International Journal of Modeling, Simulation, and Scientific Computing

This paper aims to comprehensively explore the pivotal role of simulation and modeling in the field of Artificial Intelligence (AI). It focuses on elucidating the diverse applications of simulation and modeling in training AI systems, optimizing algorithms, and enhancing decision-making processes. To achieve this objective, we conducted an extensive review of the literature from the Scopus database, employing a well-defined selection process. We utilized keywords such as “simulation,” “modeling,” “Artificial Intelligence,” and related terms to identify relevant papers published within the last 10 years. The selection criteria included assessing the relevance, quality, contribution, and recent citations of the papers. After a rigorous screening process, we selected 40 papers with the highest overall scores for inclusion in our review. The selected papers encompass a wide range of domains where simulation and modeling play a vital role in advancing AI applications. These domains include manufacturing, healthcare, energy consumption prediction, public sector decision-making, education, environmental modeling, and more. Our review highlights how AI leverages simulation and modeling to improve predictive accuracy, optimize resource allocation, and enhance decision-making processes across diverse sectors. We also discuss the potential future directions in the integration of simulation and modeling with AI, emphasizing its significance in various fields.

For the charged-particle dynamics under a strong and constant magnetic field, we consider and analyze exponential integrators in this paper. We derive two kinds of exponential integrators and study their long-time behavior by using the technique of modulated Fourier expansions. It is shown that a symmetric two-stage exponential integrator and some one-stage symplectic exponential integrators approximately conserve the energy and magnetic moment of the charged-particle dynamics over long times.

The MeanShift algorithm is a nonparametric method based on gradient ascent and it can effectively handle complex variations in lychee orchard scenes as well as changes in lychee tree canopies due to its adaptability, multi-scale analysis capabilities, and robustness, making it widely used in the segmentation processing of drone-based remote sensing images of lychee orchards. However, due to the high computational complexity of the MeanShift algorithm, its performance in processing large-scale drone remote sensing images of lychee tree canopies is not highly efficient, leading to low segmentation efficiency, which hampers its broader application. To address these issues, this study proposes high-speed MeanShift parallel segmentation algorithms for drone remote sensing images of lychee tree canopies based on MapReduce and Spark distributed computing frameworks. In this study, a cluster consisting of four nodes with Hadoop and Spark was set up, and 4000 drone remote sensing images were used as test data to evaluate the algorithm. Experimental results show that, the MeanShift algorithm based on MapReduce reduced the task execution time by 86.1% compared to the traditional MeanShift algorithm, while the MeanShift algorithm based on Spark reduced the task execution time by 88.0%, without compromising segmentation accuracy. The MeanShift parallel segmentation algorithm based on Hadoop and Spark platform can overcome the bottleneck of task execution efficiency and significantly enhance computational speed on a single machine.

This paper studies the dynamics of interacting Tilapia fish and Pelican bird population in the Salton Sea. We assume that the diseases spread in Tilapia fish follows the Holling type II response function, and the interaction between Tilapia and Pelican follows the Beddington–DeAngelis response function. The dynamics of diffusive and delayed system are discussed separately. Analytically, all the feasible equilibria and their stability are discussed. The criterion for Turing instability is derived. Based on the normal form theory and center manifold arguments, the existence of stability criterion and the direction of Hopf bifurcation are obtained. Numerical simulation shows the occurrence Hopf bifurcation, double Hopf bifurcation and transcritical bifurcation scenarios. The snap shot shows the spot, spot-strip mix patterns in the whole domain. Further, the stability switching phenomena is observed in the delayed system. Our comprehensive study highlights the effect of different parameters, multiple time delay and extinction in Pelican populations.

Milk is considered to be one of the most perfect natural foods for human beings. The analysis of effective components and determination of important components in milk is an important practical work. Based on the previous measurement data of fat volume concentration and refractive index of milk in the market, this paper forms a fitting formula through mathematical fitting, which can solve the corresponding relationship between nonmeasurement points of milk concentration and refractive index. Because the actual measurement points are limited, its practical application is not convenient. However, fitting formulas can calculate the corresponding values of infinitely many points, which can be easily applied. However, judging from the goodness of fit of the fitting formula, a theoretically good fitting formula may produce underfitting and overfitting in practical application. Through the analysis of the first-order to sixth-order fitting formulas of milk concentration and refractive index, this paper tells us that the fitting formula with high goodness of fit is not necessarily a good formula. The specific problems should be analyzed in the application process to make the theory more consistent with the actual situation. From the calculation and analysis in this paper, it can be seen that the fitted sixth-order formula has the best effect when the fat volume concentration of milk is 0–10%. When considering a fat volume concentration of 0–34% in milk, the fitted second-order formula performs better.

