This paper makes use of the RBF to calculate the interpolated displacement of vertexes on a 3D human body mesh. Through the integration of points of human body landmarks, lines, and customized measurements, a new 3D human body deformation model is created. At the same time, an RBF control point set is formed for the deformation of the 3D human body model. In addition, parametric cubic splines and bicubic surfaces patches are utilized to form a 3D garment model. This approach controls the constraints between 3D model parameters, enabling predictable control over the 3D prototype. Finally, garment design solutions under different methods are compared using SSIM, PSNR, and FID image evaluation metrics. The results show that 3D garment prototypes created using the method successfully imitate prototype designs, proving the usefulness of computer-aided garment designing solutions. The images produced through this method perform extremely well in all aspects and produce the best visual effects in garment designing.
Insects are highly adaptable and efficient in their ability to synthesize and metabolize diverse chemical compounds. The firefly Pyrocoelia spp., (Coleoptera: Lampyridae), is confronted with various chemical assaults throughout its life cycle, the defensive toxins secreted by snails and slugs during firefly larval predation, synthetic pesticides during adult stage, and the risk of auto-toxicity from endogenously produced lucibufagins. Despite the ecological, cultural and economic significance and conservation concerns surrounding fireflies, a clear understanding of their detoxification processes at gene and molecular level is lacking. This review discusses molecular basis of major detoxification enzyme superfamilies, viz., Cytochrome P450 monooxygenases (P450s), Glutathione S-transferases (GSTs), UDP-glycosyltransferases (UGTs), Carboxylesterases (CarEs), and ATP-binding cassette (ABC) transporters. Further the transcriptional regulatory networks, including various pathways that contribute to the induction and expression of these defense genes are discussed in detail. This provides a rationale and a roadmap for developing molecular biomarkers for ecotoxicological risk assessment, deriving conservation strategies for declining firefly populations, and helps in understanding the evolution of chemical adaptation in non-model insects.
For solving the problems which exist when building correct mathematical models for strong coupling, nonlinear, time-changing systems such as robot operating arms, and these problems often make traditional control methods have not good enough decoupling results and bad dynamic performance, therefore we put forward one improved RBF neural network inverse model based decoupling control method. Firstly, the particle swarm optimization algorithm is utilized by us to carry out offline optimization for initial neighborhood clustering radii, therefore obtaining relatively better network parameters. These optimized radius values are then employed for real-time nearby region grouping to dynamically build the RBF neural network reverse model, thus overcoming the restriction of traditional methods which need pre-set network structures. Second, the inverse model that has been found is connected in series with the system being controlled to construct a pseudo-linear system, therefore it enables dynamic decoupling between the joints of robotic manipulators. In the end, one PD controller is combined into a compound closed-loop control system for reducing inverse model errors and strengthening robustness. The simulation which uses a two-degree-of-freedom manipulator to carry out validation proves that our strategy can effectively realize dynamic decoupling and high-precision track following, whose tracking accuracy and robustness all exceed the traditional PD control methods.
Under the background of city-station combination, the different interaction between "fast going to work" and "slow staying" passenger streams inside big city railway hubs causes serious time-space conflicts; The traditional passive, static control approaches have difficulty in dealing with the impulse surges which are generated by incoming trains. For the solving of the zero-sum game problem that lies between traffic evacuation and commercial attraction, this thesis puts forward a dynamic collaboration optimization method for passenger flow on the basis of Model Predictive Control (MPC). Firstly, the complicated space arrangement of the hub is by us abstracted to be a direction queuing network. An innovative multidimensional Macroscopic Fundamental Diagram (MFD) evolution model—incorporating the proportions of heterogeneous passenger flows—is then constructed to precisely quantify the nonlinear frictional impedance that commercial lingering behaviors impose on main-line traffic flow. Second, a multivariable feedforward MPC architecture is established to implement proactive flow restrictions and dynamic path guidance by simulating the trajectories of congestion shockwaves through a rolling prediction mechanism. Simulation results, based on the Chongqing Shapingba Hub as a case study, demonstrate that this system can reduce the maximum queue length at core ticket gates by 58.1% and narrow the variance in travel delays by 67.6%. Furthermore, by identifying a balanced solution on the Pareto frontier—trading a marginal increase in travel delay for a substantial boost in commercial attraction—the system achieves a deeply adaptive collaboration between anti-stampede safety protocols and the economic efficiency of the urban micro-center.
