Stochastic modeling of human learning behavior
Stochastic modeling of human learning behavior in manual control task
SEARCH · Search NASA
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Stochastic modeling of human learning behavior in manual control task
In a large enterprise it is difficult for cyber security analysts to know what services and roles every machine on the network is performing (e.g. file server, domain name server, email server). Understanding the roles of the systems in the network provides analysts with a situational awareness that will allow them to detect consequential changes in the network, initiate an incident response plan, and optimize their security posture. Using the network flow data, already collected by most enterprises, we developed a tool that enables analysts to automatically detect/classify services and roles of every machine that’s operating on a network (e.g. file server, domain name server, email server) for better situational awareness of potential threats to the network. his tool queries Elasticsearch for network flow data, creates a temporal behavior model of each system, uses unsupervised machine learning to cluster the models with a set of labeled temporal behavior models, and the resulting information can be printed to the console or programmatically accessed. The results include the likelihood that a machine has a labeled role and lists other machines that are most similar in behavior.
Numerous new flight deck configurations for caution, warning, and alerts can be conceived; yet testing them with human-in-the-Ioop experiments to evaluate each one would not be practical. New sensors, instruments, and displays are being put into cockpits every day and this is particularly true as we enter the dawn of the Next Generation Air Transportation System (NextGen). By modeling pilot behavior in a computer simulation, an unlimited number of unique caution, warning, and alert configurations can be evaluated 24/7 by a computer. These computer simulations can then identify the most promising candidate formats to further evaluate in higher fidelity, but more costly, Human-in-the-Ioop (HITL) simulations. Evaluations using batch simulations with human performance models saves time, money, and enables a broader consideration of possible caution, warning, and alerting configurations for future flight decks.
Monolithic plate-type fuel is a fuel form that is being developed for the conversion of high performance research and test reactors to low-enrichment uranium fuels. These fuel-plates are comprised of a high density, low enrichment, U-Mo alloy based fuel foil encapsulated in an aluminum cladding. To benchmark this new design, number of plates has been irradiated with satisfactory performance. As a part of continuing evaluation efforts, a set of plates covering range of operational parameters is scheduled to be tested during MP-1 irradiation experiments. It is necessary to evaluate the thermo-mechanical performance of plates during irradiation. For this, selected plates with distinct operational histories; covering low power, high power and high fission density were simulated. Fully coupled three-dimensional models of plates with a capability to evolve mechanical and thermal properties of constituent materials with irradiation time and burn-up were developed. The models input used projected parameters, including plate geometry, irradiation history and coolant conditions as input. The model output included temperature, displacement and stresses in the fuel, cladding and diffusion barrier. The fuel behavioral model considered inelastic behavior including volumetric swelling due to solid and gaseous products, irradiation induced creep, thermal expansion, conductivity degradation and plasticity. A visco-plastic behavioral model was used for the cladding that included thermal creep, irradiation hardening, growth due to fast neutrons and Mises plasticity. The plates were then simulated by using projected irradiation parameters. The resulting temperature, displacement and stress-strains were comparatively evaluated on the selected paths. The results were then compared with those of plates from previous RERTR experiments.
Monolithic plate-type fuel is a fuel form that is being developed for the conversion of high performance research and test reactors to low-enrichment uranium fuels. These fuel-plates are comprised of a high density, low enrichment, U-Mo alloy based fuel foil encapsulated in an aluminum cladding. To benchmark this new design, number of plates has been irradiated with satisfactory performance. As a part of continuing evaluation efforts, a set of plates covering range of operational parameters is scheduled to be tested during MP-1 irradiation experiments. It is necessary to evaluate the thermo-mechanical performance of plates during irradiation. For this, selected plates with distinct operational histories; covering low power, high power and high fission density were simulated. Fully coupled three-dimensional models of plates with a capability to evolve mechanical and thermal properties of constituent materials with irradiation time and burn-up were developed. The models input used projected parameters, including plate geometry, irradiation history and coolant conditions as input. The model output included temperature, displacement and stresses in the fuel, cladding and diffusion barrier. The fuel behavioral model considered inelastic behavior including volumetric swelling due to solid and gaseous products, irradiation induced creep, thermal expansion, conductivity degradation and plasticity. A visco-plastic behavioral model was used for the cladding that included thermal creep, irradiation hardening, growth due to fast neutrons and Mises plasticity. The plates were then simulated by using projected irradiation parameters. The resulting temperature, displacement and stress-strains were comparatively evaluated on the selected paths. The results were then compared with those of plates from previous RERTR experiments.
