order to solve the impacts of model uncertainty on the reliability analysis results, while considering computational efficiency and accuracy, based on the Kriging model and Monte Carlo simulation(MCS) method, a structural reliability calculation method combining adaptive ensemble strategy and active learning function was proposed. Considering the modeling uncertainty of Kriging models, combined with multiple Kriging models, this methed constructed an active learning function that comprehensively considered sample point contribution and sample point distance. The ensemble Kriging model was iteratively updated through the active learning function until the convergence conditions were satisfied. Finally the structural reliability analysis was carried out by the constructed ensemble Kriging model and MCS method. The validity of the proposed method was verified by numerical and engineering examples, and the results show that the proposed method is more robust than other major methods, and the computational efficiency is higher while ensuring the computational accuracy.

Aiming at the shortages of effective historical data in actual manufaction， a carbon emission prediction and multi-objective optimization model driven by small sample data was proposed. The Box-Behnken experimental design was used to collect processing data， and then the back propagation neural network was used to establish a prediction model for carbon emissions and processing efficiency， which ensured the prediction accuracy with less historical sample data. Aiming to optimize the total carbon consumption and makespan， the improved gray wolf optimization algorithm and entropy-TOPSIS comprehensive evaluation were used for determining the optimal processing parameters. Finally， the effectiveness of the proposed method was verified by machining experiments. 

 To solve the low-carbon scheduling problems of multi-objective flexible job-shops taking equipment energy consumption， tool wear and cutting fluid consumption as carbon emission sources and energy consumption and labor cost as processing cost， a low-carbon scheduling model was formulated to minimize carbon emission， makespan and processing cost， and an improved elitist NSGA-Ⅱ was proposed to solve the problem. Firstly， the chromosome composition was dynamically adjusted by encoding based on Tent chaotic map and greedy decoding based on analytic hierarchy process to improve the quality of the initial population. Then， an adaptive genetic strategy was proposed based on genetic parameters， which adjusted the crossover and mutation rates according to the population evolution stage and the population non dominated state dynamically. Finally， based on external archives an improved elite retention strategy was designed to improve the population diversity in the later stages of the algorithm and retain high-quality individuals in the evolution processes. The effectiveness of the improved algorithm was verified by standard scheduling examples and a practical case. 

In order to solve the problems of product consistency and low production efficiency caused by manually reinforced carbon/carbon crucible preforms， a flexible winding forming system of carbon fiber reinforced crucible preform was proposed. The line type design process of crucible core die for special rotary body structure with one end plane head and one end ellipsoid head was presented based on the non-geodesic method. Each module and implementation method of computer aided line type design were introduced， and the design line type was simulated. A special winding machine with four degrees of freedom was designed according to the winding characteristics. The control system of four-axis winding machine was designed based on programmable controller and touch screen， and the winding tests were carried out. It is indicated that the winding machine runs stably and the fiber may be wound continuously and stably on the surfaces of the core die， and the system may realize automatic winding of carbon fiber reinforced carbon/carbon crucible preforms. 

FISH technology was a common method to detect the nucleic acid sequence of malignant tumor cells. It was widely used in the field of cancer diagnoses and treatments. In order to realize the full automation of FISH experiments, a full-automatic pathological staining system was designed, which was provided with multi axis control arm, reagent sampler, slide, cover slide fixture, auxiliary module and so on. The positioning erorrs of the multi axis control arm are less than ±0.1 mm, the accuracy of extracting and discarding the pipette head is more than 99.5%, the success rate of extracting the cover glass is more than 99%, and the erorrs of micro reagent sampling are less than 0.6 μL. The errors of adding a large amount of reagent are less than 0.5 mL, the temperature control erorrs of slide clamp are less than ±1 ℃, and there is no reagent leakage. The results of biological experiments show that the system has good fluorescence staining effectiveness, the signal points are clearly visible, and the interpretable rate of fluorescence image is more than 90%.

