In recent years, the increasing frequency and intensity of extreme rainfall events have significantly amplified the risk of urban flooding. Such events not only cause severe damage to surface infrastructure but also pose critical threat to underground systems, inluding metro networks and underground parking facilities . This study reviews representative metro flood incidents over the past decade, with particular focus on the Hong Kong "9·8" rainfall event and the Zhengzhou "7·20" extreme rainstorm. Through a comparative analysis of the flooding processes and associated impacts, deficiencies in flood prevention planning and emergency response systems are identified, particularly in the Zhengzhou case. Based on these case studies, the predisposing environmental conditions and governing mechanisms of metro flooding are examined. To effectively prevent flood intrusion, metro entrances and exits should be located in relatively elevated areas and designed in strict accordance with flood protection standards. At the same time, protective facilities should be installed at the exit or entrance of the metro station. As an innovative flood control facility, the hydrodynamic automatic floodgate is specially designed for underground infrastructures, which can effectively resist flood intrusion during extreme rainfall events. In addition, this study puts forward prospects for metro flood control countermeasures, emphasizing the establishment of an early warning system, preemptive intelligent early warning of urban flood risks, strengthening urban emergency management, and public disaster awareness. The findings provide a theoretical basis and practical guidance for urban flood mitigation and the safe operation of metro systems.
Existing ensemble-based positive and unlabeled learning (PUL) methods often fail to adequately account for the impact of noise in both positive and unlabeled samples during classifier construction. To address this issue, this paper proposes a naïve Bayesian ensemble-based PUL algorithm (NBEB-PUL), which consists of two stages: label assignment and noise filtering. In the label assignment stage, NBEB-PUL employs naïve Bayesian classifiers as base learners and integrates them using the AdaBoosting ensemble strategy to construct a strong classifier, named Ada-NBC. This strong classifier is then utilized to compute the mean posterior probabilities of a validation set for unlabeled samples, enabling explicit modeling of labeling uncertainty. Based on low-uncertainty predictions a set of reliable positive samples is iteratively identified, along with a residual unlabeled sample set. In the noise filtering stage, NBEB-PUL leverages the ensemble classifier generated in the first stage to reclassify the reliable positive set and the residual unlabeled set, resulting in Ada-NBC-predicted positive and negative sample sets. These sets are then intersected with the first-stage positive and unlabeled sets to extract overlapping high-confidence samples, forming the final positive and negative sample sets. The samples pruned from the dataset during this process are identified as noise. The feasibility, rationality, and effectiveness of NBEB-PUL were validated on 23 benchmark datasets from UCI and KEEL. Experimental results demonstrate that the algorithm exhibits stable convergence during the training process as the number of iterations increases. Moreover, NBEB-PUL outperforms six state-of-the-art PUL algorithms (S-EM, Biased-SVM, Modified-PUL, PU-LP, LP-PUL, and AdaPU) in terms of classification accuracy under varying positive sample proportions of 0.45, 0.40, 0.35, and 0.30. These results confirm that NBEB-PUL provides an effective and robust solution for positive and unlabeled learning in the presence of noise.
Accurate characterization of the microscopic occurrence states of remaining oil is essential for improving oil recovery in low-permeability reservoirs during the high water-cut stage. To advance the characterization of remaining oil from static description to dynamic quantitative analysis and to clarify the evolution of remaining oil within rock pore spaces during water flooding, a multi-parameter coupled dynamic characterization method is developed. Microscopic displacement experiments are conducted using computed tomography (CT) scanning technology. By comprehensively analyzing parameters including the total number of remaining oil clusters, average cluster volume, contact area ratio, shape factor, and Euler number, the coupled evolution of remaining oil quantity, spatial distribution, and morphological characteristics throughout the entire displacement process is quantitatively characterized. The results show that during water flooding, the total number of remaining oil clusters continuously increases while the average cluster volume decreases, resulting in an increasingly dispersed distribution; and the greater the core permeability, the more pronounced these trends become. As displacement proceeds, the oil phase is progressively detached from pore walls, leading to a continuous reduction in the oil-rock contact area ratio. Although the cores are water-wet, strong reservoir heterogeneity causes portions of residual oil remain in contact with the pore walls during the late development stage, suggesting the further recovery can be achieved by reducing interfacial tension in subsequent development stages. In addition, well-connected cluster-type remaining oil is mobilized and gradually transforms into other occurrence types during water flooding. At the residual oil stage, cluster-type remaining oil dominates, forming a complex occurrence system in which cluster residual oil coexists with other morphologies. This observation indicates that the enhanced oil recovery strategies should shift from a single displacement mechanism to a comprehensive mobilization mechanism targeting multiple residual oil morphologies. This study provides critical insights linking microscopic mechanisms to macroscopic development strategies and offer an important scientific basis for the efficient development of the G76 fault-block reservoir in the JD Oilfield during the high water-cut stage.
