Greg L. Zacharias

A computational situation assessment model for nuclear power plant operations

This paper presents a computational situation assessment (SA) model and a model-based SA metric for nuclear power plant operations. The model and metric development starts with a definition of the plant operators SA centered decision making behavior. A computational SA model and a model-based SA metric are then developed to quantify and measure operator SA. Using the SA model as a core, we further develop an integrated operator/plant model that provides for explicit representation of the operators fundamental functions of information processing, situation assessment, and decision making in a closed-loop plant/operator simulation environment. We evaluate the model and metric in a model-based simulation of a selected emergency scenario, and a model-based analysis of a range of contemplated nuclear power plant automation/aiding options.

Hybrid fuzzy logic flight controller synthesis via pilot modeling

This paper presents an investigation of a hybrid technique developed for synthesizing fuzzy logic controllers as stability augmentation systems. The hybrid technique combines the control capabilities of fuzzy logic with the learning capabilities of genetic algorithms, to yield a fuzzy logic controller augmentation system optimized to satisfy desired handling qualities requirements. An optimal control model is used to provide the closed-loop handling qualities metrics. The optimal control model is an analytic model of the pilot which can support the computation of in-flight handling qualities metrics, such as the standard Cooper-Harper rating. A genetic algorithm is used to optimize the attributes of the fuzzy logic controller. These attributes include the parameters of the fuzzy logic controller membership functions and the rule structure. The hybrid technique was implemented and tested in an offline engineering simulation using a wide-envelope F/A-18 longitudinal model. The tests included examining the robustness of the fuzzy logic controllers, the robustness of the genetic algorithms optimization technique, and the effect of changing the number of rules used in the fuzzy logic controller. Results indicate that the approach provides a robust design technique for fuzzy logic controller stability augmentation system synthesis and also show that the synthesized fuzzy logic controllers possess good robustness qualities.

Fuzzy logic gain scheduling for flight control

Demonstrates the potential capabilities of fuzzy logic to augment the design of flight control systems for high performance aircraft. Specifically, the authors detail the procedure for developing a fuzzy logic gain scheduler for use over the full flight envelope. The fuzzy logic approach has been implemented and tested using a full-envelope nonlinear F/A-18 simulation. Comparisons are made with a conventional approach that uses numerical interpolation. Results demonstrate the feasibility and flexibility of the approach to provide adequate control performance across the flight envelope, outside the flight envelope, and in the presence of noisy air data measurements.<<ETX>>

Malware Analysis and attribution using Genetic Information

As organizations become ever more dependent on networked operations, they are increasingly vulnerable to attack by a variety of attackers, including criminals, terrorists and nation states using cyber attacks. New malware attacks, including viruses, Trojans, and worms, are constantly and rapidly emerging threats. However, attackers often reuse code and techniques from previous attacks. Both by recognizing the reused elements from previous attacks and by detecting patterns in the types of modification and reuse observed, we can more rapidly develop defenses, make hypotheses about the source of the malware, and predict and prepare to defend against future attacks. We achieve these objectives in Malware Analysis and Attribution using Genetic Information (MAAGI) by adapting and extending concepts from biology and linguistics. First, analyzing the “genetics” of malware (i.e., reverse engineered representations of the original program) provides critical information about the program that is not available by looking only at the executable program. Second, the evolutionary process of malware (i.e., the transformation from one species of malware to another) can provide insights into the ancestry of malware, characteristics of the attacker, and where future attacks might come from and what they might look like. Third, functional linguistics is the study of the intent behind communicative acts; its application to malware characterization can support the study of the intent behind malware behaviors. To this point in the program, we developed a system that uses a range of reverse engineering techniques, including static, dynamic, behavioral, and functional analysis that clusters malware into families. We are also able to determine the malware lineage in some situations. Using behavioral and functional analysis, we are also able to identify a number of functions and purposes of malware.

Air traffic controller agent model for free flight

The objective of this research was to design and demonstrate an agent-based modeling and analysis tool for evaluating air traffic controller (ATC) performance under free flight air traffic management (ATM), and to develop a distributed decision-making model to investigate the potential for collaborative problemsolving under free flight. We developed a set of intelligent agent models representing ATC and pilot behavior in free flight. These models contain simplified representations of information processing and situation assessment applied to air traffic conflict detection. They also contain detailed air traffic conflict resolution models applying collaborative decisionmaking via inter-agent negotiation. We demonstrated the operation of a limited-scope prototype in a set of simulation trials designed to exercise the multi-agent decision-making model under a spectrum of free flight air traffic configurations. The results showed that our agent models are capable of performing on-line learning to support global situation awareness (SA) in the air traffic environment. The nature of conflict resolution maneuvers negotiated by the agents varied with conflict geometry in a manner consistent with intuitive expectation.