It was rather difficult to measure the centroid side slip angle and the tire lateral forces of the four-wheel-drive electric vehicles directly. Considering the unmodeled dynamic characteristics of the systems, model parameter perturbations, system process noises and measurement noises, a joint estimation method was proposed based on FFRLS and RCKF. Based on applying FFRLS to estimate the mass of the vehicle in real time and the estimation error minimization in the background of the maximum value embedded into the standard cubature Kalman filter to realize RCKF. The improved strategy of the joint estimation algorithm was proposed, which effectively improved the anti-interference ability of the filter to the model parameter perturbations and the unmodeled noises under composite conditions. It may  realize the accurate estimation of the centroid sideslip angle and tire lateral forces. By CarSim/Simulink, the accuracy, robustness and anti-interference of the algorithm were verified in different conditions. Through the actual vehicle platform of four-wheel-drive electric vehicle, the validity of the algorithm was verified. Research show that the results of the proposed method are more accurate than that of RCKF and the standard Cubature Kalman filter. The problems of joint estimation of centroid side slip angle and tire lateral forces of four-wheel-drive electric vehicle are solved under composite conditions.

Based on the four dimensions of thermal-pressure parameters,lubricant oil conditions,vibrations and cylinder pressures,data acquisition and feature extraction were carried out,and a method for diesel engine condition monitoring was proposed based on OCSVM anomaly detection algorithm and Fisher discriminant analysis.The CUSUMMR was used for parameter trend detection and the D-S evidence theory and weight of evidence method were used for multi-source information fusion.the RUL (remaining useful life) of diesel engine lubricating oil was predicted by LSTM.The Paper solves the problems such as low early warning rate,poor adaptability of the model under diesel engine dynamic conditions.

To solve the problem that it was difficult to obtain a large number of high-quality rolling bearing fault data in the actual industrial environment， and the generalization performance of the intelligent diagnosis model was poor， a fault diagnosis method was proposed based on simulation data driven and domain adaptation. Firstly， a physical model was established to obtain rich simulation data， which simulated different failure forms of bearings according to bearing parameters and corresponding operating conditions. Secondly， the transfer learning method was used to solve the problem of inconsistent data feature distributions between simulation and actual fault data. The residual channel attention mechanism network was used to extract the transfer fault features of different domains， and the adaptive operation of different domains in the network training processes was carried out through the condition maximum mean discrepancy metric criterion， which taken into account the conditional distribution discrepancies between different domains. Finally， different transfer model tests were carried out on the bearing data sets damaged by man-made damage and accelerated life test. The results show that the method proposed may obtain better recognition accuracy when the target domain contains a small number of labels. 

At present， there was an urgent need for an accelerated life test method of harmonic reducer in engineering to replace the full life test in order to reduce the test cycles and costs. Based on the principles of accelerated life test， a constant stress accelerated life test program  was proposed based on the failure characteristics of the harmonic reducer. The flexible wheel that was most likely to fail in the harmonic reducers was taken as the object， and the definite failure judgment， acceleration stress， acceleration factor， method of accelerating model were pointed out. In the small sample test data processing， a more accurate Weibull distribution parameter estimation was obtained by using the maximum likelihood method and Markov Monte Carlo method. The experimental results of the prototype show that the scheme may effectively describe the life of the harmonic reducers. The reliability evaluation standard of the basic rated life is 62% higher than that of the evaluation standard of the median life. The life index of the harmonic reducers was effectively evaluated. The research results have certain engineering application values. 

 For the current problems for complex modeling and long development cycle of virtual entities such as industrial robots in digital twin workshop construction， a modular construction method of virtual-real drive systems for industrial robots in digital twin workshops was proposed， which divided the virtual-real drive systems into an interaction layer for setting model parameters and a control layer for designing configurations according to functional requirements， and then abstracted the physical industrial robots， etc. into a simulation model from coupling single functional atomic model. The modular and hierarchical approach to building virtual-reality drive systems may quickly and effectively realize the modeling of digital twin virtual entities such as industrial robots, as well as the simulation of industrial robots operating in virtual space and the simultaneous operation of virtual-reality.