To address the finite-time control problem of a nonlinear connected vehicle platoon subject to both matched and unmatched disturbances, we propose a control strategy to ensure the finite-time stability of the connected vehicle platoon under complex disturbance conditions. It is assumed that all following vehicles in the platoon can obtain the state information of the lead vehicle through vehicle-to-vehicle communication. A disturbance observer is designed to accurately estimate two types of disturbances within finite time. Subsequently, based on a constant inter-vehicle spacing strategy and terminal sliding mode theory, a finite-time sliding mode control algorithm is proposed. Finally, numerical simulations are conducted to evaluate the effectiveness of the proposed control strategy. The results show that, even in the presence of both types of disturbances, the observer can rapidly estimate two types of disturbances within 0.5 s, and the position, velocity and acceleration tracking errors converge within finite time, thereby effectively ensuring stability and robustness of platoon motion. The comparative studies with existing consensus methods and proportional integral derivative (PID) control methods show that the proposed algorithm achieves an average root mean square error (RMSE) of 0.199 m in position, 0.163 m/s in velocity, and 0.296 m/s2 in acceleration, with the maximum absolute value of position tracking error remaining below 0.9 m. These values are consistently smaller than those obtained using PID control. Furthermore, robustness analyses under diverse conditions, including varying communication delays, sensor errors, vehicle dynamics parameters, packet loss rates, and disturbance magnitudes confirm that the proposed method maintains satisfactory performance under these challenging scenarios.
Extreme learning machines (ELMs) are prone to numerical instability and overfitting when dealing with ill-conditioned matrices, primarily due to the use of the Moore-Penrose pseudoinverse for output weight estimation. To address these issues, this paper proposes an ELM-based on Q-learning (Q-ELM), in which the determination of output weight is reformulated as a Markov decision process (MDP) by utilizing a reinforcement learning agent to iteratively searches for optimal solution within a discrete action space. A weight clipping mechanism is introduced to impose implicit hard-constrained regularization, effectively suppressing noise amplification. Theoretical analysis establishes the convergence of the proposed algorithm under convex objective functions and derives its noise robustness bounds. Experimental results demonstrate that Q-ELM significantly outperforms the standard ELM, achieving accuracy improvements of up to 5.39 percentage points on the small-sample scenarios (Glass dataset) and 2.19 percentage points on the high-dimensional ill-posed task (SynapseMNIST3D dataset). These gains highlight the enhanced robustness and classification performance of the proposed model. Furthermore, across 19 benchmark datasets and in comparison with nine competing algorithms, Q-ELM attains an average ranking of 4.16, securing the top position among all single-hidden layer feed forward network (SLFN) algorithms. Notably, statistical significance tests confirm that Q-ELM yields superior results compared to conventional ELM and its SLFN variants.
Drill-and-blast tunneling generates large volumes of rock spoil, and efficient and accurate characterization of block size distribution is critical for enhancing resource utilization. However, traditional manual measurement methods are time-consuming and limited in spatial coverage. To address these limitations, a UAV-based image acquisition approach combined with deep learning was employed to automatically extract rock block contours. The block size distribution and resource utilization potential of tunnel spoil were investigated using a case study from Dangshun Tunnel spoil yard in Qinghai Province, China. The results show that the median block diameter is approximately 300 mm, with maximum sizes exceeding 1 500 mm. Both the size and morphology of rock blocks exhibit systematic spatial variation along the slope, with particle size gradually increasing from fine to coarse from the slope crest to the slope toe, while block shape transitioning progressively from subrounded to angular. In recently dumped areas, rock blocks largely preserve their original blast-induced fragmentation features, with a relatively high proportion of large fragments exceeding 1 000 mm in size. This suggests that optimization of charging structure and blasthole spacing is necessary. Based on gravity-induced sorting characteristics, a zoned resource utilization strategy is proposed in which medium- and fine-grained materials at the crest and slope face zones can be directly used as fill or road construction materials, whereas coarse blocks accumulated at the slope toe should be crushed for recycling. The findings provide a technical basis for efficient resource utilization of tunnel spoil and optimization of blasting design in tunnel engineering.