In the surfaces correlation study, several different volumetric and drainage measurement techniques for classifying surface texture were evaluated as part of a major study to develop and improve methods for predicting tire friction performance on all types of pavement. The objective of the evaluation was to seek relationships between the different techniques, and to relate those results to surface frictional characteristics. We needed to know how each of the tests could be related to each other. Another of my assigned projects was to make a tire behavior math model for the High Speed Civil Transport (HSCT) using the same methods used for the space shuttle a few years ago. A provided third order equation with two variables was used. This model will also be used for studies with the Boeing 777. Only a few changes will be necessary to adapt it for this other aircraft, which is the newest offered by Boeing. In my final project I was involved with testing the tires for this new aircraft using the Aircraft Landing Dynamics Facility (ALDF) test carriage within the carriage house at LaRC. A 50 inch diameter radial tire manufactured by Michelin Aircraft Tire Corporation had to be tested to double overload of 114,000 pounds. The rated load of each tire is 57,000 pounds, but Boeing required tests assuming failure of a companion tire that could have cost Michelin approximately $12 million to build a facility to provide the required test capability. Here at LaRC, only minimum modifications to the facility were required to perform this specific test.
Instrumented nanoindentation was used in this paper to investigate the hardness, elastic modulus, and creep behavior of an austenitic Fe-20Cr-25Ni model alloy at room temperature, with the indented grain orientation being the variant. The samples indented close to the {111} surfaces exhibited the highest hardness and modulus. However, nanoindentation creep tests showed the greatest tendency for creep in the {111} indented samples, compared with the samples indented close to the {001} and {101} surfaces. Scanning electron microscopy and cross-sectional transmission electron microscopy revealed slip bands and dislocations in all samples. The slip band patterns on the indented surfaces were influenced by the grain orientations. Deformation twinning was observed only under the {001} indented surfaces. Finally, microstructural analysis and molecular dynamics modeling correlated the anisotropic nanoindentation-creep behavior with the different dislocation substructures formed during indentation, which resulted from the dislocation reactions of certain active slip systems that are determined by the indented grain orientations.
The population is increasingly becoming tractable as more and more people carry handheld devices as part of their everyday activities. Recent studies have shown that handheld devices' generated traffic share is now more than 50% of total global online traffic. This has created an unprecedented opportunity for modeling human mobility behavior. For example, aggregate check-ins and dwell time can reveal building level occupancies. However, there are clear limits to accurate modeling (e.g. reproducible, repeatable, and realistic), unless we decipher the underlying reason causing typical mobility patterns. We know that human behavior is a reflection of a set of activities, such as going to the gym or work, and which can be seen as a catalyst for humans to move from one location to another. This work envisions the use of activity characterization for modeling human mobility by introducing a context that maps activities to certain mobility patterns. In the end, we highlight the efficacy of the proposed approach by analyzing the impact of public policies surrounding stay at home order on human mobility.
Brittle behavior of metal alloys is often critical to modeling ballistic impact and penetration. The ALEGRA multiphysics finite element software incorporates calibrated models for the equation of state, elasticity, yield stress, plasticity and fracture, but simulations do not always capture expected metal fracture. Here we report concerted efforts to do so for one important case where experiments clearly show shear fractures: a tungsten sphere impacting a steel plate at various angles. Our best simulations show fractures that are qualitatively similar to experiments, but there are significant differences in quantitative metrics. Specifically, velocities of tracers used to quantify simulated plug parameters consistently fall short of measured plug velocities. Also, simulated plugs break apart more than expected from experimental evidence. We attribute these shortfalls to the lack of an explicit shear fracture mechanism in the material models, leading to over-estimated resistance to plug formation and movement.