Unbalancing training dataset caused by the difficulty in obtaining fault samples seriously affectsed the robust and accuracy of fault diagnosis model. A data generation model was proposed based on self-adaptive auxiliary classifier GAN， which adaptively adjusted the generator loss by measuring the relative performance between discriminator and generator， accelerated the converge speed of training processes， and improved the quality of generated data. The raw data， data generated by auxiliary classifier GAN method， and data generated by proposed method were used as the input data of the BP neural network. The results show that the model trained by data of the proposed method was superior. Comparison results of the proposed method with 1D-CNN，e2e-LSTM，CFVS-SVM， and FFT-CNN fault diagnosis methods manifest that the proposed method is better in fault diagnosis accuracy and data processing time

In order to explore the changes and distributions of the temperature of PDMS sample being impacted and cooled by the liquid nitrogen jet during the cryogenic micro-abrasive air jet machining the PDMS microchannels, this paper proposed a systematic research method of temperature that composed of temperature measurement experiments, Becks inverse estimation method fitting and APDL temperature field simulation. The errors between the final simulation results and the temperature measurement ones are only 1.735%. The simulations find that it will take 5.747 s for the surfaces to reach the obvious embrittlement temperature -147 ℃. PDMS may not reach a machinable embrittlement state at the moment of contact with liquid nitrogen， so it is necessary to pay attention to the pre-cooling time before machining. The simulation data was used to fit the cooling rate of PDMS to -147 ℃, and the cooling rate was used as a reference to predict the pre-cooling time required in the actual machining. By comparing the cooling rate, it is found that the cooling rate in the depth range of 0~100 μm from the surface of the PDMS specimen is 22.054% faster than that in the depth range of 0~1500 μm, and the cooling rate of the aluminum alloy workbench is about 22.311% higher than that of the adiabatic workbench.

In the intelligent photovoltaic tracker brackets， cold-formed purlins were used to support the photovoltaic panels， and located spannig the horizontal single-axis and the module frame. Firstly， the minimum compliance of the structures was taken as the target and relative densities of elements were taken as the design variables， and the topology optimum design models were constructed under the given volume and the first natural frequency constraints. Optimal material distributions of the purlins were obtained based on SIMP (solid isotropic material with penalization) method， and this topology optimization structure was engineering designed and manufactured. Then， test load conditions were designed according to the practical environment where the photovoltaic tracker brackets were applied under different wind loads. The static and dynamic finite element analyses of the original and optimized purlins were carried out respectively， the simulation results show that the optimized purlins are improved in terms of bending resistance， torsion resistance， and natural frequency. Thus， the effectiveness of the optimization design method is verified. After that， the optimal purlins whose mass is reduced by 8.8% were also manufactured by engineering methods， and the mechanics performances were verified by the experiments. 

 Aiming at the problems in flexible job shop scheduling， an improved multi-objective migrating bird optimization（MOMBO）algorithm was proposed to solve the high-dimensional multi-objective scheduling problem with the consideration of maximum completion time， total delay period， total load of machine， and total energy consumption. On the basis of migrating bird optimization algorithm， MOMBO algorithm introduced a selection operator based on Pareto domination and reference point to give bird population selection pressure， and the combined weight method based on attribute hierarchical mode and gray relation analysis was used to select the most suitable solution from the optimal solution sets. The effectiveness and practicability of MOMBO were verified by test instances and case study. 

Facing the magnetic rail relationship of high temperature superconducting levitation train, the suspension force characteristics between dual-peak permanent magnetic orbit and high temperature superconducting block combination were tested at different field cold heights, and it is found that the magnitudes of suspension forces first increase then decrease as the distance increases, and existe the obvious hysteresis characteristics. In order to optimize the parameters of high temperature superconducting block combination and permanent magnetic track pair, the equivalent processing method for high temperature superconducting block was proposed based on image dipole model. The simulation model was established in order to obtain the trapped magnetic fields. The attractive force and repulsive force of the combination of high temperature superconductor block combination were calculated by way of trapped magnetic field and low relative permeability, and the relationship between suspension force and suspension gap at different field cold heights was gained. The results show that the simulated results are in good agreement with experimental ones at position of near different field cold heights, which verifies the equivalent processing method. The aim of paper is to provide a simple and effective method for the suspension force calculation of high temperature superconducting flux pinned magnetic levitation trains.