A second-order time-stepping method for numerically solving the linearized time fractional KdV equation (TFKDVE) whose solution has initial singularity is investigated. Alikhanov’s scheme is used to discretize the Caputo fractional derivative on temporal graded mesh, and the space is approximated by a Petrov–Galerkin spectral method. Detailed stability and error estimate of the full-discrete scheme are given, and the final error bound is $\alpha$-robust. Numerical examples are given to illustrate the sharpness of the error analysis.

The Machine Learning (ML) models are prone to a curse of dimensionality. The dataset with a greater number of features involves more computational cost and it may lead to low performance in the context of prediction accuracy. Therefore, in this research work we have predicted diabetes with more accuracy by using a smaller number of features. The heuristic methods Sequential Forward Selection (SFS), Sequential Backward Selection (SBS) and metaheuristic evolutionary methods — Whale Optimization Algorithm (WOA) and Genetic Algorithm (GA) are used for performing feature subset selection. The Gini index is also used as a filter evaluator. The performance of the feature subsets is analyzed by applying three different types of ML models, Random Forest (RF), Multi-Layer Perceptron (MLP) and K-Nearest Neighbor (KNN). We have predicted type-2 diabetes with an accuracy of 96.82%. Also, we have reduced the number of features up to 67.44% i.e., identified 32.56% most relevant features.

This research paper aims to study the impact of COVID-19 on the economy, where we have extended the standard epidemiological model, the Susceptible–Infected–Recovered (SIR) model. In the economy’s dynamics, we have incorporated aggregate demand and aggregate supply and further set up a system of three-dimensional nonlinear differential equations, which is more likely the Labor–Infected–Quarantine (LIQ) model, an extension of the SIR model. Here, we have derived the steady-state condition and analyzed the model’s stability. In addition, we have done a comparative static analysis of parameters with some policy implications. We aim to achieve the twin objective, i.e., controlling the pandemic and reviving the economy under the set of the policy mix.

This research paper develops an algorithm to solve a neutrosophic linear fractional optimization problem, where the cost of the objective functions, the technology coefficients, and the resources are neutrosophic numbers (triangular and trapezoidal numbers). Two defuzzification methods are used here to convert the neutrosophic linear fractional optimization problem into a multi-objective linear fractional optimization problem (MOLFOP). The fractional objective function is defuzzified using the component-wise optimization method, and the linear constraints are defuzzified using the aggregation ranking function strategy. Hence, the MOLFOP is then solved using the fuzzy goal programming method. In order to further explain our suggested algorithm, two numerical examples are solved at the end, and the outcomes obtained by our method are compared with the outcomes obtained using the other algorithm. 

Modular robots carry the unique advantage of changing their shape according to different tasks and environments. In order to study the lightweight and flexible combination characteristics of modular robots, this paper presents a two-degree-of-freedom pluggable modular robot and its simulation method, named SnackBot. SnackBot is composed of several independently operating modules. Each module contains a control unit, a micro battery, and two DC motors that drive the module joints. In order to predict robot actions and monitor the state of the robot, the simulation model of SnackBot is built and a virtual environment is established in Ignition Gazebo. The virtual robot and physical robot are synchronously controlled in the environment of the Robot Operating System 2 (ROS2). Finally, experimental results verify the feasibility of the SnackBot.

This paper focuses on the online control of a class of nonlinear dynamical systems, specifically robotic manipulators. Solutions utilizing Proportional–Integral–Derivative (PID) control schemes are employed to control the joints of robotic manipulators. However, the existing control strategies utilize fixed gains, which do not fully account for the inherent nonlinearity of the dynamical structure or the dynamics of reference-tracking error. Additionally, the individual joint’s dynamic performance is optimized independently from the performance of other joints. This work introduces an adaptive integral Reinforcement Learning algorithm to control a four-DoF robotic arm in real time. This is done using a model-free Value Iteration process implemented in a continuous-time mode. The solution does not assume any knowledge of the dynamics of the robot arm and does not require any initial admissible control strategy to proceed with the adaptive learning solution. The self-learning algorithm provides adaptable strategies to control the turntable, forearm, bicep, and wrist joints of the robotic arm. The performance of the adaptive learning solution is compared with those of Proportional–Integral–Derivative and high-order model-free adaptive control schemes to highlight its effectiveness.