Ground collapse poses a significant challenge to urban infrastructure safety and sustainable development, often resulting in severe economic losses and threats to public safety. Damage to underground pipelines is recognized as one of the primary triggers of such failures. Focusing on sandy soil conditions in Guangzhou, a series of laboratory model tests were conducted to simulate the full evolution process of subsurface cavity formation and subsequent ground collapse induced by pipeline leakage. The effects of key factors, including pipeline burial depth, defect size and internal flow velocity, on the soil erosion process were systematically investigated. The results show that leakage from damaged pipelines induces outward seepage flow, leading to progressive erosion of surrounding soil and particle migration. This process results in the gradual formation of arch-shaped underground cavities, which continuously expand with ongoing soil loss. Collapse occurs when the soil arch can no longer sustain its self-weight and the overlying load, ultimately forming characteristic hourglass-shaped erosion cavity. Parametric analysis shows that increasing burial depth, reducing defect size and lowering the flow velocity can effectively decrease the scale of ground collapse and shorten the evolution time of subsurface cavities, thereby mitigating collapse risk. The findings provide a quantitative basis for understanding the mechanisms of pipeline leakage-induced ground collapse and offer theoretical guidance and technical support for the prevention and mitigation of urban subsidence hazards.
To address the lack of clear criteria for well and layer selection in the offshore applications of fracturing-flooding technology, this study develops a refined numerical simulation method to quantitatively characterize the oil enhancement mechanisms associated with high-pressure fracture-induced permeability improvement, surfactant-assisted oil displacement, and soak-imbibition processes. Based on numerical simulations, selection criteria for wells and layers suitable for fracturing-flooding are systematically investigated, and the relative influence of key selection indicators on incremental oil production is analyzed. Results show that the oil-saturation condition play a more critical role than original reserves and physical properties of the reservoir in well and layer selection. Specifically, a pressure maintenance level greater than 0.4 and an oil saturation of exceeding 40% are identified as the lower limits for candidate wells and layers. After fracturing-flooding treatment of the optimized target wells and layers, the initial incremental oil production reaches 50 m3/d, and the productivity index of the target layer increases from 1.2 m3/(d·MPa) to 9.2 m3/(d·MPa), and the cumulative incremental oil production reaches 1.01 × 104 m3 within six months, indicating that the fracturing-flooding performance meets the design requirements. Field practice shows that the proposed refined numerical simulation approach—which incorporates formation energy replenishment, pore and permeability enhancement, pressure-induced fracturing, surfactant-assisted washing oil, and imbibition displacement mechanisms—together with the established well and layer selection criteria, provides effective technical support for fracturing-flooding operations in offshore low-permeability reservoirs.
To address the low utilization of regenerative braking energy and significant energy waste in conventional railway systems, this study aims to improve energy utilization efficiency by developing an energy-efficient timetable optimization framework incorporating energy storage systems. The proposed approach enables coordinated utilization of regenerative braking energy through both immediate and delayed mechanisms. Based on operational characteristics of conventional railways, an optimization model is formulated with the objective of maximizing total regenerative braking energy utilization. The decision variables include train departure intervals at the origin station and dwell times, subject to multiple constraints such as safety headway requirements, dynamic overtaking rules, and energy storage capacity limits. A simulated annealing-based solution algorithm is developed and validated through a case study of the Lanzhou-Wuwei section of the Lanxin railway. A test scenario involving 15 stations and 43 passenger and freight trains during a certain daytime period shows that the total regenerative energy utilization reaches 4 134.38 kWh, corresponding to a utilization rate of 56.28%. Sensitivity analysis is conducted to examine the effects of departure intervals, dwell times, storage capacity, train frequency, and the proportion of regenerative braking-capable locomotives on energy-saving performance. Results indicate that under high-density operation of regenerative braking-capable locomotives (approximately 58%), coordinated optimization of timetable parameters and storage capacity can achieve a maximum energy-saving rate of 86.33%. An economic evaluation based on net present value (NPV) method shows consistently positive returns on energy storage investment over a 10-year horizon. These findings demonstrate that integrating energy storage systems into timetable optimization not only enhances energy utilization efficiency but also yields significant economic benefits.