 Sediment penetration test was an important in-situ testing technique for deep-sea sediment strength. In view of the extreme environmental characteristics of the 7000 meter deep sea and the scientific demands for fine in-situ measurement， a deep-sea non-contact strong magnetic sediment penetration strength in-situ measurement device was developed based on the Jiaolong manned submersible， which broke through the problem that the original penetration measurement techniques in the 7000 meter deep sea structure was complex and could not achieve in-situ and fine measurement， A new method of solving deep-sea pressure resistance and strain gauge deformation conduction was proposed based on the non-contact principle of permanent magnetic force balance conduction of manned submersible. The prototype development and deep-sea sea tests were completed， and good test results were obtained. The whole system adopted self-contained working mode to collect and store data， and might be connected with the computer to realize in-situ data reading. The operating water depth is as 7000 m， the maximum penetration depth is as 250 mm， the measuring range is as 0～100 kPa， and the accuracy may reach 5%~10% FS(full scale). In March 2021， the Jiaolong manned submersible completed 5200 m and 6650 m sea trials in 181 and 185 dives in the Parihivila basin， and successfully obtained geotechnical data. 

A dot matrix offshore wind turbine integrated foundation design scheme was established based on the concept of wind turbine integration. The overall structural parameters and mechanics model were analyzed， and the feasibility and advantages of the overall scheme of the new dot matrix foundation（DMF） were established. Combined with the traditional OC4 semi-submersible wind turbine system， the hydrodynamic simulation was carried out under wind wave current loads. Results show that the stability of DMF in pitch motion is 70% higher than that of the traditional OC4 system. The small-scale prototype processing of DMF and the simulation experiments of wind and wave trough were completed based on similarity theory. Experimental results are in good agreement with the simulation data. 

 Aiming at the problems that the standard unscented Kalman filter（UKF） localization algorithm could not meet the high-precision localization requirements of mobile robots when moved on uneven ground， an ARUKF localization algorithm was proposed based on robust estimation theory. The ARUKF adaptively adjusted the predicted value of UKF according to the dynamic residual， reduced the influences of external interference on the predicted values of the systems， improved the accuracy and robustness of the system， speeded up the operation by reducing the computational complexity of the sampling processes， and improved the real-time performance of the system. The simulation and field test results show that the ARUKF algorithm may converge faster for the disturbance generated by uneven ground， and have better accuracy， robustness， and real-time performance， compared with the UKF algorithm and the improved UKF algorithm based on Sage-Husa. The average distance error is less than 2 mm， and the average angle error is less than 0.016 rad， which may meet more stringent requirements of the construction site. 

Improving the surface integrity of bearing grooves was an important solution to solve technical problems such as low reliability life, high friction power consumption and unstable vibrations and noises of bearings. On the basis of comprehensive research of generalized surface integrity by domestic and foreign scholars, the definition and connotation of characteristic parameters of bearing groove surface integrity were proposed, and the significance of characteristic parameters and the influence mechanism on bearing performance were expounded. The influences of different processing processes such as forging, turning, heat treatment, grinding and ultra-precision on the surface characteristic parameters of bearing raceways were analyzed in details, and the corresponding control measures were given. The new technologies, new processes, and new methods for improving surface integrity were briefly introduced. Finally, the summary and prospect of improving the surface integrity of bearing grooves were made to provide some references for improving the contact fatigue lifes of bearings.

Inland waterway navigation was an important part of the modern comprehensive transportation systems.The real-time and high-precision ship trajectory prediction method was helpful to effectively avoid water traffic accidents and enhance the ability of automation and intelligent supervision.Aiming at the problems that the accuracy of the existing inland ship trajectory prediction was not high,in order to improve the short-term prediction accuracy of ship trajectory,comprehensively using the recent AIS (automatic identification system) data and historical AIS data of the ships,and based on the relationship among trajectory and speed,course,and the characteristics of inland waterway,the temporal convolutional network model for speed and course prediction,ship trajectory dynamics equation model and adaptive double-hidden layer RBF network were constructed.The ship trajectory prediction method based on multi-model fusion was proposed.Experimental results show that the proposed method has obvious improvement in trajectory prediction accuracy and may meet the real-time requirements.

In response to the unmanned inspection requirements of mining wire ropes, a rope-twisting climbing inspection robot was designed and developed. Compared to traditional axial climbing robots, which required approximately 91.5% of the driving force. When carrying a load of 3 kg, the robot may overcome obstacles with a height 0.6 mm higher than that of axial climbing robots. With an obstacle height of 3 mm, the maximum load capacity exceeds that of axial climbing robots by 0.4 kg. Climbing experiments were conducted under simulated deep mine conditions with wire rope vibrations. The results show that the climbing robots exhibite stable climbing performance when the wire ropes are stationary, achieving a maximum climbing speed of 8.25 m/min and capable of continuous climbing for 500 m. Under low-frequency large-amplitude vibration conditions, the climbing speed of the robot is higher than that when stationary, while under high-frequency small-amplitude vibration conditions, slight fluctuations in climbing speed are observed due to wire rope vibrations.

