Aiming at the difficulty of tracing Internet rumors, a Node Feature-Enhanced Traceability Model (NFETM) is proposed based on information carrier model and in-depth mining of user characteristics. This paper aims to use deep learning method to obtain high-order multi-scale features of information nodes (high-order neighbors, neighbor states, different state connection structures), and combine SEIR propagation mechanism to learn node states into information sources (I states) and non-information sources (S, E, R states). Firstly, multiple node centrality indexes are used to expand and enrich node characteristics. Secondly, an anti-noise enhancement module is used to reconstruct the expanded node features, and dynamically learn the features of the node itself and its first-order neighbors. Thirdly, the metric learning method is used to adjust the node feature space, so that the distance between nodes in the same state is reduced, so as to distinguish the categories and characteristics of nodes. Finally, the multi-dimensional features of nodes are fused and classified to determine the information source. The experimental results show that the proposed model achieves relatively good results in both the simulated generation network and the real network.

To reveal the transmission characteristics of respiratory infectious diseases clustering epidemic and explore the crucial role of network science in infectious disease control, we constructed and analyzed transmission networks for clustering epidemic, social relationship transmission networks, directed weighted bipartite networks between two gender-based age groups, and their corresponding null model networks based on structured post-epidemiological investigation data. The results indicate that by extracting key indicators from epidemiological investigations, constructing transmission networks, and analyzing them, it is possible to accurately focus on epidemiological characteristics and understand the infection risk among different populations. The application of network science has the potential to enhance our understanding of and response to the risk challenges posed by emerging infectious diseases.

The current research on supply chain risk propagation has limitations in the description of enterprise status setting and the relationship between enterprises. A cellular space is mapped to represent the supply chain network, fully considering the differences in operating capabilities of node enterprises under interruption risks and the mutual influence of neighboring cells. The improved risk transmission rule and a simulation model based on cellular automata and SEIRD infectious disease models are constructed. According to the results of multi-scenario simulation and research expansion, the potential reasons for the advancement of supply chain network disruption risk propagation process are analyzed, as well as the influence of market heat, market competition degree, risk early warning and control, and government macro-control on the propagation trend. The research conclusions can provide strategic reference for supply chain disruption risk management decision-making.

In order to explore the impact of intelligent inventory decisions trained by deep reinforcement learning algorithms on the dynamic evolution of supply chain systems, this paper considers the perspective of real-world decision makers and utilizes system dynamics modeling to reproduce the logical structure of a four-order supply chain model constructed using deep reinforcement learning. The decision results are visualized to assess their impact on the system. The experiments illustrate that the algorithm can make better ordering decisions based on the setting of its objective function, but it fails to achieve the lowest cost for the members who apply the algorithm synchronously. The evolutionary process reveals that the Sterman strategy has the role of maintaining the stability of the system during dynamic evolution; establishing a reasonable number of iterations helps to obtain a lower total supply chain cost.

Magnetronic memristor is introduced into the Morris-Lecar neural system to describe the induced current induced by magnetic flux, and the effect of memristor parameters on the dynamical characteristics of the system is explored. Based on the relationship between charge and magnetic flux, a memristive Morris-Lecar neuron model is established, the discharge patterns of the system are numerically simulated, and the synchronization transition process of the coupled system is researched by using the similarity function, and it is concluded that the strength of the electric coupling and the memristor parameters have a modulating effect on the synchronization state of the system. An asymmetric locally coupled memristive neural network model is constructed, and the effect of coupling strength on the existence of neural network embedded states is researched by using spatio-temporal dynamical diagrams, which further reveals the intrinsic mechanism of the information encoding and transmission process of complex neural network systems.

For constructing a five-dimensional chaotic system with complex chaotic properties and improving the security of image data protection, this paper adopts a cubic nonlinear type magnetronic memristor as the feedback term of the system, combines a four-dimensional chaotic system with the memristor and adds a fixed constant term and a linear feedback controller to obtain a five-dimensional magnetronic memristor chaotic system, and the system generates a coexisting hidden attractor under the influence of a parameter, and it can observe the Limit rings, hidden attractors, weak chaos and other types of attractors are observed. A chaotic image encryption algorithm using chaotic sequence encryption is proposed in combination with the disruption-diffusion algorithm, and the experimental results show that the algorithm has high security.

To further investigate the regulatory role of coupled time-delay on the modes of discharge activity in a half-center oscillator (HCO), based on the Hindmash-Rose (HR) neuronal model, the DHCO (delayed HCO) model with bursting behavior is constructed. By calculating phase difference between clusters of the bursting, dynamical evolution of the DHCO nervous system is studied with time-delay regulation. The DHCO system presents in-phase and anti-phase bursting oscillations in different parameter spaces. The synchronization transition of the bursting is determined under time-delay controlling. The results show that the DHCO model, regarded as the functional unit of the CPG (Central pattern generator) system can generate and switch multiple locomotion gaits by adjusting time-delay.