In order to solve the selection problems such as multiple products, complex scenarios, and difficulty in supporting complex solutions, the product selection push model for complex solutions was proposed, the selection cloud platform was set up, and the platform architecture, composition, and operating mechanism were given. Based on the digital twin five-dimensional framework, the modeling process of the product selection cloud platform was detailed, and in the demand stage, the function model of the model was determined by the collaborative filter algorithm, which gave a comprehensive description of its functions. Through the construction of multi-platform fusion and algorithm model, the active push, scheme automatic process, and verification function of solution product selection have been realized. The solution has been applied in the selection of the smart security community scheme, verifying the effectiveness and practicality of the platform.

The diversity, infinity, and nonuniform description of human motion make it challenging for computers to understand human activities. To explore and reuse captured human motion data, this work defines a more comprehensive hierarchical theoretical model of human motion and proposes a standard human posture encoding scheme. We construct a domain knowledge graph (DKG) named the human motion encoding knowledge graph (HME-KG) based on posture codes and action labels. Community detection, similarity analysis, and centrality analysis are used to explore the potential value of motion data. This paper conducts an evaluation and visualization of HME-KG.

In the digital era, social media platforms have become the focal point for public discourse, with a significant impact on shaping societal narratives. However, they are also rife with mis- and disinformation, which can rapidly disseminate and influence public opinion. This paper investigates the propagation of mis- and disinformation on X, a social media platform formerly known as Twitter. We employ a multidimensional analytical approach, integrating sentiment analysis, wavelet analysis, and network analysis to discern the patterns and intensity of misleading information waves. Sentiment analysis elucidates the emotional tone and subjective context within which information is framed. Wavelet analysis reveals the temporal dynamics and persistence of disinformation trends over time. Network analysis maps the intricate web of information flow, identifying key nodes and vectors of virality. The results offer a granular understanding of how false narratives are constructed and sustained within the digital ecosystem. This study contributes to the broader field of digital media literacy by highlighting the urgent need for robust analytical tools to navigate and neutralize the infodemic in the age of social media.

Biological experiments have confirmed that the locomotion system includes the nervous and musculoskeletal systems, and central pattern generator (CPG). The motor unit (MU) is the core part of the musculoskeletal system. Based on the biological experiments, it is confirmed that the CPG can generate and distribute the motor rhythms and patterns to different MUs through flexible weights, and the MUs feedback can affect the output of CPG. Then, the MU model, the CPG model, and their coupling model are investigated to explain the biological phenomena. Within this work, it was found that there is consistency between the shape of states generated by the three MUs in this work and the experimental results in previous studies. When the MUs correspond to the parts in limbs, and the CPG connects to the MUs via synapses, the locomotion pattern can be adjusted by CPG to use the variable synaptic weights. Besides, the model with fractional order (FO) has better efficiency than the model with integer order. Therefore, the model investigates the relationship between the CPG and the MU and provides a novel technique to introduce the FO coupling model for controlling the locomotion state.

Due to technical limitations and the characteristics of the launch power system propellant, its support capability is affected by many factors and links. In the process of mission support, it is necessary to consider both its quality characteristics and the safety during the support process, which bring difficulties to the accuracy of its support capability evaluation. In this regard, the evaluation indexes system is constructed according to the requirements of total quality management theory, process safety management theory and equipment support characteristics. In the process of setting the indexes system, the technique for order preference by similarity to an ideal solution method (TOPSIS method) is used to screen the indexes to make indexes setting more scientific. The subjective weights obtained by the order relation analysis method (G1 method) and the objective weights obtained by the entropy weight method (EW method) are combined by the game theory combination weighting method to obtain the indexes combination weights, which reduce the error of weight calculation. The two-dimensional cloud model is used to evaluate support capability of the launch power system propellant from the two dimensions of quality management capability and safety management capability, which can reflect the support capability more comprehensively and accurately. Through example verification, the closeness of support capability of the launch power system propellant to the evaluation grades of ‘excellent’, ‘good’, ‘average’, ‘poor’ and ‘very poor’ is calculated to be 0.2333, 5.1169, 0.3579, 0.1821 and 0.1041, respectively, indicating that the evaluation results are closer to good. Finally, compared with the existing one-dimensional cloud model evaluation method, the superiority of this research method is highlighted, which has a certain guiding role in improving its support capability.

This paper provides a comprehensive exploration of physics-informed neural networks and their core features. It delves into their role in tackling inverse problems inherent in ordinary differential equation-based models. Within this context, we introduce a two-group epidemiological model, elucidating its fundamental attributes. The central objective of this research is to accurately estimate the model parameters for both groups in the epidemiological model. We offer a detailed exposition of the adopted methodology, providing insights into the algorithm and the techniques employed for its implementation. Through this analysis, we illuminate the complexities of our study, contributing to the growing body of knowledge in this field, which intersects epidemiology and neural network-based parameter estimation for an enriched understanding of infectious disease dynamics.