SEARCH · Engineering Papers
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.
Rule-based models of human problem solving have typically not been tested for statistical significance. Three methods of testing rules - analysis of variance, randomization, and contingency tables - are presented. Advantages and disadvantages of the methods are also described.
Recent studies of relationships between subjective ratings of mental workload, performance, and human operator and task characteristics have indicated that these relationships are quite complex. In order to study the various relationships and place subjective mental workload within a theoretical framework, we developed a production system model for the performance component of the complex supervisory task called POPCORN. The production system model is represented by a hierarchial structure of goals and subgoals, and the information flow is controlled by a set of condition-action rules. The implementation of this production system, called POPEYE, generates computer simulated data under different task difficulty conditions which are comparable to those of human operators performing the task. This model is the performance aspect of an overall dynamic psychological model which we are developing to examine and quantify relationships between performance and psychological aspects in a complex environment.
Progress was made in the three following areas. In the rule-based modeling area, two papers related to identification and significane testing of rule-based models were presented. In the area of operator aiding, research focused on aiding operators in novel failure situations; a discrete control modeling approach to aiding PLANT operators was developed; and a set of guidelines were developed for implementing automation. In the area of flight simulator hardware and software, the hardware will be completed within two months and initial simulation software will then be integrated and tested.
Two project areas were pursued: the intelligent cockpit and human problem solving. The first area involves an investigation of the use of advanced software engineering methods to aid aircraft crews in procedure selection and execution. The second area is focused on human problem solving in dynamic environments, particulary in terms of identification of rule-based models land alternative approaches to training and aiding. Progress in each area is discussed.
The goal of the Autonomous Power System (APS) program is to develop and apply intelligent problem solving and control to the Space Station Freedom Electrical Power System (SSF/EPS) testbed being developed and demonstrated at NASA Lewis Research Center. The objectives of the program are to establish artificial intelligence technology paths, to craft knowledge-based tools with advanced human-operator interfaces for power systems, and to interface and integrate knowledge-based systems with conventional controllers. The Autonomous Power EXpert (APEX) portion of the APS program will integrate a knowledge-based fault diagnostic system and a power resource planner-scheduler. Then APEX will interface on-line with the SSF/EPS testbed and its Power Management Controller (PMC). The key tasks include establishing knowledge bases for system diagnostics, fault detection and isolation analysis, on-line information accessing through PMC, enhanced data management, and multiple-level, object-oriented operator displays. The first prototype of the diagnostic expert system for fault detection and isolation has been developed. The knowledge bases and the rule-based model that were developed for the Power Distribution Control Unit subsystem of the SSF/EPS testbed are described. A corresponding troubleshooting technique is also described.
This chapter will provide a thorough end-to-end description of the process for evaluation of three different data-driven algorithms for anomaly detection to select the best candidate for deployment as part of a suite of IVHM (Integrated Vehicle Health Management) technologies. These algorithms were deemed to be sufficiently mature enough to be considered viable candidates for deployment in support of the maiden launch of Ares I-X, the successor to the Space Shuttle for NASA's Constellation program. Data-driven algorithms are just one of three different types being deployed. The other two types of algorithms being deployed include a "nile-based" expert system, and a "model-based" system. Within these two categories, the deployable candidates have already been selected based upon qualitative factors such as flight heritage. For the rule-based system, SHINE (Spacecraft High-speed Inference Engine) has been selected for deployment, which is a component of BEAM (Beacon-based Exception Analysis for Multimissions), a patented technology developed at NASA's JPL (Jet Propulsion Laboratory) and serves to aid in the management and identification of operational modes. For the "model-based" system, a commercially available package developed by QSI (Qualtech Systems, Inc.), TEAMS (Testability Engineering and Maintenance System) has been selected for deployment to aid in diagnosis. In the context of this particular deployment, distinctions among the use of the terms "data-driven," "rule-based," and "model-based," can be found in. Although there are three different categories of algorithms that have been selected for deployment, our main focus in this chapter will be on the evaluation of three candidates for data-driven anomaly detection. These algorithms will be evaluated upon their capability for robustly detecting incipient faults or failures in the ground-based phase of pre-launch space shuttle operations, rather than based oil heritage as performed in previous studies. Robust detection will allow for the achievement of pre-specified minimum false alarm and/or missed detection rates in the selection of alert thresholds. All algorithms will also be optimized with respect to an aggregation of these same criteria. Our study relies upon the use of Shuttle data to act as was a proxy for and in preparation for application to Ares I-X data, which uses a very similar hardware platform for the subsystems that are being targeted (TVC - Thrust Vector Control subsystem for the SRB (Solid Rocket Booster)).