The practice of elderly service ecological cannot be separated from the co-creation and integration of multiple subjects. Based on the theory of service ecosystem, we constructed a three-party evolutionary game model of "elderly users-digital support platform-senior care service enterprises" and introduced the government intervention, and deduced and verified the dynamic evolution process and influencing factors of the stabilization strategies of each participating subject. The study shows that the government, as an important regulator of the elderly service ecosystem, can help the overall smart elderly service ecosystem to move towards a stable and win-win situation through the creation of a high-quality business environment, the introduction of digital technology and equipment, the implementation of reasonable rewards and penalties, the reputational evaluation system, and the practice of a dynamic regulatory network.

To overcome the problem of insufficient feature extraction capabilities of traditional target recognition algorithms in complex scenarios, a new target recognition algorithm based on a hybrid convolutional neural network is proposed. The core of the algorithm lies in combining the learning capability of non-Euclidean domains with traditional convolutional neural networks to enhance the depth and breadth of feature representation. The algorithm in this paper can extract and strengthen the key feature information in target recognition, and significantly improve the accuracy and robustness of recognition.

In order to discover new chaotic systems, the generalized Lorenz-Stenflo chaotic system is constructed based on the Lorenz-Stenflo chaotic system by adding the disturbance parameters. The basic chaotic characteristics of this system are analyzed by means of the dissipation, equilibrium point and stability, bifurcation diagram and Lyapunov exponential spectrum. Based on Lyapunov stability theory, the analytical expression of the global exponential attractive set of the system is given. The global exponential synchronization is achieved by adding linear feedback control to two generalized Lorenz-Stenflo systems by using the estimator of system solution boundness. Finally, the numerical simulation of the synchronization process is carried out, and the computer simulation results confirm the theoretical feasibility of global exponential synchronization.

In order to enrich the chaotic system model, a four-dimensional hyperchaotic system is constructed based on the Chen chaotic system. By analyzing the dissipation, equilibrium points, stability, sensitivity to initial values, bifurcation diagram, Lyapunov exponent spectra, LE dimension, power spectra and poincaré mapping of the new chaotic system, the rich and complex dynamic characteristics of the new chaotic system are revealed. In order to further enhance the security of information in secure communication, the new chaotic system and Qi hyperchaotic system are synchronized by using the transform modified function projective synchronization method, and a synchronization controller is designed to realize fast synchronization between hyperchaotic systems with different structures. The theoretical analysis and simulation results are consistent. Compared with the adaptive synchronous controller, the time required for the drive system and the response system to achieve synchronization is obviously shortened, which reflects the superiority of the method proposed in this paper.

A voice encryption algorithm based on chaotic sequences and dynamic Joseph rings is proposed. The digitized voice data is used as input to the SHA-256 algorithm to generate a Hash value. The dynamic key and the external input key are combined to generate the initial value of the chaotic system, and the chaotic sequence key stream is generated by the system iteration. The chaotic sequence key stream is used to replace the step size parameter of the Joseph ring. The dynamic Joseph ring with variable step size is used to generate a scrambling matrix. The scrambled data and the chaotic sequence key stream are used for XOR operation to realize the diffusion encryption of voice information. The numerical simulation and security analysis verified the feasibility and security of the voice encryption algorithm.

In order to explore the trust evolution of infrastructures and effectively prevent the online public opinion crisis in extreme climate emergencies. Selecting the “July 20” extreme heavy rainstorm in Zhengzhou, China as the case. Public opinions of Sponge City projects on Sina Weibo were captured. Content analysis, sentiment analysis, and curve fitting method were conducted to identify trust dimensions, users’ behaviors, and evolution patterns. The results indicated that the crisis began with the opinions of integrity-based trust damage from media practitioners, then broke out with opinions of compound-dimensions trust in the guidance of elites and media practitioners. Afterwards, it declined in the authoritative explanation of news organization. In the end, the public opinion crisis subsided with the continuous but low-impact opinions of trust damage by individuals.

To study the impact of government subsidies on different recycling channels in the new energy vehicle remanufacturing supply chain, a multi-objective model under dual recycling channels is established, an improved robust optimization method is used to solve the problem of uncertainty in demand and recycling volume during recycling, and a convolutional neural network (Conv-GLU network) method is proposed to solve the model. By comparing the performance of online and offline recycling channels, joint recycling channels and recycling channels under government intervention, the multi-objective optimization under government intervention is optimal. Therefore, the government can reasonably intervene in recycling under the background of big data to help new energy vehicle enterprises establish a dual recycling channel remanufacturing green supply chain.

