In order to extract pure cell populations from multiple cell populations or to extract the required components from complex samples， a novel regulation method of induced?charge electro?osmotic （ICEO） is proposed， based on the superposition effect of dual electric fields， to study the remodeling mechanism of the ICEO vortex and its particle control performance. Firstly， a multi?physical coupling simulation model is established and the asymmetric evolution mechanism is studied. Secondly， the particle control device is designed and processed， and the particle control experimental system is built. Then， the aggregation and longitudinal migration characteristics of single particle induced by asymmetrically ICEO vortices at different voltages are studied. Finally， aggregation and separation characteristics of various particles within the asymmetric ICEO vortices are explored. The results show that this method can achieve the aggregation， migration and separation of micro?scale particles in a simple control way， and it has great application potential in the field of environmental detection and disease diagnosis.

Aiming at magnitude estimation in earthquake early warning， a segmented estimation method based on convolutional neural network （CNN） is proposed. The input is the waveform starting from the P wave onset and lasting 3 s. The output is the estimated magnitude range （local magnitude M_L≥5.0 for large earthquake and M_L<5.0 for small earthquake）. If the waveform belongs to the large earthquake range， the alarm will be sent directly； if the waveform belongs to the small earthquake range， the specific magnitude value will be estimated. For the estimation of magnitude range， the accuracy of the CNN model can reach 98.04%. The mean absolute errors （MAE） of estimating small earthquake magnitudes based on parameters τ_c and P_d are 0.20 and 0.31， respectively. The results demonstrate the efficacy of the proposed segmented magnitude estimation method in accurately early warning large earthquakes and reducing the probability of missed warnings. Additionally， it enhances the precision of small earthquake magnitude estimation.

Finite element simulations were used to analyze the influences of the position， number， size， and shape of process holes on the forming performance of cold stamping tailor rolled blank U?channel parts. The results indicated that tailor rolled blanks are advantageous in reducing residual stresses in the sidewalls when holes are positioned in the thin zone of the sidewalls. The springback of the tailor rolled blank was found to be less dependent on the distribution of process holes but more influenced by their number， radius， and shape. From the standpoint of forming performance and springback impact， it is preferable to locate process holes in the thick zone and transition zone areas at the end of the U?channel part of the tailor rolled blank. If holes must be drilled in the sidewall， the thin zone area is the preferred location. In adjacent areas， the number of process holes should ideally not exceed two， with smaller sizes being more suitable. A shorter distance along the rolling direction is also preferred for the shape of the process hole.

In order to achieve stable combustion of gas in micro?burners and obtain high and uniform temperature distribution on the outer wall， a combustion model of gas in micro?burners with backward steps was established. Firstly， the influence of the number of backward steps on micro?combustion was studied. Secondly， the combustion characteristics of methane/hydrogen in micro?burners with three backward steps under different hydrogen ratios were studied. The results show that the backward step structure can stabilize the flame. When the inlet velocity of the gas is 3 m/s， the equivalent ratio range （0.3~5.0） of the gas with three backward steps is larger than that （0.5~4.6） with one backward step. Although the backward step structure increases the temperature of the outer wall， the temperature difference of the outer wall becomes larger， and the temperature uniformity is reduced. Adding a certain proportion of methane to hydrogen not only slows down the combustion reaction rate of the mixture， but also obtains higher outer wall temperature and energy output.

Under the working conditions of heavy load and alternating load， spiral bevel gears are prone to tooth root cracks， which lead to tooth fractures. It is necessary to clarify the influence mechanism of cracks on meshing characteristics and provide theoretical basis for fault diagnosis. The finite element model of bevel gear pairs is constructed， and the static contact simulation analysis is carried out by ANSYS software. The fault models of various crack types are set up by the method of unit node replacement， and the influence of cracks on the meshing characteristics of spiral bevel gear pairs is discussed. The results show that when the contact ellipse of gear teeth moves to the crack area of gear teeth， the meshing stiffness gradually decreases with the increase of the severity of the crack， and the maximum reduction of stiffness under the condition of plane crack， space crack and broken teeth is 27.58%， 14.12% and 32.82% respectively. The contact ellipse position， crack position and depth of gear meshing will make the contact stress of tooth surface and the bending stress of tooth root have different increasing and decreasing trends.

In response to the technical development needs of improving the vibration resistance and crossing ability of outdoor AGV（automated guided vehicle）， a follow?up wheel vibration isolation system for outdoor AGVs is proposed and designed. Based on the vehicle suspension dynamics theory， a 7?DOF dynamic model and differential equations for motion in all directions are constructed. A simulation model is established using Simulink software environment. Based on this， three evaluation indicators for AGV stability are simulated and analyzed using triangular wave input as road excitation. Experimental tests are conducted on the obstacle?crossing capability of the AGV， along with efficiency comparison experiments between different obstacle?crossing simulation methods. The experimental results show that the mean maximum displacement of spring?loaded mass in the AGV is 13.42 mm， and the prediction error of the established simulation model is 4.6%， which is lower than that of the traditional Adams software’s simulation prediction error. The use of the proposed Simulink?based obstacle?crossing simulation method saves approximately 28% working time for designers compared to the traditional Adams simulation method.

