Behavior Modeling Method Research Articles

Pedestrian safety behavior modeling method in human-vehicle-intelligence shared road space

In the context of the development of autonomous vehicles sharing road resources with traditional traffic participants in the future, modeling the safe behavior of pedestrians in the shared road space between people, vehicles and intelligence is crucial for the simulation development and reliability and safety testing of autonomous vehicles. To meet this demand, the safe movement behavior of pedestrians is first analyzed, and on this basis, the traditional pedestrian social force model is improved to construct a momentum model for the interactive movement of pedestrians with other pedestrians, traditional vehicles and autonomous vehicles in the shared road space; then, the safety parameters of the model are calibrated using a data set of human-vehicle interactive movement and a genetic algorithm, and a typical shared road space scenario is built to verify the effectiveness of the model. The simulation results show that the model can simulate the real safe movement behavior of pedestrians, and has better simulation effects when used in tasks such as human-vehicle-intelligence shared road space simulation.

Reinforcement learning for occupant behavior modeling in public buildings: Why, what and how?

Effective control of energy consumption in public buildings holds paramount significance for global sustainable development. However, uncertainty in occupant behavior during operational phase often leads to a substantial discrepancy between the designed and actual energy consumption. While there is existing research on occupant behavior in public buildings utilizing stochastic modeling, statistical modeling, data mining, and agent-based modeling methods, the connotations and formation mechanisms have not been systematically revealed, rendering these developed models weak in applicability and reliability in practical scenarios. Reinforcement learning, which can perceive aspects of external environment and learn from stochastic interactions spontaneously, has been distinguished as a promising method for occupant behavior modeling. This research aims to develop a paradigm for reinforcement learning application in occupant behavior modeling in public buildings. Semi-structured interviews and literature review were combined to collect relevant information, while reasons for using reinforcement learning and applicable algorithms were discerned through qualitative analysis. Finally, detailed frameworks were proposed and validated through focus groups, offering a comprehensive guide that will serve as a reference for future research. This includes steps for collecting occupant-related data, selecting appropriate algorithms, and training and deploying reinforcement learning agents effectively. This research innovatively proposed a “panorama” of reinforcement learning for occupant behavior modeling in public buildings, laying a solid foundation for indoor environment and energy efficiency improvement in public building operations.

A Function-Based Behavioral Modeling Method for Air Combat Simulation

A Function-Based Behavioral Modeling Method for Air Combat Simulation

DC‐bias and temperature included CSWPL model for RF power transistors

AbstractA novel frequency domain behavioral modeling method for gallium‐nitride (GaN) devices is presented in this article. By utilizing a multi‐dimensional polynomial function, the proposed technique interpolates DC‐bias voltage and temperature based on the Canonical section‐wise piecewise linear (CSWPL) model framework. A detailed description of the model's theory is provided. With data from 10‐W GaN devices, the model was implemented in commercial software and validated through both DC and radio frequency (RF) tests. A robust predictive ability is demonstrated by the obtained results, thus proving the accuracy of the developed modeling method. This model is superior to the standard CSWPL model in that it is capable of predicting transistor behavior at different bias voltages and temperatures using a single set of parameters, thereby greatly reducing the complexity of the model and the time required for extraction.

Development of a Modified Generalized Prandtl–Ishlinskii Modeling and Compensation Method for Complex Hysteresis Behaviors of the Flexible Ureteroscope

Development of a Modified Generalized Prandtl–Ishlinskii Modeling and Compensation Method for Complex Hysteresis Behaviors of the Flexible Ureteroscope

Behavioral Modeling of Collaborative Problem Solving in Multiplayer Virtual Reality Manufacturing Simulation Games

Abstract Engineering is an inherently creative and collaborative endeavor to solve real-world problems, in which collaborative problem solving (CPS) is considered one of the most critical professional skills. Hands-on practices and assessment methods are essential to promote deeper learning and foster the development of professional skills. However, most existing approaches are based on out-of-process procedures such as surveys, tests, or interviews that measure the effectiveness of learning activity in an aggregated way. It is desirable to quantify CPS dynamics during the learning process. Advancements in virtual reality (VR) provide great opportunities to realize digital learning environments to facilitate a learning-by-doing curriculum. In addition, sensors in VR systems allow us to collect in-process user behavioral data. This paper presents a multiplayer VR manufacturing simulation game for virtual hands-on learning experiences, as well as a behavioral modeling method for monitoring the CPS skills of participants. First, we developed the Virtual Learning Factory, where users play simulation games of various manufacturing paradigms. Second, we collected action logs from a sample of participants and used the same pattern to generate more data. Third, the behavioral data are modeled as dynamic networks for each player. Last, network features are calculated, and a CPS scoring method is driven from them. Experimental results show that the proposed behavioral modeling successfully captures different patterns of CPS dynamics according to manufacturing paradigms and individuals. This detailed assessment contributes to the development of appropriate student-specific interventions to improve learning outcomes.