This modelization starts from the following hypotheses: pilot's behavior is a time discrete process, he can perform only one task at a time and his operating mode depends on the considered flight subphase. Pilot's behavior was observed using an electro oculometer and a simulator cockpit. A FORTRAN program has been elaborated using two strategies. The first one is a Markovian process in which the successive instrument readings are governed by a matrix of conditional probabilities. In the second one, strategy is an heuristic process and the concepts of mental load and performance are described. The results of the two aspects have been compared with simulation data.
Models from Coupled Model Intercomparison Project Phase 6 produce too much precipitation over the Andes but too little over the Amazon or the Wet Andes-Dry Amazon (WADA) bias pattern. Unlike the conventional view that convection parameterization and land model deficiencies can contribute to Amazonian rainfall biases, we approach this long-standing biased model behavior through the lens of Andean convection. Using Community Earth System Model v1.1 and focusing on the wet season, our mechanism-denial experiments demonstrate that Andean convection notably reduces precipitation over the Amazon during austral summer. The Andean forced Amazonian response operates on weather timescale. Furthermore, the reduction of Amazonian rainfall is detectable within a few hours after initial Andean forcing. The precipitation response is primarily driven by variations in the moisture budget and is moderated by changes in convective available potential energy over the Amazon. Changes in the total advection of moisture over the Amazon are dominated by the vertical advection term and can be attributed to discrepancies in the dynamic omega field. In the experiments, the Andean east flank region is scrutinized where the vertical velocity and moisture fields play an intermediary role for the Andean driven WADA connection. The Andean forcing induces descending anomalies on the Andean east flank. The disturbances of wind and geopotential fields over the Andean east flank propagate eastward via Kelvin waves. Over the Amazon, descending anomalies and advective drying lead to reduction of mid-to-high level cloud, increase of shortwave cloud forcing and surface net radiation, and enhancement of themodynamic stability and rainfall reduction.
Atomistic molecular dynamics (MD) and a microstructural dislocation density-based crystalline plasticity (DCP) framework were used together across time scales varying from picoseconds to nanoseconds and length scales spanning from angstroms to micrometers to model a buried copper–nickel interface subjected to high strain rates. The nucleation and evolution of defects, such as dislocations and stacking faults, as well as large inelastic strain accumulations and wave-induced stress reflections were physically represented in both approaches. Both methods showed similar qualitative behavior, such as defects originating along the impactor edges, a dominance of Shockley partial dislocations, and non-continuous dislocation distributions across the buried interface. The favorable comparison between methods justifies assumptions used in both, to model phenomena, such as the nucleation and interactions of single defects and partials with reflected tensile waves, based on MD predictions, which are consistent with the evolution of perfect and partial dislocation densities as predicted by DCP. This substantiates how the nanoscale as modeled by MD is representative of microstructural behavior as modeled by DCP.
Behavioral models are developed that closely reproduced pulsive control response of two pilots using markedly different control techniques while conducting a tracking task. An intriguing find was that the pilots appeared to: 1) produce a continuous, internally-generated stick signal that they integrated in time; 2) integrate the actual stick position; and 3) compare the two integrations to either issue or cease a pulse command. This suggests that the pilots utilized kinesthetic feedback in order to sense and integrate stick position, supporting the hypothesis that pilots can access and employ the proprioceptive inner feedback loop proposed by Hess's pilot Structural Model. A Pilot Cost Index was developed, whose elements include estimated workload, performance, and the degree to which the pilot employs kinesthetic feedback. Preliminary results suggest that a pilot's operating point (parameter values) may be based on control style and index minimization.
A supertanker is considered as a nonlinear system which responds very slowly to changes in the rudder position. Moreover this type of ship is often unstable in loaded condition. In order to model the helmsman's behavior, a number of tests were performed using a ship maneuvering simulator. The trained subjects had to steer a 200,000 tons tanker along a varying course. The results obtained from these trials are presented.
Human and social modeling has emerged as an important research area at Sandia National Laboratories due to its potential to improve national defense-related decision-making in the presence of uncertainty. To learn about which sensitivity analysis techniques are most suitable for models of human behavior, different promising methods were applied to an example model, tested, and compared. The example model simulates cognitive, behavioral, and social processes and interactions, and involves substantial nonlinearity, uncertainty, and variability. Results showed that some sensitivity analysis methods create similar results, and can thus be considered redundant. However, other methods, such as global methods that consider interactions between inputs, can generate insight not gained from traditional methods.