The DX gas reservoir is characterized by ultra-high pressure, extremely low porosity and permeability, well-developed fractures, and a strong bottom-water drive. During depletion, the decline in formation pressure intrudes rapid bottom-water invasion through fracture networks, resulting in early water breakthrough and significant trapping of natural gas within micro-fractures and pore spaces, thereby reducing overall recovery. To investigate the bottom-water invasion behavior in fractured tight gas reservoirs, physical simulation experiments were conducted under high-temperature and ultra-high-pressure conditions (136 ℃ and 106 MPa) using full-diameter fractured core samples. The effects of depletion rate and aquifer sizes on key indicators, including recovery factor, water-gas ratio, water-free production period, and gas production per unit pressure drop, were systematically analyzed. Results indicate that both increasing aquifer size and accelerating depletion rate lead to higher water production, increased water-gas ratio, shortened water-free period, and reduced gas recovery. Aquifer size exerts a significantly stronger influence than the depletion rate. Specifically, increasing the aquifer size from 4.5 to 20.0 times the hydrocarbon pore volume results in a nearly tenfold increase in cumulative water production, an rise in water breakthrough pressure by 18 MPa, and a value of 21.61% reduction in gas recovery. In contrast, doubling the depletion rate results in approximately a twofold increase in cumulative water production, a 6 MPa increase in breakthrough pressure, and a 10.97% reduction in recovery. These findings indicate that appropriate control of rational depletion rate and effective water management strategies are critical for improving economic performance of fractured tight gas reservoirs. This research provides a crucial theoretical and experimental basis for understanding and managing bottom-water invasion and offers significant practical guidance for optimizing development strategies in fractured gas reservoirs.
To address the limitations of hydroxyapatite (HA) microspheres, such as insufficient drug-loading capacity and severe aggregation, high-performance drug-loaded microspheres via compounding functional materials were prepared and the underlying synergistic enhancement mechanisms were explored. In this study, HA was combined with graphene oxide (GO), gelatin (Gel), and β-cyclodextrin (β-CD) to fabricate a series of drug-loaded composite microspheres, including HA-Gel, HA-GO-Gel, HA-GO-β-CD, HA-GO-β-CD-Gel, using hydrothermal and emulsion crosslinking methods. After loading with curcumin (Cur) drug, the characterization and test results showed that the drug-loading performance of ternary and quaternary composite microspheres was significantly enhanced by introducing β-CD and GO into the HA-Gel binary composite system. Among them, the HA-GO-β-CD-Gel quaternary composite microspheres exhibited more uniform particle size and better dispersion. The encapsulation efficiency and drug loading capacity reached (40.46±1.41)% and (5.58±0.69)%, respectively, with microsphere particle size ranging from 6.9 to 17.4 μm. Molecular dynamics (MD) simulations were further conducted to elucidate the mechanism underlying the enhanced drug-loading performance. Taking the HA-β-CD binary composite microsphere as an example, the results indicated that the adsorption capacity of HA-β-CD composite microspheres for Cur drug molecules was significantly stronger than that of pure HA microspheres, demonstrating improved drug-loading capacity. The simulation results were consistent with the experimental observations. Overall, HA multiphase composite microspheres, particularly the HA-GO-β-CD-Gel quaternary composite microspheres, can effectively improve the drug-loading performance and show promising application potential.
Coal-tight sandstone composite reservoirs without interlayer barriers are prone to interlayer fracture interference,ineffective stimulation,and difficulties in production evaluation due to strong contrasts in lithology,petrophysical properties,and in-situ stress.To address these challenges in Well WFD1,a differentiated staged fracturing and testing strategy was proposed and implemented based on reservoir characteristics.The coal rock interval was treated using moderate pumping rates(6.0-10.0 m3/min)with medium-scale fracture network stimulation,combined with variable-rate control of fracture height and multi-size proppant placement to enhance conductivity while mitigating water sensitivity damage.The tight sandstone interval was fractured using a moderate-scale,low proppant concentration design with composite proppants(150 μm ceramic fines+212 to 425 μm quartz sand+300 to 850 μm quartz sand),coupled with liquid nitrogen-assisted injection to improve flowback efficiency.During the production stage,a dual-packer pumping string was deployed to achieve effective zonal isolation,enabling independent flowback and production testing of the coal and sandstone intervals.Field results indicate that fracture height growth was effectively controlled,no interlayer communication occurred,and both reservoir intervals were successfully stimulated.Reliable production capacity and fluid properties of the coal-tight sandstone were obtained for each layer.These results indicate that the proposed differentiated staged fracturing and testing technology is effective for coal-tight sandstone composite reservoirs without interlayer barriers and provides a practical reference for safe stimulation and accurate zonal evaluation of similar coal-measure gas reservoirs.