The pace of large wind turbine units was accelerating, and the reliability of core components was increasingly important for wind turbine operations. Sliding bearings had the advantages of high load capacity, long life, easy maintenance, scalability and small size, and they had advantages and great potential for the reliable replacement of wind turbine main bearings key components produced at home. The problems of main shaft rolling bearings in high-power wind turbines and the advantages of using sliding bearings on the main shaft were analyzed herein. The technical methods and application status of wind turbine main shaft sliding bearing design, materials, lubrication, and experimental verification were present in detail, and the existing problems of high-power wind turbine main shaft sliding bearings and future development trends were summarized. It is expected to provide reference for the digital design and industrial development of high-power wind turbine main bearings.

In order to reduce equipment failure rates and downtime loss， a method for equipments multi-condition monitoring and fault diagnosis was proposed herein based on cointegration analysis. Based on the data collected by the supervisory control and data acquisition system， the random forest feature selection algorithm was used to extract the key feature variables related to equipments failure. By cointegration analysis of the key feature series， the cointegration coefficient was calculated and the cointegration model was established to obtain the optimal residual series， which might reflect the changes of equipments status. Probability plot was applied to analyze the optimal residual series and obtain the interval division of multiple operating situations. The residual warning thresholds corresponding to each operating conditions were determined so as to achieve state monitoring and fault warning. The research results of a certain type of direct-drive wind turbine show that the proposed method may analyze non-stationary time series effectively， monitor the faults of motors and identify the operating situations of the wind turbines by residual threshold， improve the accuracy of fault diagnosis. 

Aiming at the large deviations between the actual paths and the predetermined ones of underwater gliders affected by ocean current, a neural network ocean current prediction model with long-term and short-term memory and attention mechanism was established based on the traditional long-term and short-term memory network model.The dynamic Q-table of underwater glider motions was generated by depth neural network, and the optimal motion attitude was selected by reinforcement learning algorithm. Considering the influences of ocean current, an underwater glider path tracking algorithm was constructed based on depth reinforcement learning. The results show that the long-term and short-term memory network based on attention mechanism has less mean square errors and root mean square errors in ocean current prediction than that of the traditional integrated moving average autoregressive model and long-term and short-term memory network.Compared with the traditional PID control, the deep reinforcement learning model may reduce the root mean square errors of the underwater glider trajectory by 50.9%, and significantly improve the path tracking accuracy.

Due to influences of noise and complex transmission paths， the wind turbine gear fault signatures were generally weak. To effectively diagnose the gear faults， a new blind deconvolution method  was proposed based on reweighted-kurtosis maximization. The reweighted-kurtosis possessed great robustness to single or few strong impulse interferences and did not require any prior knowledge of the fault impulse train to be restored. The proposed deconvolution method may effectively solve the problems that the classical kurtosis maximization-based methods tend to restore a single dominant impulse rather than the gear fault impulse train. At the same time， the proposed method has stronger applicability in gear fault diagnosis for industrial equipment in comparison with the common non-fully “blind” methods（relying on the prior knowledge of the fault characteristic frequency）. The analysis results of the simulated signals show that the proposed method is effective in restoring fault impulse trains. The applications in wind turbine fault diagnosis demonstrate the effectiveness of the proposed method for gear fault diagnosis. 

Aiming at distributed flexible assembly job-shop environments, the machine selection flexibility, worker scheduling flexibility, and process sequence flexibility in scheduling were comprehensively considered, and a model of DMFAJSP was constructed taking minimizing the makespan and minimizing the total energy consumption as optimization objectives. To solve the DMFAJSP model, MDMA was proposed using estimation of distribution algorithm as the global search component and the neighborhood search operators as the local search component. Finally, comparative experiments were conducted between the proposed algorithm and other algorithms, and the results show that the MDMA has significant advantages in solving DMFAJSP model.