To solve the problems of high cost， poor stability， and difficulty in deployment， a new indoor positioning system combining time of arrival （TOA） and angle of arrival （AOA） is proposed. The system is composed of the positioning base station and the controlled positioning unit. It is characterized by using the ultrasonic sensor arranged in the firing type to obtain the distance characteristics between the positioning base station and the controlled positioning unit， and using the angle sensor to obtain the angle characteristics of the controlled positioning unit relative to the positioning base station. A single base station is used to realize the accurate indoor positioning process. The basic structure and principle of the system are analyzed， the positioning model is established， and the fixed point positioning accuracy and tracking positioning accuracy are experimentally verified in a certain range. The experimental results show that the system has a simple structure， easy installation and strong robustness. The maximum fixed?point positioning error within the test range is less than 5 cm， and the tracking positioning error is less than 15 cm.

To address the conflict between tracking accuracy and vehicle stability in the path tracking task of autonomous vehicles， a path tracking controller considering lateral and roll stability was proposed. Firstly， a four?wheel independently?driven intelligent vehicle is taken as the research object to develop an integrated model predictive controller based on active steering and motor?driven torque distribution to ensure vehicle stability and tracking accuracy under extreme conditions. Secondly， vehicle stability constraints are designed using the phase plot method and the maximum side slip angle of tires to optimize dynamic performance， while roll constraints are established using the zero?moment method to prevent rollover. Finally， simulations comparing the proposed controller with the existing ones demonstrate that the new controller can maximize vehicle dynamics to the fullest， and enhance path tracking precision while concurrently ensuring vehicle stability.

Based on the study of the motion characteristics of a single abrasive particle under two?dimensional ultrasonic vibration， the average chip thickness was determined according to the principle of constant volume. Based on the geometric shape of the abrasive grain and the elastic recovery rate of the material， the chip deformation force model is derived. According to the friction force from the elastic contact between the abrasive particles and the workpiece and the chip outflow， the friction model is established. Considering the impact of high frequency vibration on the total grinding force， the impact force model is obtained. Combining the chip deformation force， friction force and impact force model， the grinding force model of two?dimensional ultrasonic grinding is obtained. Through the experimental study of two?dimensional ultrasonic grinding of alumina ceramic materials， the constants in the model are determined， and the rationality of the grinding force model is verified. The results show that the average errors between the experimental values and the theoretical values of the normal force and the tangential force are 11.09% and 8.07%， respectively， and the maximum error does not exceed 20%. Thus， the model has a predictive effect.

In order to improve the surface quality of 2.5D C_f /SiC composites in micro?grinding， a three?factor and five?level micro?grinding orthogonal experiment was carried out with 500# diamond abrasive and 0.9 mm diameter electroplated micro?grinder. The influence of micro?grinding speed v_s， grinding depth a_p and feed speed v_w on the grinding performance evaluation parameters （grinding force， surface roughness and surface morphology） was analyzed by range and variance. The variation law of grinding performance evaluation parameters with process parameters was analyzed through the test results at different levels. The results showed that the grinding depth has the greatest influence on grinding performance， and the feed speed has the least influence. When the grinding depth and feed speed increase， the surface roughness and grinding force gradually increase， and there are more surface defects； when the micro?grinding speed is increased， the surface roughness and grinding force gradually decrease， the surface morphology is flat and uniform， and there are fewer defects.

Defects in laser melting deposition are key problems restricting its development. Achieving precise automatic identification of defects is a crucial approach to enhance the application level of laser melting deposition technology. A novel algorithm for extracting the melt pool’s transient characteristics was presented， and the relationship between transient characteristics and lack of fusion defects of the deposition layers was found. Moreover， a dataset of the melt pool’s transient characteristics was established. The mainstream recognition algorithms were trained and tested， leading to the identification of the most effective model， ResNet 34. In order to solve the poor fitting training loss effect and slow calculating speed of ResNet 34， a hybrid LRCN 64 model was proposed combining the traditional convolutional networks and LSTM（long short?term memory） networks. It exhibited remarkable accuracy and significant calculating speed. The testing accuracy of the LRCN 64 model reaches 95.8%， thereby realizing the identification of lack of fusion defects， which provides valuable technical support to facilitate online non?destructive testing of deposited parts.

Given the uncertainty in the development characteristics of rock discontinuities， a method for searching for irregular sliding surfaces in rock slopes was proposed. Using the rock masses of rock discontinuities III and IV as research objects， and utilizing the block cutting and key block identification modules in the GeoSMA-3D （geotechnical structure and model analysis-3D） system， the sliding mass is set as a geometric polyhedron. The internal block combinations of the rock mass form the polyhedron， with the surface of the polyhedron representing the sliding surface. The analytic hierarchy process is introduced to establish semi?quantitative evaluation criteria for the sliding surface. Using the Xiaopanling slope as a case study， the criteria for determining the irregular sliding surface were integrated into the GeoSMA-3D system. The study identifies six key blocks and three combined sliding surfaces. The potential sliding surface has a safety factor of 1.046， and the geometric shape of the sliding surface aligns with the actual slope with a fitness of up to 95%. This validates the rationality of the proposed method for searching sliding surfaces in rock slopes.