Driver behavior significantly influences vehicle fuel consumption and exhaust emissions.Clarifying the relationship between inefficient driving habits(e.g.,idling and rapid acceleration/deceleration)and vehicle carbon emissions is essential for urban carbon-reduction strategies.Using extensive,high-precision trajectory data,this study applies the vehicle specific power(VSP)method to calculate carbon-emission characteristics at various scales and constructs a comprehensive driving-behavior feature set.The distribution of vehicle carbon emissions is examined,while a t-distributed stochastic neighbor embedding(t-SNE)is employed for feature reduction,followed by k-medoids clustering to identify common patterns in high-emission driving behaviors.Moreover,the maximal information coefficient(MIC)method is applied to analyze the effects of driving behaviors under different road-priority conditions on carbon emissions.Results show that vehicles can be classified into three types,which are idling,normal and aggressive acceleration,accounting for 28.8%,44.9%and 26.3%,respectively.Carbon emissions are strongly correlated with road priority,with significant differences in emission patterns between primary and secondary roads.For high-priority road segments,carbon emissions are strongly associated with acceleration-related features;for low-priority road segments,heading-angle variation,idling proportion,and minimum speed exhibit stronger correlations with emissions.By identifying the driving patterns associated with high carbon emissions,this study proposes targeted strategies to reduce carbon emissions under dynamic urban-traffic conditions,contributing to the improvement of urban carbon-reduction systems.This analysis highlights the importance of addressing driving habits in the development of effective carbon reduction policies in urban environments.
This study models the regulatory process of biomacromolecular phase separation facilitated by biological membranes.Utilizing molecular dynamics simulations based on the coarse graining(CG)Martini force field,we investigate the phase separation of components within polyelectrolyte(PE)and cholesterol-containing dipalmitoyl phosphatidyl choline(DPPC)lipid membrane CDM.Combined with mean field theory,we reveal the intricate physical nature of component phase separation in the PE-CDM system.Our molecular dynamics simulations provide profound insights into the multiphase separation characteristics of the PE-CDM system and identify the driving factors behind phase separation.Furthermore,we delve deeper into the simulation results through mean field theory,elucidating the entropy effects during phase separation in the PE-CDM system and the regulatory mechanisms of component interactions.Our findings illustrate that biomembranes can indeed regulate the phase separation of biomacromolecules at their surfaces,with components such as cholesterol and DPPC playing crucial regulatory roles.Additionally,biomacromolecules can induce phase separation of biomembrane components.The results presented in this paper align with experimental observations,offering a deeper understanding of the critical mechanism by which biomembranes regulate intracellular biomacromolecule phase separation through prewetting.This study can provide valuable references for further research and potential applications.
To extend the diversity of identifiable spatial point patterns,this study is based on the spatial chromatic model(SCM)and investigates the correspondence between spatial chromatic codes derived from singular spatial chromatic tessellations and spatial point patterns.The results show that both the magnitude and statistical characteristics of the spatial chromatic codes can effectively indicate the distribution patterns of spatial points.The proposed method is capable of identifying not only common point pattern characteristics,such as randomness and clustering,but also special configurations including collinearity,cocircularity,and symmetry.Moreover,it facilitates the integration of point pattern recognition with other spatial analysis functions provided by SCM,enabling the analysis and processing of entities and their spatial relationships within a unified framework.The findings of this study provide new insights and analytical approaches for spatial point pattern recognition.
To solve the problem of excessive basic nitrogen compounds in the benzene feedstock of ethylbenzene unit,the adsorption performance of HZSM-5 zeolite,activated clay and waste fluid catalytic cracking catalyst as absorbents was systematically investigated.The structural features,morphological characteristics,and physicochemical properties of the adsorbents were characterized by a combination of analytical techniques,including X-ray diffraction(XRD),scanning electron microscopy(SEM),low-temperature nitrogen adsorption-desorption,and ammonia temperature-programmed desorption(NH3-TPD).The correlation between the adsorption performance and material structure was also established.Results show that HZSM-5 zeolite exhibits highest adsorption capacity and removal efficiency for basic nitrogen.After treatment with HZSM-5 zeolite,the residual basic nitrogen concentration in the raw benzene feedstock was reduced to 0.08 mg/kg,meeting the industrial specification.The structure-activity relationship analyses reveal that the adsorption performance is positively correlated with the specific surface area of the adsorbent,while acidity is not the dominant influencing factor.Breakthrough experiments confirm that the adsorption kinetics follow the Boltzmann model.This research aligns with the trends of green chemistry and sustainable development,demonstrates good potential for industrial application,and provides new insights for adsorbent design.