A fault monitoring and diagnosis expert system called Faultfinder was conceived and developed to detect and diagnose in-flight failures in an aircraft. Faultfinder is an automated intelligent aid whose purpose is to assist the flight crew in fault monitoring, fault diagnosis, and recovery planning. The present implementation of this concept performs monitoring and diagnosis for a generic aircraft's propulsion and hydraulic subsystems. This implementation is capable of detecting and diagnosing failures of known and unknown (i.e., unforseeable) type in a real-time environment. Faultfinder uses both rule-based and model-based reasoning strategies which operate on causal, temporal, and qualitative information. A preliminary evaluation is made of the diagnostic concepts implemented in Faultfinder. The evaluation used actual aircraft accident and incident cases which were simulated to assess the effectiveness of Faultfinder in detecting and diagnosing failures. Results of this evaluation, together with the description of the current Faultfinder implementation, are presented.
The seminar presents papers on modeling and simulation methodology, artificial intelligence and expert systems, environments for simulation/expert system development, and methodology for simulation/expert system development. Particular attention is given to simulation modeling concepts and their representation, modular hierarchical model specification, knowledge representation, and rule-based diagnostic expert system development. Other topics include the combination of symbolic and discrete event simulation, real time inferencing, and the management of large knowledge-based simulation projects.
An approach to natural language meaning-based parsing in which the unit of linguistic knowledge is the word rather than the rewrite rule is described. In the word expert parser, knowledge about language is distributed across a population of procedural experts, each representing a word of the language, and each an expert at diagnosing that word's intended usage in context. The parser is structured around a coroutine control environment in which the generator-like word experts ask questions and exchange information in coming to collective agreement on sentence meaning. The word expert theory is advanced as a better cognitive model of human language expertise than the traditional rule-based approach. The technical discussion is organized around examples taken from the prototype LISP system which implements parts of the theory.
The modifications to the rule-based CAME program which allow it to more accurately model the fault-handling processes of fault-tolerant systems are described. This new capability is added to the CAME program by modeling the fault-handling processes of fault-tolerant systems. The integrated airframe/propulsion control system architecture (IAPSA II) reference configuration currently under development is detailed.
A basic paradigm that allows representation of physical systems with a focus on context and time is presented. Paragon provides the capability to quickly capture an expert's knowledge in a cognitively resonant manner. From that description, Paragon creates a simulation model in LISP, which when executed, verifies that the domain expert did not make any mistakes. The Achille's heel of rule-based systems has been the lack of a systematic methodology for testing, and Paragon's developers are certain that the model-based approach overcomes that problem. The reason this testing is now possible is that software, which is very difficult to test, has in essence been transformed into hardware.
The rule-based Computer-Aided Markov Evaluator (CAME) program was expanded in its ability to incorporate the effect of fault-handling processes into the construction of a reliability model. The fault-handling processes are modeled as semi-Markov events and CAME constructs and appropriate semi-Markov model. To solve the model, the program outputs it in a form which can be directly solved with the Semi-Markov Unreliability Range Evaluator (SURE) program. As a means of evaluating the alterations made to the CAME program, the program is used to model the reliability of portions of the Integrated Airframe/Propulsion Control System Architecture (IAPSA 2) reference configuration. The reliability predictions are compared with a previous analysis. The results bear out the feasibility of utilizing CAME to generate appropriate semi-Markov models to model fault-handling processes.