AI models are criticized as being black boxes, potentially subjecting climate science to greater uncertainty. Explainable artificial intelligence (XAI) has been proposed to probe AI models and increase trust. In this review and perspective paper, we suggest that, in addition to using XAI methods, AI researchers in climate science can learn from past successes in the development of physics-based dynamical climate models. Dynamical models are complex but have gained trust because their successes and failures can sometimes be attributed to specific components or sub-models, such as when model bias is explained by pointing to a particular parameterization. We propose three types of understanding as a basis to evaluate trust in dynamical and AI models alike: (1) instrumental understanding, which is obtained when a model has passed a functional test; (2) statistical understanding, obtained when researchers can make sense of the modeling results using statistical techniques to identify input–output relationships; and (3) component-level understanding, which refers to modelers' ability to point to specific model components or parts in the model architecture as the culprit for erratic model behaviors or as the crucial reason why the model functions well. We demonstrate how component-level understanding has been sought and achieved via climate model intercomparison projects over the past several decades. Such component-level understanding routinely leads to model improvements and may also serve as a template for thinking about AI-driven climate science. Currently, XAI methods can help explain the behaviors of AI models by focusing on the mapping between input and output, thereby increasing the statistical understanding of AI models. Yet, to further increase our understanding of AI models, we will have to build AI models that have interpretable components amenable to component-level understanding. We give recent examples from the AI climate science literature to highlight some recent, albeit limited, successes in achieving component-level understanding and thereby explaining model behavior. The merit of such interpretable AI models is that they serve as a stronger basis for trust in climate modeling and, by extension, downstream uses of climate model data.
This manuscript focuses on the modeling of electric vehicle (EV) driver’s range anxiety, a fear that a vehicle does not have sufficient range, or state of charge (SOC) of the battery pack, to reach its destination and would strand its occupants. Despite numerous research studies on the modeling of charging behaviors, modeling efforts to understand at what battery percentages do EV drivers charge their vehicles, and what are the associated contributing factors, are rather limited. To this end, an ensemble learning model based on gradient boosting is developed. The model sequentially fits new predictors to new residuals of the previous prediction and, then, minimizes the loss when adding the latest prediction. A total of 18 features are defined and extracted from the multisource data, which cover information on driver, vehicles, stations, traffic conditions, as well as spatial-temporal context information of the charging events. The analyzed dataset includes 4.5-year’s charging event log data from 3,096 users and 468 public charging stations in Kansas City Missouri, and the macroscopic travel demand model maintained by the metropolitan planning organization. Here, the result shows the proposed model achieved a satisfactory result with a R square value of 0.54 and root mean square error of 0.14, both better than multiple linear regression model and random forest model. To reduce range anxiety, it is suggested that the priorities of deploying new charging facilities should be given to the areas with higher daily traffic prediction, with more conservative EV users or that are further from residential areas.
In this paper we report an X-ray diffraction study on the phase behavior of binary lipid mixtures of 1-palmitoyl-2-docosahexaenoyl-sn-glycero-3-phosphoethanolamine (DHA-PE) and 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) at low concentrations below 5.0 mol% DHA-PE. Our results show that DHA-PE induces phase separation into a DHA rich liquid crystalline (L α ) phase and a DHA poor gel (L β' ) phase at overall DHA-PE concentrations as low as 0.1 mol%. In addition, we find that the structure of the L β' phase, from which the DHA-PE molecules are largely excluded, is modified in the phase separated state at low DHA-PE concentrations, with a decrease in bilayer thickness of 1.34 nm for 0.1 mol% at room temperature, compared to pure DPPC bilayers. This result is contrary to that seen in similar studies on mono-unsaturated lipids where an increase in bilayer thickness is observed. The surprising effect of such low DHA-PE concentrations on membrane structure may be important in understanding the role of highly polyunsaturated lipids in biological membrane-based structures and similar artificial surfactant systems.