Behavioral Modeling Method of Macromodels for Interconnected Systems With Frequency Characteristics and Nonlinear Termination Networks

This study presents a behavioral macromodeling method for electronic or electrical interconnected systems, when there are frequency-characterized interconnects loaded by nonlinear terminations in systems. Analog behavioral modeling (ABM) was leveraged to capture frequency-dependent characteristics of multiconductor interconnects and properties of nonlinear terminations. Black-box macromodels were developed from the external mathematical behavior of parts, and real systems could be simulated in the SPICE circuit solver. Also, several critical points and constraints concerning ABM models were analyzed and proposed.From the validation results, the relative errors of proposed models are all less than 3%, and the execution time reduces by a factor over 10 concerning the existing interconnect model, which implies the proposed interconnect model can significantly improve the simulation efficiency, and the analysis accuracy can be guaranteed using proposed models of interconnect systems.

Multi-Output Recurrent Neural Network Behavioral Model for Digital Predistortion of RF Power Amplifiers

In this letter, we propose a method for behavioral modeling and digital predistortion (DPD) of RF power amplifiers (PAs) based on multi-output recurrent neural networks (RNNs). RNN has high modeling accuracy, but it also has high running complexity due to the recurrent mechanism. For this reason, we propose a multi-output model architecture, which means that the DPD model produces multiple adjacent outputs simultaneously for a single input sample group. This approach greatly reduces the running complexity of DPD based on RNNs with essentially no deterioration in performance. The proposed multi-output mechanism is applied to both long short-term memory (LSTM) and gate recurrent unit (GRU), and excellent linearization performances are maintained.

Agent behavior modeling method based on reinforcement learning and human in the loop

Computer generated force (CGF) is one of the increasingly important research topics in the field of simulation. However, low modeling efficiency and lack of adaptability are acute problems of traditional CGF modeling. In this study, a method for modeling the agent behavior based on reinforcement learning and human in the loop is proposed to improve the ability and efficiency of agent behavior modeling. First, an overall framework for modeling the behavior of intelligent agents is constructed based on the deep reinforcement learning algorithm Soft Actor Critic (SAC) framework. Second, in order to overcome the slow convergence speed of the SAC framework, a method for human interaction and value evaluation in the loop is introduced, and the specific algorithm flow is designed. Third, in order to verify the performance of the proposed method, experiments are conducted and compared with algorithms using a pure SAC framework based on an example of agent completing specific tasks. Result shows that after 100 episodes of training, the task completion rate of the agent can approach 100% while a pure SAC framework require at least 500 episodes of training to gradually improve the completion rate. Finally, the results demonstrate that the proposed method can significantly improve the efficiency of agent behavior modeling and the task completion rate increases with the number of human interventions in the loop.

Nonlinear behavior mechanism and modeling method of wound SMD inductor

AbstractIn the actual use process, the characteristics of the wire‐wound power inductor will change. On the one hand, the DC bias will fluctuate because the working current of the wire‐wound power inductor is not constant; on the other hand, there is a variable magnetic field, which, according to the principle of electromagnetic, affects structural size, core permeability, and flux density, both of which affect the amount of inductance. As a result, this research examines the wire‐wound SMD inductor's nonlinear behavior and modeling approach. Under the condition of zero bias at normal temperature, the static model is determined by simulating Y parameters and comparing its main performance parameters. The influence factors of temperature and DC bias are studied. The nonlinear model circuit is written in Verilog‐A format, and the inductor‐current curve is fitted by polynomial. The dynamic model is verified by ADS and compared with the test data. SWPA series power inductor measurement data from Sunlord is used to verify the validation of the model proposed in this paper. The Y parameter and performance index under zero DC bias at room temperature, inductance characteristics under different currents and inductance characteristics under different temperatures were compared respectively. From the comparison results, it can be concluded that, The fitting accuracy of the model is good in the frequency range of 100 KHz ~ 100 MHz, DC bias range of 0 ~ 0.28 A, and different temperatures (−40, 25, 125°C).