 The examines automatic fiber placement on point cloud surfaces was researched. The direct projection method with moving least-squares projection technique was proposed to project the initial path. A point cloud slicing method was improved to propose a point cloud projection slicing method. A cubic B-spline interpolation algorithm was used to generate initial paths by fitting projection points， and an equidistant offset algorithm for chord length subdivision projections was proposed. Based on stretching the path splines end， a boundary processing method was proposed to solve the problem that the offsetting paths could not reach the surface boundary. Finally， the algorithm was visualized by reverse engineering module of CATIA， and the systematic relationship between the path planning algorithm and the 3D modelling software was established. 

In order to solve the problems of large minimum turning radius and inadequate steering maneuverability of Ackermann steering-based off-road vehicles， a road adaptive pivot steering control strategy was developed by taking advantages of the independent control of vehicle torque driven by in-wheel motors. A seven-degree-of-freedom pivot steering dynamics model was constructed to explain the evolution of the longitudinal and transverse coupled motion tire forces during pivot steering， and a quantitative model was established to quantify the pivot steering resistance moment and transverse sway moment with wheel slip rate and road adhesion coefficient. The desired trajectory of transverse sway angular velocity under different adhesion conditions was designed with steering power responsiveness as the optimization objective， and the safety threshold of each wheel slip rate was used as the stability constraint to reduce the steering center offset. The executive layer tracked the transverse angular velocity based on the model prediction algorithm， while the adaptive sliding mode controller was introduced to adjust the wheel slip rate to ensure the stability of the longitudinal and transverse motions. Simulation tests and real vehicle tests show that the developed pivot steering control strategy achieves accurate tracking of the desired pivot steering trajectory under high， medium and low adhesion surfaces， and limits the steering center offset to within 500 mm， which improves the pivot steering flexibility and lateral stability of the off-road vehicles and realizes "fast and stable" pivot steering. 

A new design method of a macro-biological pressure-retaining sampling-culture system at full-ocean depth(operating water depth of 11 000 m) was presented. The system might realize macro-biological pressure-retaining sampling in the full-ocean-depth environment, and at the same time met the requirements of pressure-free transferring and in-situ culture in the laboratory. The basic structure and working principles of the full-ocean-depth macro-biological pressure-retaining sampling-culture system were introduced in details, and the full-ocean-depth macro-biological pressure-retaining sampling-culture system were designed and analyzed. With the developed engineering prototype of the full-ocean-depth macro-biological pressure-retaining sampling, the pressure-retaining performance and sample transfer tests of the full-ocean-depth macro-biological pressure-retaining sampling were carried out. The experimental results show that the full-ocean-depth macro-biological pressure-retaining sampling may bear the pressure of 127 MPa, and the greater the internal pressure of the sampler, the better the pressure-retaining performance, and the transfer without pressure drop is realized in the ultra-high pressure environments, to provide high-quality macro-biological pressure-retaining samples for deep seabed scientific research.

In the automatic cutting processes， accurate and reliable detecting of tool wear states was the key to ensure the processing quality and efficiency. Aiming at the problems of tedious feature extraction with low accuracy and traditional deep learning network could not extract the hidden information of data comprehensively， an improved model was proposed based on integration of CNN and BiLSTM by adding batch standardization layer to CNN and using two BiLSTM layers. The model could automatically extract the spatial and temporal features of cutting forces， vibration and sound signals after wavelet threshold denoising and down sampling to realize tool wear detection. Compared with CNN-BiLSTM model and traditional deep learning model， the accuracy and stability of the improved model are greatly improved. The proposed method provides technical support for accurately detecting the tool wear states in the automatic machining processes and improves the production efficiency and machining quality. 

The feed system of CNC machines driven by PMSM servo system was susceptible to parameters variation, external disturbances and other uncertain factors, an AFOSMC scheme was designed based on NDO. First, the dynamic mathematical model of PMSM with uncertainties was established. Then, FOSMC and adaptive control were combined to suppress the chattering phenomenon of integer order sliding mode control, and the switching gain might be adjusted in real time to improve the control accuracy of the system. However, external disturbances would have a great impact on the system. Therefore, NDO was used to identify the external disturbances in real time, and the observed values were introduced into AFOSMC as feedforward compensation to improve the anti-interference ability of the controller. Finally, the experimental results show that the designed AFOSMC may weaken the chattering phenomenon effectively based on NDO, improve tracking performance and anti-interference ability of the feed systems.