To reveal the disastrous failure and precursory characteristics caused by the instability of the residual coal pillar under the deep mining， the microseismic （MS） characteristics from the dynamic response of the lower mining strata are analyzed by using the statistical method. The control characteristics of multi?source parameters in the prediction of rock fracture are explored. The key tangent slope and multi?parameters are established as the control index model and prediction method of strong dynamic response of rock stratum. The results show that the data discreteness can be avoided with the statistical method for source parameters and the aggregated damage area of rock stratum can be delineated. In the process of dynamic response from weak to strong， the source parameters have typical time characteristics： the energy index presents three stages： from sharp rise to steady to sharp decline， and the cumulative apparent volume is in the stage of smooth transition to sharp increase， and the b value is in the trend of decreasing to low value and stable. Compared with the ground pressure behaviour characteristics of the residual coal pillar area， the response model and prediction method of taking the logarithm value of energy index and the change rate of cumulative apparent volume trend change as the main control precursor index， and multi?source parameters such as total energy， b value and ground pressure as the secondary control precursor index are feasible.

To address the limitation of the existing super?resolution reconstruction of remote sensing image algorithms in fully extracting and utilizing features and coping with high computational complexity in complex scenes， a Transformer network model for super?resolution reconstruction of remote sensing image based on multi?scale feature fusion was proposed. The multi?scale residual Swin Transformer module was introduced to fully extract features and reduce the module redundancy used for flat feature extraction. A feature fusion refinement module was established that can fully extract image features to improve network performance. Based on the public UC Merced Land Use dataset， the experimental results show that the number of parameters required by the proposed model is only 61.6% of the parameters compared with the current mainstream super?resolution reconstruction method EDSR model. The peak signal?to?noise ratio and structural similarity of the reconstruction results at different scales are increased by 0.82 dB and 0.024 on average compared with the EDSR model. Through comparative analysis， it is proved that the model proposed can effectively reduce the redundancy of network parameters while improving the quality of the image. It can significantly improve the quality of the reconstructed image to meet the requirements of high?resolution remote sensing image processing.

To suppress the thermal runaway（TR） and its propagation in lithium?ion batteries， a nitrogen and water mist（NWM） biphasic flow system is constructed. By changing the pressure and duration of the system， comparing the surface temperature and the volume fraction of flue gas emitted from the batteries， and analyzing the cooling power of NWM and the heat accumulation in the battery module， the mechanism and effectiveness of NWM system in suppressing TR propagation in the lithium?ion battery module are studied. The results show that under the synergistic effect of NWM cooling the battery module surface and the suffocation mechanism， duration of the jet fire of the battery module is significantly reduced. Moreover， higher NWM pressure and longer duration of the system leads to lower peak temperature on the battery surface， and time intervals for TR propagation between adjacent batteries increases， demonstrating a significantly enhanced effectiveness of NWM in suppressing the propagation of TR in lithium?ion batteries.

Based on the reality that Chinese listed enterprises generally rely on suppliers and the important background of the construction of “Digital China”， text analysis technology is used， and the data of A?share listed companies from 2007 to 2021 are selected to investigate the consequences and internal mechanism of supplier relations on corporate digital transformation. The research results indicate that， a significant “U”?shaped relationship between supplier concentration and corporate digital transformation. This “U”?shaped relationship is more remarkable in industries with low levels of competition， non?state?owned enterprises， and specialized new enterprises. Supplier concentration has an impact on corporate digital transformation by dynamically affecting trade credit financing.

Based on the resource conservation theory， the cross?domain network use behaviors of 742 employees were classified by the latent profile analysis （LPA） method. The difference of the influence of different cross?domain network use behavior on work?family relationship was further analyzed. The research results showed that employees’ cross?domain network use behavior can be divided into three types. They are respectively named avoidant cross?domain network use behavior， moderate cross?domain network use behavior and active cross?domain network use behavior. Among them， the level of work?family relationship of employees with active cross?domain network use behavior is the highest. Meanwhile， compared with work?family conflict， employees’ cross?domain network use behavior has a greater impact on work?family enrichment.

Basing on the static and dynamic panel models，the effect of the development of sci?tech finance on regionat sci?tech innovation in China is empirically studied， and the moderating effect of digital economy on that effect is explored. It is found that： sci?tech finance plays an obvious role in supporting sci?tech innovation， but this is mainly reflected in the short term， and the long?term effect is reduced. The effect of different sci?tech finance is discrepant at different stages of sci?tech innovation. Fiscal expenditure and financial institution loans on sci?tech finance plays the main supporting effect on the stage of technology research and achievement transformation； financial institution loans on sci?tech finance still plays an important role， and the role of venture capital is growing at the stage of technology promotion and industrialization； in the process of supporting sci?tech innovation through sci?tech finance， digital economy has a positive moderating effect.