Environment-sensitive crowd behavior modeling method based on reinforcement learning

Environment-sensitive crowd behavior modeling method based on reinforcement learning

Low-Cost and Highly Accurate Behavioral Modeling of Antenna Structures by Means of Knowledge-Based Domain-Constrained Deep Learning Surrogates

The awareness and practical benefits of behavioral modeling methods have been steadily growing in the antenna engineering community over the last decade or so. Undoubtedly, the most important advantage thereof is a possibility of a dramatic reduction of computational expenses associated with computer-aided design procedures, especially those relying on full-wave electromagnetic (EM) simulations. In particular, the employment of fast replacement models (surrogates) allows for repetitive evaluations of the antenna structure at negligible cost, thereby accelerating processes such as parametric optimization, multi-criterial design, or uncertainty quantification. Notwithstanding, a construction of reliable data-driven surrogates is seriously hindered by the curse of dimensionality and the need for covering broad ranges of geometry/material parameters, which is imperative from the perspective of design utility. A recently proposed constrained modeling approach with knowledge-based stochastic determination of the model domain addresses this issue to a large extent and has been demonstrated to enable quasi-global modeling capability while maintaining a low setup cost. This work introduces a novel technique that capitalizes on the domain confinement paradigm and incorporates deep-learning-based regression modeling to facilitate handling of highly-nonlinear antenna characteristics. The presented framework is demonstrated using three microstrip antennas and favorably compared to several state-of-the-art techniques. The predictive power of our models reaches remarkable 2% of a relative rms error (averaged over the considered antenna structures), which is a significant improvement over all benchmark methods.

GLOBEM

There is a growing body of research revealing that longitudinal passive sensing data from smartphones and wearable devices can capture daily behavior signals for human behavior modeling, such as depression detection. Most prior studies build and evaluate machine learning models using data collected from a single population. However, to ensure that a behavior model can work for a larger group of users, its generalizability needs to be verified on multiple datasets from different populations. We present the first work evaluating cross-dataset generalizability of longitudinal behavior models, using depression detection as an application. We collect multiple longitudinal passive mobile sensing datasets with over 500 users from two institutes over a two-year span, leading to four institute-year datasets. Using the datasets, we closely re-implement and evaluated nine prior depression detection algorithms. Our experiment reveals the lack of model generalizability of these methods. We also implement eight recently popular domain generalization algorithms from the machine learning community. Our results indicate that these methods also do not generalize well on our datasets, with barely any advantage over the naive baseline of guessing the majority. We then present two new algorithms with better generalizability. Our new algorithm, Reorder, significantly and consistently outperforms existing methods on most cross-dataset generalization setups. However, the overall advantage is incremental and still has great room for improvement. Our analysis reveals that the individual differences (both within and between populations) may play the most important role in the cross-dataset generalization challenge. Finally, we provide an open-source benchmark platform GLOBEM- short for Generalization of Longitudinal BEhavior Modeling - to consolidate all 19 algorithms. GLOBEM can support researchers in using, developing, and evaluating different longitudinal behavior modeling methods. We call for researchers' attention to model generalizability evaluation for future longitudinal human behavior modeling studies.

A Low Running Complexity Model for Digital Predistortion of RF Power Amplifiers

In this letter, two basis function multiplexing-based behavioral modeling methods for digital predistortion (DPD) of RF power amplifiers (PAs) are proposed to reduce the running complexity of DPD. The proposed full basis-propagating selection (FBPS) model and reduced-complexity FBPS (RC-FBPS) model give two reasonable ways to multiplex even-order basis functions, extending the basis-propagating selection (BAPS) model which only uses basis function delay and odd-order basis functions. The experimental results confirm that both the proposed FBPS and RC-FBPS models can achieve a good tradeoff between running complexity and performance.

Improved EMI Behavioral Modeling Method of Three-Phase Inverter Based on the Noise-Source Phase Alignment

The electromagnetic interference (EMI) behavioral model (BM), including active one-port networks, is widely used to guide the EMI filter design. To build BM, the series and shunt impedance insertion (SSII) based modeling method is normally adopted. During the SSII-based modeling, the time-domain measurements under different impedance insertion cases are introduced. However, the noise-source phases are time-varying, which means that the noise-source phases are different under multiple measurements. The different noise-source phases cause the errors of BM. To address this shortcoming, this article investigates the effect of the noise-source phases on BM for the first time. The improved BM method based on the noise-source phase alignment is proposed. The improved BM method effectively improves the EMI prediction accuracy of BM with an error of 6 dB or less. The effectiveness of the improved modeling method for the conducted emissions estimation is verified in the three-phase inverter test prototype.