To tackle such problems as the failure to consider diverse behaviors of students and improper dynamism of the curriculum system when teaching the topic of Fundamentals of Accounting, this paper presents a hybrid algorithm approach as the approach to optimize the teaching material and curriculum system. Through the integration of Gaussian mixture models and clickstream data analysis, this can be used to determine the characteristics of student learning behaviors and the model parameters are optimized using the Expectation-Maximization (EM) algorithm. According to the ARCS motivation model, the learning design is based on an all-encompassing teaching evaluation system including both formative and summative evaluations. With the help of empirical mode decomposition, students multi-dimensional behavioral time-frequency features were extracted with the use of 2022 Basic Accounting course at H University School of Accounting as an empirical subject. The connection between patterns of behavior and academic achievement was confirmed and compared the results of teaching in experimental and non-experimental classes. Results show that: The hybrid algorithm is capable of identifying the multimodal distribution features of student behavior. The non-experimental class mean (63.15), median (67), and pass rate (0.77) were significantly lower than the respective measures in the experimental class full sample. There was a significant correlation (0.201) between learning performance and formative assessment at the 1% statistical level.
Aiming at the problems of high dimension of heavy gas turbine operation data, rapid change of working conditions, unbalanced fault samples and high training cost of traditional deep models, a large fault diagnosis model construction method based on LoRA low-rank modulation was proposed. The method takes multi-source operation data as input, establishes a fault semantic representation module, and maps time series signals such as temperature, pressure, flow, speed, vibration and control feedback into high-dimensional representations with context correlation. On this basis, the LoRA low-rank modulation adaptation mechanism is introduced to efficiently fine-tune the parameters of the pre-trained time series large model, and the feature fusion of time domain, frequency domain and working condition is combined to improve the complex fault recognition ability. Experimental results show that the accuracy, recall rate and macro-average F1 value of the proposed model on the test set reach 96.8%, 95.1%and 95.7%respectively, which are better than those of SVM, CNN, LSTM and standard Transformer models. At the same time, the number of trainable parameters is reduced from 88.4M to 9.6M, which reduces the computational overhead while ensuring the diagnosis performance. The research shows that the proposed method has good accuracy, robustness and deployment feasibility in the intelligent diagnosis scenario of heavy gas turbine. Povzetek: Raziskava predlaga metodo velikega modela z LoRA nizko-rang modulacijo za diagnostiko okvar težkih plinskih turbin. Model učinkovito združuje večvirovne časovne podatke in dosega visoko natančnost, priklic in F1, ob zmanjšanem številu parametrov.
In response to the problem that traditional empirical statistical methods are difficult to accurately predict the degree of seismic damage to masonry structures and have a large deviation from actual seismic damage patterns, this paper takes a six story dormitory building with masonry on the bottom frame as the research object, and conducts seismic damage prediction and numerical simulation research. Firstly, a comprehensive modeling strategy is adopted to establish a finite element model of the masonry structure based on ABAQUS. The concrete damage plasticity (CDP) model is selected to describe the material constitutive behavior, and a three line hysteresis restoring force model considering stiffness degradation is introduced to characterize the interlayer mechanical properties; Secondly, the effectiveness of the model was verified through quasi-static experiments. Five amplitude modulated seismic waves were selected for dynamic elastoplastic time history analysis, and the seismic performance of the structure was evaluated based on plastic energy dissipation theory. The local deformation and energy dissipation laws of the wall under different construction measures were compared. Finally, experimental analysis shows that the fusion method of the established mathematical model and finite element simulation can accurately predict the evolution law of seismic damage in masonry structures. The construction of columns is a key measure to improve the seismic performance and collapse prevention ability of masonry structures. The research results can provide theoretical basis for seismic identification and reinforcement of masonry structures.