An intelligent tutoring system for simulator-based helicopter flight training

The development of an intelligent tutoring system for helicopter flight training is described. The Intelligent Flight Trainer (IFT) is a simulator-based system designed to assist students in developing proficiency on a suite of initial entry rotary wing maneuvers. It encapsulates instructor pilot domain knowledge in an expert system shell that provides tutorial and performance monitoring functions through a synthetic voice generator. The expert system shell works in concert with a variable stability augmentation control law that makes it easier for a neophyte student to control the motion of the simulated vehicle. Experimental verification of the IFT is currently under way at the UH-1 Training Research Simulator at Fort Rucker. NOMENCLATURE airborne flight training, and they have been used widely for both civilian and military training. Simulators operate indoors and are not affected by weather, and they are not subject to the problems and accidents that may occur in real flight (Gonzales and Ingraham, 1994). They do, however, require intensive supervision by instructors or check pilots, which can limit availability and increase expenses. Intelligent Tutoring Systems (ITSs) offer the potential to reduce this dependence on instructor pilots by automating the instructional process. The objective of this work is to hybridize flight simulation and ITS technologies to develop a system capable of teaching Army flight students how to perform basic helicopter flight maneuvers. E(.) Expected value h Altitude, ft p Roll rate, rad/sec q Pitch rate, rad/sec r Yaw rate, rad/sec t Time, sec x Distance along x-axis, ft/sec x State vector u Body x-axis velocity, ft/sec u Control vector v Body y-axis velocity , ft/sec V Airspeed, ft/sec or knots w Body z-axis velocity, ft/sec Intelligent tutoring systems are designed to train and instruct a user in a computerized environment. They are generally developed to take the place of a human instructor. They draw upon a body of domain knowledge that is embedded in the system as an expert system of rules (Farr and Psotka, 1992). ITSs present this expertise to a learner under the control of some appropriate pedagogical strategy tailored to his or her changing states of knowledge and understanding. Ideally, the ITS monitors and diagnoses the student’s progress to improved expertise in the form of an evolving student model. y Distance along y-axis, ft y Output vector z Distance along z-axis, ft col Collective input, inches lon Longitudinal cyclic input, inches lat Lateral cyclic input, inches ped Antitorque pedal input, inches Roll attitude, rad Much of the ITS work to date has concentrated on conventional curriculum areas, such as elementary subtraction (Ohlsson, 1990), fractions (Gutstein, 1992), basic Newtonian mechanics (Teodoro, 1990), and the like. They have also been used for teaching electronic troubleshooting (Brown, Burton, and deKleer, 1982). In these applications, the focus has been on building up the student’s knowledge base and on teaching the skills for manipulating that knowledge. Pitch attitude, rad

Approaches for modeling individuals within organizational simulations

The human behavior modeling community has traditionally been divided into those addressing individual behavior models, and those addressing organizational and team models. And yet it is clear that these extremes do not reflect the complex reality of the mutually-constraining interactions between an individual and his/her organizational environment. In this paper we argue that realistic models of organizations may require not only models of individual decision-makers, but also explicit models of a variety of individual differences influencing their decision-making and behavior (e.g., cognitive styles, personality traits, and affective states). Following a brief review of individual differences and cognitive architectures research, we describe two alternative approaches to modeling the individual within an organizational simulation: a cognitive architecture and a profile-based social network. We illustrate each approach with concrete examples from existing prototypes.

Distributed course-of-action planning using genetic algorithms, XML, and JMS

Future command and control (C2) systems must be constructed in such a way that they are extensible both in terms of the kinds of scenarios they can handle and the type of manipulations that they support. This paper presents an open architecture that uses commercial standards and implementations where appropriate. The discussion is framed by our ongoing work with a course of action planner and generator that uses genetic algorithms together with an abstract wargamer to suggest a small number of possible COAs (FOX).

AGENT-BASED PERFORMANCE ASSESSMENT TOOL FOR GENERAL AVIATION OPERATIONS UNDER FREE FLIGHT

The objective of this research is to design and demonstrate an agent-based modeling and analysis tool for evaluating General Aviation (GA) pilot situation awareness under free flight air traffic management (ATM). A computational tool is developed to assess free flights potential effect on GA operators, by combining an agent-based representation of the overall pilot/vehicle/ATM system with quantitative modelbased metrics of pilot SA. The models performance is demonstrated in a set of simulation trials designed to measure the pilot agents ability to recognize and correctly assess protected zone conflicts in free flight ATM, using information available from a hypothetical cockpit display of traffic information. A set of simulations is presented to examine the effect of sensor accuracy and attention allocation on pilot awareness of protected zone conflict hazards posed by intruder aircraft. The results show that reducing sensor accuracy leads to an increase in overall SA error, and that the pilot agent divides its attention over multiple traffic hazards in proportion to each intruders hazard potential. This attention-sharing varies dynamically as the conflict situation changes, in a manner that is consistent with intuitive expectations.

Multistage Integration Model for Human Egomotion Perception

Human computational vision models that attempt to account for the dynamic perception of egomotion and relative depth typically assume a common three-stage process: first, compute the optical flow field based on the dynamically changing image; second, estimate the egomotion states based on the flow; and third, estimate the relative depth/shape based on the egomotion states and possibly on a model of the viewed surface. We propose a model more in line with recent work in human vision, employing multistage integration. Here the dynamic image is first processed to generate spatial and temporal image gradients that drive a mutually interconnected state estimator and depth/shape estimator. The state estimator uses the image gradient information in combination with a depth/shape estimate of the viewed surface and an assumed model of the viewers dynamics to generate current state estimates; in tandem, the depth/shape estimator uses the image gradient information in combination with the viewers state estimate and assumed shape model to generate current depth/shape estimates. In this paper, we describe the model and compare model predictions with empirical data.