The application of the rule identification technique to the analysis of human performance data is proposed. The relation between the language and identifiable consistencies is discussed. The advantages of production system models for the description of complex human behavior are studied. The use of a Monte Carlo significance testing procedure to assure the validity of the rule identification is examined. An example of the rule-based analysis of Palmer's (1983) data is presented.
The model of the system assumes a constant power source and loads (experiments) whose power demands exceed the supply. Experiments are described by their name, power consumption, time for a complete run, present status and the state of the load. The power consumption of each load is set at a constant level but can be dynamically modified by the operator. The status specifies if the experiment is running, paused, completed or failed. The state compensates for the lack of actual feedback sensor data, by signifying the stability of the load. Experiments are scheduled to keep as many running as possible with the current system limitations. A graphics oriented user interface is embedded into the rule-based system to enable an operator to easily experiment with the system.
A system for fault detection and isolation called LES, developed at the Kennedy Space Center for the Space Shuttle's Launch Processing System, is a well-developed diagnostic system that is simultaneously model-based and sensor-based. This experiment has led to a surprising result: the failure of a sensor can not only be handled in precisely the same way as the failure of any other object, but may present an especially easy case. Classical rule-based diagnostic systems need to find out whether or not their sensors are telling them the truth before they can safely draw inferences from them. By contrast, while LES does treat sensors as a special case, it does so only because there may exist a short cut that allows them to be handled more simply than other objects. LES uses both structural and functional knowledge, and has found cases in which the structural knowledge can be economically replaced by the judicious use of functional relationships; LES' functional relationships are stored in exactly one place, so they must be inverted to determine hypothetical values for possibly faulty objects. The inversion process has been extended to include conditional functions not normally considered to have inverses.
The High Energy Astrophysics Costing Tool (HEACT) is a combined Expert System/Numerical Simulation to evaluate sensors and experiments proposed for space platforms in order to determine their expected performance and to assess the amount and quality of scientific information likely to be gathered by such experiments. The end product of this tool will be both a cost estimate for the experiment package as deployed and also an estimate of the amount of new scientific knowledge (both new data and reductions in the uncertainties of previous measurements) which would result from the actual deployment and operation of that experiment package. The numerical simulation part of HEACT will contain components to: (1) calculate the high energy signatures of both actual and proposed classes of astrophysical objects, and (2) model the response and resolution of the candidate sensors for observing those objects. The rule based part of the HEACT will be based on knowledge in the areas of experiment design, instrument selection, and system integration developed from experience with existing high energy astrophysics observatories, both orbiting and lofted by rockets or balloons. This paper presents mainly results from the design of the rule-based part of HEACT. It focuses on the structure of the rules and the structures used to represent the knowledge related to the design and integration of such experiments. Methods for integrating the results from numerical simulations into the expert system will also be discussed.
The ability to articulate, collect, and automate the application of the expertise needed for the analysis of space shuttle main engine (SSME) test data would be of great benefit to NASA liquid rocket engine experts. This paper describes a project whose goal is to build a rule-based expert system which incorporates such expertise. Experiential expertise, collected directly from the experts currently involved in SSME data analysis, is used to build a rule base to identify engine anomalies similar to those analyzed previously. Additionally, an alternate method of expertise capture is being explored. This method would generate rules inductively based on calculations made using a theoretical model of the SSME's operation. The latter rules would be capable of diagnosing anomalies which may not have appeared before, but whose effects can be predicted by the theoretical model.
A rule-base generation procedure is developed for expert systems used to diagnose anomalies in the performance of mechanical plants and similar engineering systems. The method is based on construction of a failure-mode information-propagation model (FIPM). Details of the FIPM procedure are discussed and illustrated with diagrams; reference is made to a sample application involving the turbopump of the high-pressure oxidizer for the Space Shuttle main engine.