To investigate the influence of initial moisture content on the mechanical properties of unsaturated loess, this paper focuses on unsaturated loess with varying initial moisture content. Isotropic compression tests at constant moisture content and true triaxial shear tests were conducted. By controlling suction and net mean stress, the relationship between void ratio and net mean stress was analyzed to determine the yield point and yield stress. Furthermore, the effects of confining pressure, saturation, dry density, and clay content on yield characteristics were explored. The experimental results indicate that under isotropic compression at constant moisture content, the void ratio decreases with increasing net mean stress. The e-lnp curve is approximately composed of two intersecting straight lines, with the intersection point representing the isotropic compression yield stress. Both yield stress and yield suction increase with the increase of initial suction. The slopes of the straight lines before and after yielding are less affected by initial suction. Moisture weakens the structural yield characteristics of loess, and the yield stress of different soil samples exhibits varying trends with increasing saturation. The study reveals that initial moisture content significantly affects the LC yield characteristics of unsaturated loess. Yield stress is closely related to saturation, dry density, and clay content. The research findings provide a theoretical basis for the engineering application of unsaturated loess.
A deep learning joint decision model of variable speed limit and on-ramp control for traffic flow control was proposed to solve the problems of mutual isolation, feedback lag and insufficient coordination between the main line speed limit regulation and on-ramp control in the bottleneck area of expressway. The model integrates roadside sensing data, short-term state prediction and closed-loop feedback update mechanism, and synchronously generates the main line speed limit level and ramp release strategy in a unified state space to achieve collaborative suppression of speed attenuation, density accumulation and queue diffusion. Experiments were carried out based on 12.4 km continuous bottleneck road, two on-ramps and 16 weeks of traffic data. The results show that compared with the fixed speed limit strategy, the proposed method increases the average speed of the main line by 10.4%, increases the traffic per unit hour by 10.5%, reduces the congestion duration by 50.6%, and decreases the average queue length of the ramp by 46.4%. The research shows that the joint control framework driven by deep learning can improve the real-time performance, stability and overall efficiency of highway traffic regulation.
This paper utilizes intelligent technology to assist repertoire teaching and singing instruction, combining with the existing computer music software, to explore the development path of the traditional choral teaching mode in the intelligent era. For the teaching of choral repertoire, a real-time music beat tracking algorithm is proposed, which carries out wavelet transform on the pre-processed music signals, detects the peaks of the resulting detail coefficients, constructs and solves the smooth histogram of the music beats, and obtains the real-time value of the music beats. In addition, in order to improve teachers' guidance to students' singing, deep convolutional networks with powerful dimensionality reduction and feature learning ability are used to embed high-dimensional and time-sequential vocal spectral features into the 3-dimensional timbre embedding space, to realize the characterization and similarity metrics of vocal timbre in the 3-dimensional timbre embedding space, and to build a vocal timbre characterization model. The experimental samples are selected and the experimental group and control group are set up, in which the students in the experimental group have 1-6 different semitones of range broadening with the technical assistance of the algorithms and models in this paper, which verifies the important technical roles of real-time music beat tracking algorithms and vocal coloration characterization models in reconstructing the traditional choral teaching mode.
In recent years, with the widespread use of generative artificial intelligence image technology by young people, numerous examples have emerged of these images being employed to convey emotions and gender identity through visual symbols. Using both semiotics and affective computing, this study will conduct content analysis, experimental research and in-depth interviews to explore the symbolic encoding features of gender representation in AI-generated images and adolescents' emotional responses, as well as their underlying connections. Based on the above results, both traditional gender stereotypes in terms of clothing, posture and environment are still visible; at the same time, technology is also creating various forms of gender expression. Adolescents' Emotional Responses to Stereotypic and Pluralistic Representations: Cognitive Conformity vs. Curiosity/Identification. Collect physiological and psychological data on young people using affective computing technology to provide objective support for studying the impact of gender symbols on them. Combine semiotic analysis and affective computing to build a new research system for exploring how gender cognition develops in AI environments; at the same time, provide theoretical support and practical suggestions for regulating the creation of AI images and promoting the healthy emotional growth of adolescents.
A Base Converter was established in an anaerobic environment, and KOH served as an activator for sesame-straw-based materials at a high temperature. D-BC carboxyl-functionalized biochar was prepared by grinding DL-malic acid with BC to adsorb the azo dye reactive bright red X-B, and the promoting effect of carboxyl groups in the adsorption process was explored. The biochar before doping, after doping, and after adsorption were all characterized and analyzed by SEM, BET, XPS and FTIR. Single-factor experiments and model fitting were used to study the adsorption processes of BC and D-BC, and it was found that the surfaces of BC and D-BC consist of dense rhombic pores with a large specific surface area. Adsorption of X-B on BC and D-BC is primarily electrostatic adsorption, hydrogen bond interaction and $\pi$-$\pi$ interaction; the surface functional groups of BC are relatively scarce, so physical adsorption dominates. The relatively large amount of -COOH in D-BC promotes the occurrence of electrostatic attraction and hydrogen bonding, and is therefore considered to have some chemical adsorption characteristics. Both BC and D-BC belong to monolayer adsorption, and the maximum adsorption capacity of D-BC can reach 2877.59 mg/g, which is more than 1.5 times that of BC.