To enhance the prediction accuracy for both average corrosion rate and pitting corrosion rate of oil well tubing in deep complex environments,and to address the issue of insufficient consideration of base learner heterogeneity in traditional Stacking ensemble learning,an improved Stacking ensemble learning algorithm based on the coefficient of determination(R2)is proposed.This algorithm integrates four machine learning models as base learners:extreme gradient boosting(XGBoost),random forest(RF),support vector regression(SVR),and gradient boosting decision tree(GBDT).The outputs of these base learners are weighted according to their respective R2,and the weighted combination forms the input dataset for the meta-learner.Experimental results demonstrate that,compared with the traditional Stacking ensemble method,the improved model achieves a 25.9%reduction in mean absolute error(MAE)and a 9.7%reduction in mean squared error(MSE)for average corrosion rate prediction,alongside a 2.3%increase in the R2.For pitting corrosion rate prediction,it yields reductions of 11.6%for MAE and 2.0%for MSE,respectively,with a 2.7%increase for R2.These results validate the effectiveness of the proposed algorithm,and the research findings provide valuable support for corrosion prevention,control and safe operational maintenance of deep oil well tubing.
To optimize the crystalline quality of phase-change heterostructure(PCH)thin films and investigate the effects of different substrates and buffer layers on the growth of Sb2Te3 films,we systematically compared the crystalline properties of Sb2Te3 films deposited on Si,SiO2,and Al2O3 substrates.It was found that introducing W or Ti as a buffer layer significantly influences film growth.In particular,the combination of an Al2O3 substrate with W buffer layer approximately 10 nm-thick was shown to markedly promote the high-quality growth of Sb2Te3 films while preserving their orientation consistency.This approach enabled the successful fabrication of PCH films with near-atomic-level flatness.This study provides important experimental evidence and theoretical support for the preparation of high-performance PCH materials.
To enhance the green and efficient treatment and disposal of waste slurry and soil generated during urban engineering construction,we explore the key technologies for their high-efficiency resources utilization and outline future research directions.A coupling theoretical method for phase separation and resource utilization of waste slurry and soil is proposed.Based on the development of material systems,the establishment of process-technology-evaluation methods,and the advancement of intelligent equipment,a complete set of green and efficient resource utilization technologies has been developed to address the large-scale treatment and disposal challenges of waste slurry and soil.Future research on waste slurry and soil can be conducted at three levels:large-scale disposal and treatment,functional engineering utilization,and the development of advanced materials derived from solid waste.Furthermore,a concept of soil-water coupled utilization based on waste slurry and soil is introduced.By employing geopolymer-based solidification and mesoporous functional regulation,waste soil can be transformed into advanced filling materials with permeability,breathability,and multifunctional such as catalysis,carbon sequestration,and pollutant removal.While achieving the treatment and disposal of waste slurry and soil,this approach also generates positive environmental effects for soil and water pollution control.This study provides new solutions to the treatment and disposal challenges of waste slurry and soil in engineering construction and has significant implications for the management of urban ecological environments.
To achieve precise end-effector control from the initial position to the target position of a planar three-link PAA(passive-active-active)system with underactuated first joint,a position control strategy based on the same angular velocity constraint and neural networks is proposed.The dynamic model of the system is established using the Euler-Lagrange equation,and the constraint equations of underactuated joints are analyzed to obtain the angular velocity relationships among the system links.Next,under the same angular velocity constraint for the active links,a Lyapunov-based controller is constructed.A simulation platform is constructed to collect angular data of both active and passive links.The collected data are augmented using a generative adversarial network(GAN),and the mapping between passive and active link angles is established through a deep neural network(DNN).Considering the geometric constraints among link angles,the target angles for all system links corresponding to the end-effector position are optimized using a genetic algorithm,the desired end-effector position are optimized using a genetic algorithm.Based on the Lyapunov function,the designed controller synchronously stabilizes the second and third links to their target angles while achieving passive link angle control.Consequently,the position control objective of the system end-effector is realized through a non-switching control strategy.Simulation results show that,compared with the switching control method,the system end-point under the non-switching control strategy converges to(-0.698 0,1.003 0)within 8 s,with shorter convergence time and smaller errors,avoiding complex integration processes.The non-switching control strategy achieves precise control of the system end-point from the initial position to the target position.