Improved behavioral modeling using augmented LSTM networks for ultra‐broadband mmWave PA

AbstractIn this article, a new behavioral modeling method based on augmented long‐short term memory (ALSTM) networks for ultra‐broadband millimeter‐wave power amplifiers (mmWave PAs), is proposed. The basic theory and modeling procedure of this technique are provided. Different optimization cores of the models are tested, and the optimal algorithm is chosen. Tradeoffs have been made to fix the optimal topology for the proposed modeling technique. To validate the proposed method, a 4‐carrier 320 MHz modulated signal was employed to excite a mmWave PA with the center frequency located at 41 GHz. Experimental results show that the proposed ALSTM model has better predictive capability when compared with the traditional and other existing machine learning modeling techniques.

Reducing Power Consumption of Digital Predistortion for RF Power Amplifiers Using Real-Time Model Switching

In this article, we propose a new behavioral modeling method to reduce the running complexity and power consumption of digital predistortion (DPD) models for radio frequency power amplifiers. By employing the proposed method, different cross terms in a DPD model can be switched dynamically in real time. Each cross-term branch can further choose the model coefficients from multiple coefficient sets, which improves the DPD performance with little extra complexity. The switch of both cross-term branches and model coefficients is realized using a decision tree-based switch controller, leading to very low run-time complexity. By optimizing the selection process, different input samples can choose the most suitable model configuration that contributes most to the linearization performance. As only one branch is activated at a time, the power consumption can be greatly reduced. Based on the experimental results, the proposed method can achieve excellent linearization performance with significantly reduced power consumption.

Digital Predistortion of Nonlinear Power Amplifier Based on Improved Spline Adaptive Filtering Method

To improve the modeling accuracy and digital predistortion compensation effect, this paper proposed a kind of polynomial based spline adaptive filter (SAF) power amplifier (PA) behavioral modeling method. The proposed model can solve the shortcomings of the traditional memory polynomial model (MP), such as higher nonlinear order, highly coefficient calculation difficulty, poor hardware stability, low modeling accuracy and poor digital predistortion compensation. Meanwhile, the proposed SAF based model can avoid the pseudo-inversion operation of the matrix when calculating the coefficients through the indirect learning structure, which can improve the compensation effect of digital predistortion. In order to compare the performance of the different models, the modeling processes of original filtering model, MP model and dynamic variation reduction (Dynamic Deviation Reduction, DDR) model based on SAF method (OSAF, MPSAF, DDRSAF) are presented respectively. The proposed models are verified by a class F PA, which are stimulated by a set of measured input and output data of an LTE signal with 20 MHz bandwidth. The measured results shows that compared with the conventional MP model, the proposed DDRSAF model presents the best model performance and improves the modeling accuracy by 6.6dB while maintaining the similar computational complexity, and the ACPR compensation effect increases by 19.58dB and 20.47dB respectively, which have the important significance for practical engineering application.

Analysis of insider threat detection methods tested on CERT dataset

Applying artificial intelligence for insider threat detection faced a number of challenges: lack of data, user behavior modeling, etc. Meanwhile, there are open datasets for research and there are effective approaches of insider threat detection. This work considers studies that utilizes one of the popular datasets — CERT. A comparative analysis contains features used for user behavior modeling and insider threat detection methods.

Numerical Modeling of the Pipeline Uplift Mechanism in Clay

Abstract Buried pipelines are a vital infrastructure and are mainly used to transport energy carriers and other essential products. The pipes are generally buried in the upper layer of soil deposits and, therefore, are highly affected by different geo-environmental conditions. The various pathological cases recorded in the world are caused by the degradation of structures in contact with swelling soils, the fact that necessitates a full understanding and investigation of such a phenomenon. This paper presents a method for the pipeline behavior modeling based on the finite element analysis by using PLAXIS 3D software, aimed at the determination of the pipe bending moment, displacement over its length, and the evaluation of vertical stresses in soil under the pipe. A parametric study has been carried out to investigate the effect of the pipe burial depth and the soil cohesion. The finite-element results have been compared with experimental data from the literature. It was found that, unlike laboratory models, the numerical analysis can account for the internal pressure in the pipe and the depth of the pipe burial. The finite-element analysis showed that the presence of fluid pressure inside the pipe results in a decrease in the maximum swelling of the soil by about 95%. The displacement of the pipe is considerably affected by the burial depth. The vertical stress at one end of the pipe can be greater than that at the other end in the case of a pipe under internal pressure, while in the case of an empty pipe, the values are very close at both ends. The numerical analysis shows that an increase in the pipe internal pressure leads to a decrease in its vertical displacement.