In this paper, MIDI data of Korean traditional music and music from the Central Plains were collected, melodic features and note features of the music data were extracted, and a multi-style melody and arrangement generation (MSMICA) model was constructed. The Multi-Sequence Generative Adversarial Network serves as the main framework of the model, including two core parts, the generator and the discriminator. The generator incorporates a reinforcement learning mechanism to reward with feedback from the discriminator, while the discriminator utilizes the GRU component for the recognition of mesogenic musical elements as well as harmonic music generation, and proves the feasibility of the model based on relevant experiments. During the training process, the Loss and Accuracy values of the model converge to about 0.44 and 0.95 successively in about 60 rounds of iterations, showing good training results. The fluctuation amplitude of the signal features of the model-generated traditional music of the Central Plains style Korean music is between ±0.8, and the signal features in the Meier spectrum, spectral roll-off and chromatic frequency are similar to the real music, and the model-generated music clip is more in line with the traditional music of the Central Plains style Korean music.
In this paper, we first construct a corpus system that can automatically crawl, clean and categorize business texts from the Internet. Then, by introducing the HowNet lexical annotation algorithm and probabilistic sentence alignment model, the corpus is deeply semantically related and structurally aligned to Chinese and English, making it a structured bilingual teaching resource. Based on this, training strategies are designed for oral and written expressions, such as contextualized quiz, role-playing, imitation writing training, and mind mapping-assisted writing, etc., so as to transform the static corpus into an interactive teaching path. A semester-long comparative teaching of 124 students found that the experimental class with AI corpus-assisted instruction had significantly higher overall business English proficiency than the traditionally taught control class, with posttest mean scores of 88.52 and 81.10, and the mean speaking score of students in the experimental class increased from 12.17 to 17.43, far exceeding that of the control class, which was 14.62. The mean score for written expression jumped from 12.27 to 17.29, again significantly higher than the 14.71 of the control class. Statistical analysis of all p-values of 0.000 confirms the significance of the differences. The questionnaire survey shows that more than 80% of the students affirmed that the model is helpful in improving their speaking and writing skills, and more than 90% of the students think that this way of learning is more interesting and easy to learn.
Reducing the waste of building energy is important for the development of green environment. In this paper, based on the dynamic calculation method of various types of cold loads in the building temperature field, a four-wire platinum RTD sensor is selected as the physical device for real-time monitoring of space temperature to improve the credibility of temperature collection. The inverse distance weight model based on local parameter optimization is constructed, combined with particle swarm optimization algorithm and K-dimensional tree, etc., to realize temperature field prediction and energy consumption optimization. The constructed model is applied to the temperature prediction simulation and energy saving optimization practice of large buildings to judge the practical value of the model. The results show that the data collection error is minimized when the search height distance is 2 and the weight coefficient is 2.5-3. In the comparison between the model simulation prediction value and the measured temperature value, the average value of temperature difference in the horizontal and vertical directions is 0.250℃ and 0.162℃ respectively, which is a small error. In 10 energy consumption optimization experiments, this paper's model is able to achieve the optimization of building energy consumption of 0.08-0.80kWh/m². Using the model of this paper can carry out high-precision prediction and energy consumption optimization of building energy-saving design.
In order to analyze the discrete dynamic modeling of intelligent teaching based on big data mining, a deterministic learning theory is put forward based on the research on the continuous excitation characteristics of radial basis function (RBF) neural network. Firstly, the hierarchical analysis framework of dynamic generative data in intelligent teaching is introduced. Then, the modeling /identification based on temporal data is discussed, and the similarity definition and fast identification method of temporal data sequence are studied. Finally, numerical experiments are carried out.The key to realize the local accurate modeling of discrete system dynamics lies in the satisfaction of some continuous excitation conditions, the exponential convergence of discrete linear time-varying systems, and the convergence of some neural network weights along the regression trajectory. These elements reveal the nature of deterministic learning for discrete dynamic systems. The local accurate modeling of discrete system dynamics can be used to time invariant representation of temporal data sequences. Numerical experiments on the fast recognition of temporal show that the error generated by the test mode is smaller than that of the other two when compared with the third training mode. The test temporal is most similar to the third training temporal.