To establish the dynamics model of complex rotor systems in turboshaft engines，the complex structure was modeled by substructure method on the basis of finite element method and piecewise linear fitting. After reducing the degrees of freedom of the systems，motion equations of the rotor systems were derived. The validity of the model was verified by theoretical analysis and experiments，and the vibration characteristics of the gas generator rotors were analyzed. The results show that the established model may significantly reduce the complexity of the overall systems and greatly shorten the calculation time under the premise of ensuring the solution accuracy.

A UWB positioning algorithm was proposed based on neural networks and self-adjusting Kalman filters for improving the positioning accuracy of current a certain three-line automatic driving rail transport system vehicles. The UWB tags and base stations were used to collect large amount of distance information between tags and various base stations and collect the actual locations of the corresponding tags, and the neural network was trained. The distance information between the tags and various base stations was sent to the centralized control center server through the network during the real-time positioning stage, and the real-time locations of the UWB positioning tags were obtained by the optimized neural network. The self-adjusting Kalman filter was used to improve the accuracy of the real-time tag positions furtherly. A set of UWB tag moving trajectories containing inclines, straight paths, and curves were designed for simulation based on the actual vehicle operation, and a UWB positioning system was built, the moving trajectories of the tags were designed, the UWB positioning algorithm combining the neural network and self-adjusting Kalman filter was verified through experiments. The results show that the maximum positioning error of the positioning algorithm combining neural network and self-adjusting Kalman filter is as 223.58 mm, and the average positioning error is as 43.16 mm, and the root mean square value of the positioning errors is as 42.06 mm. The positioning algorithm proposed combining the neural network and self-adjusting Kalman filter has the advantages of higher accuracy, better real-time performance and stability compared with the three-point positioning algorithm, Kalman filtering algorithm, and neural network algorithm, and the current positioning requirements of the three-line rail transports may be fulfilled.

At present, the collision detection of CNCMTs usually used the basic simulation function of the CNC system to detect machining G codes, which only considered whether there were collisions between the tool and the ideal workbench, or the tool and fixtures on the tool path during the processes. It was difficult to meet the actual dynamic processing environments, clamping mode and fixture change of open heavy-duty machine tools. So the digital twin was introduced into the collision detection of open heavy-duty CNCMTs, and a perception-evolutionary prediction-feedback collision detection framework was constructed. By constructing digital twin of CNCMTs and dynamically sensing machining elements such as workpieces ，fixtures and cutting tools, the evolution of the digital twin should be driven by perception data, and the potential interference phenomenon would be predicted in the machining processes. Thus, the efficiency of the CNCMTs was improved, and the potential harm was avoided. The method was applied to the collision detection of heavy-duty CNC gantry boring and milling machine ZK5520, and the effectiveness and feasibility of the proposed method were proved.

The sealing structure was the core component that kepts the wall-climbing robot running. As a new type of sealing structure with the characteristics of good wear resistance, large load, and strong wall adaptability, was increasingly studied in the field of wall-climbing robots. The failure criterion of the rolling seals under typical working conditions was proposed based on the research of the rolling seal mechanism. The failure mode of the rolling seal structure was obtained by analyzing the sealing mechanism. The specific failure modes under the three typical working conditions of horizontal translation, vertical translation, and steering were analyzed. Then the relationship between key design parameters and seal leakage was established based on the dynamics method and based on these analyses, the failure criterion of the rolling seal was proposed. Eventually, the prototype improvement and experiments were completed based on the proposed failure criterion. The experimental results show that the enhances design improves the adsorption stability, which verifies the validity and accuracy of the rolling seal failure criterion. 

The optimal speed planning for connected HEVs under urban traffics with multiple intersections was investigated. The feasibility of using the double-layer pseudo-spectral method to optimize the energy management strategy of HEVs was discussed. Firstly, based on signal phases and timing, the optimal speed was obtained via pseudo-spectral method to ensure that the vehicle might pass through the intersections without stopping. Then, based on the planned speed, the power distribution between engine and battery was arranged with the goal of minimizing vehicle energy consumption. Simulation results show that the proposed method may reduce fuel consumption by 6.9% compared with manual driving, and the effectiveness of the proposed strategy is further verified by real vehicle tests.