This paper investigates the output consensus control problem of multi-agent systems with unmeasurable states under performance constraints. Firstly, the fuzzy logic system is used to approximate the unknown nonlinear function existed in the controlled system, and the adaptive fuzzy observer is introduced to estimate the unmeasurable state. Secondly, a filter whose filtering error can be ensured to converge in a fixed time is designed, which effectively avoids the “complexity explosion” problem in the distributed controller design process. In order to further realize better transient and steady-state performance of the system, a finite-time prescribed performance function that does not depend on the initial conditions of the error is designed. By designing an appropriate barrier function, the relationship between the error performance constrained system and the unconstrained system is established. Combining backstepping and fixed-time dynamic surface techniques, a consensus control scheme with finite-time prescribed performance is proposed. This scheme not only ensures that the output of each follower and the output of the leader finally reach synchronization, but also the consensus error converges to the specified constraint range in a finite time. Finally, the effectiveness of the proposed method is verified by numerical simulation.

In order to explore the influence of driver visual angle on manual car-following decision-making, a car-following model considering the expected visual angle is proposed based on the visual angle model. Considering the vehicle width, speed and driver risk perception, the expected visual angle function is constructed, and the improved VA model is established by using the expected visual angle as the direct feedback control term. Newton iteration method is used to solve the condition of linear stability of traffic flow. Numerical simulation and parameter calibration verify the improvement effect of the model. The results show that the traffic flow stability of the model in this paper is higher than that of the VA model. Considering the expected visual angle and collision risk perception is beneficial to enhance the stability and safety of traffic flow and improve the fitting accuracy of the model.

In response to the issue of credit chaos during credit repair implementation, this study constructs a tripartite stochastic evolutionary game model which involves the government, credit agencies, and the public. Leveraging Itô stochastic differential equation theory, the stability of behavioral strategies for the subjects is analyzed, and combined with numerical simulation to analyze the influence of key variables on the dynamic evolution of game players. The research results reveal that government guidance plays a prominent role in poor social credit environments. In a favorable social credit environment, the government should implement phased and graded regulation to avoid “regulatory capture”. As public recognition and government regulation improve, the effectiveness of governance experiences significant enhancement. Moreover, credit agencies display greater sensitivity to reward parameters than penalty parameters, necessitating the implementation of reasonable incentive and punitive measures by the government. Finally, relevant suggestions are made based on the research findings, which provide effective ideas for the governance of credit chaos and the healthy development of the social credit environment.

Aiming at the problem of conflicting lane changing behaviors of intelligent connected vehicles, a vehicle lane changing decision model based on the game theory is proposed. Firstly, the scenarios of major conflicts occurring in the process of vehicle lane changing are classified, and the lane changing game model is proposed. Then the time difference to collision (TDTC) is introduced, and the speed gains, safety gains and efficiency gains are established as the game payoff function. Besides, the decision-making mechanism is established to improve the efficiency of vehicle lane-changing decision-making. Simulation experiments and comparative analysis of the safety and feasibility of two decision models in conflict situations are conducted using MATLAB software, which includes the proposed lane-changing game decision-making model and the mixed-strategy gaming decision making model. The results show that the proposed lane-changing game decision-making model has higher decision-making efficiency and safety.

To study the strategic choices of PAID’s (Poverty-Alleviation-based Industry Development) participants in poverty-stricken areas, this paper constructs a tripartite evolutionary game model of poverty alleviation enterprises, impoverished households, and local governments, based on the assumption of bounded rationality. Combining with practical cases, it explores the evolutionary stability strategies of participating entities in different contexts. The result shows that there exists a unique stable equilibrium solution in a tripartite game system, and when the expected returns of the enterprise exceed a certain threshold, it can stimulate the enthusiasm of the participating entities. The proportion of labor force in poverty-stricken areas reflects the key driving role of the implicit poverty alleviation chain. The establishment of a reasonable reward and punishment mechanism by the government helps to promote the evolution speed of enterprise strategy selection. The government’s establishment of a reasonable reward and punishment mechanism helps to promote the evolution speed of enterprise strategy choices, and the three parties work together to promote the sustainability of PAID.

In this paper, the infinite horizon linear quadratic (LQ) optimal control problem for mean-field system with unknown system information is solved by using a completely model-free reinforcement learning (RL) approach. Although the introduction of the mean-field terms in system dynamics and the cost function will destroy the adaptiveness of the control law, the optimal stabilization control is successfully obtained based on the proposed RL algorithm and the Least Squares Temporal Difference estimation. In addition, combined with the idea of introducing off-policy learning, the control policy is further improved. We also prove that the algorithm produces stable policies given that the estimation errors remain small.