The effective integration of rural eco-agriculture and leisure agriculture is an important realization form of rural revitalization strategy. This paper determines the development of ecological agriculture and leisure agriculture coupling measurement dimensions, in which the ecological agriculture level contains two dimensions of agricultural performance and agricultural input, and the leisure agriculture level has five dimensions of infrastructure, economic benefits, tourism performance, tourism input and social benefits, and the joint entropy and gray correlation methods are used for the calculation of the weight of their indicators. Based on the development content and characteristics of rural ecological agriculture and leisure agriculture, we build a framework for determining the coupling degree of the two with reference to the capacity coupling model and calculation method of physics. Under this framework, the comprehensive development level of the ecological agriculture system of the research sample in ten years is 0.253~0.859, and the comprehensive development level of the leisure agriculture system is 0.214~0.892, and the coupling degree of the two is 0.154~0.707, which reaches an intermediate level of coordination after ten years of development. Accordingly, this paper suggests that at the level of leisure agriculture, based on modern technological tools to strengthen publicity, attract tourists, and promote the growth of economic scale; and at the level of ecological agriculture in the form of economic support for the innovation of its mode of production and the maintenance of the production structure. The integration of leisure agriculture and ecological agriculture forms a high-quality sustainable development path to help rural revitalization.
This paper focuses on the digital communication security of non-heritage Bowen culture, focusing on analysing the network security risk of the platform architecture, the main types of security threats, vulnerability assessment, and proposing the key security technology system of identity authentication, data encryption, intrusion detection, and defense in depth. Combined with the security management system and emergency response mechanism, a multi-level security system is constructed to enhance the security, stability and credibility of digital communication of non-heritage culture. This study is of great practical significance for promoting the safe inheritance of intangible cultural heritage and digital cultural governance.
With the continuous growth of the demand for complex terrain operations, the efficient perception, smooth switching and stable control of wheel-legged robots in unstructured environments have become the research focus. Focusing on the problem of multimodal motion switching and cooperative control of wheeled legged robots on complex terrain, this paper constructs a system modeling and kinematic analysis framework, fuses stereo vision, IMU, encoder, wheel speed meter and foot contact information, and designs a multimodal environment perception and feature fusion method. A motion mode switching mechanism based on switching demand function, dual-threshold hysteresis determination and smooth trajectory transition is proposed, and a wheel-leg cooperative control strategy for whole-body stability constraint is constructed. The experimental results show that the terrain recognition accuracy of the proposed method reaches 97.4%, and the reasoning time is 18.7 ms. The comprehensive passing rate on four types of complex terrain reaches 94.6%, and the average passing time is 12.8 s. Under the disturbance condition, the maximum attitude deviation is controlled within 6.4°, the recovery time is shortened to 1.8 s, and the task completion rate reaches 95.4%. The results show that the proposed method can effectively improve the continuous passing ability and operation robustness of the wheel-legged robot in complex terrain, and has practical significance for promoting the intelligent development of autonomous mobile equipment in complex environments.
In response to the problem of strong subjectivity and insufficient process evidence in the evaluation of interactive quality in English classrooms, this paper constructs a multimodal intelligent evaluation model that integrates video, audio, and classroom transcribed text. Based on 48 real English classes and 1920 interactive segments, establish a five dimensional annotation system for questioning quality, feedback effectiveness, participation breadth, emotional atmosphere, and target language interaction density. The results showed that the intra class correlation coefficient (ICC) was 0.86. The macro average F1 value (Macro-F1) of the proposed model on the test set is 0.803, the mean absolute error (MAE) is 0.298, and the correlation coefficient between classroom level prediction and expert rating is 0.861. The ablation and case analysis show that there are differences in the dependence of different interaction dimensions on text, audio, and visual modalities. The model can also identify clues such as open questioning, waiting time, and participation coverage, providing interpretable basis for teachers to improve classroom interaction.
