Design Practices for ISHM
We are interested in standard formal practices and methodologies used in engineering design and propose a design environment where ISHM systems can be developed in conjunction with the system and subsystem design.
POC: Irem Tumer,  650-604-2976
PDF: Design Practices for ISHM

Model-Based Fault Identification and Recovery: Livingstone 2 (L2) is a model-based diagnosis and recovery engine that can estimate the state of the modeled system and suggest recovery actions to reach specified goals. L2 has been applied to several testbeds including the X-34 propulsion system (PITEX) and the International Space Station Command and Data Handling subsystem. L2’s predecessor, Livingstone, was part of the highly successful Remote Agent Experiment (RAX) on the Deep Space One spacecraft.  In mid 2004, Livingstone 2 will participate in a second flight experiment on Earth Observing Satellite 1 (EO-1).
POC: Sriram Narasimhan, 650-604-0832
PDF: Livingstone on EO-1, Livingstone 2

Advanced Methods for Diagnosis: We are currently developing the next generation of Livingstone called Livingstone 3 (L3). L3 extends L2’s capabilities along three dimensions: quantitative (and hybrid) reasoning, dynamic reasoning, and stochastic reasoning.
POC: Sriram Narasimhan, 650-604-0832
PDF: Advanced Methods for Diagnosis

Hybrid Diagnosis: We are developing methods for hybrid (qualitative and quantitative) model-based diagnosis based on particle filters.  These methods solve discretization problems inherent in purely qualitative approaches to diagnosis and fault identification. 
POC: Richard Dearden, 650-604-5616
PDF: Hybrid Diagnosis

Data Mining for ISHM: We are developing methods to automatically detect unusual or anomalous data in either historical or real-time sensor data, so that people can direct their attention to the unusual data. These methods can also be used to construct monitors for use with a model-based diagnosis system such as Livingstone.
POC: Mark Schwabacher, 650-604-4274
PDF: Data Mining for ISHM

 Statistical and Signal Analysis for Prognostics:  Statistical/signal analyses are being conducted to determine how complex periodic signals, and physical0 debris indicators, respond under nominal operating conditions.  Typically, multivariate signal analyses are used within a defined parametric state-space.
POC: Ed Huff, 650-604-4870
PDF: Healthwatch-2

Data-Driven Modeling for Prognostics: Prognosis of future failure states and predicting the type of failure is extremely difficult due to the high dimensionality of the problem. The number of relevant dimensions for prognosis of spacecraft failures is in the tens of thousands.  There are several scientific approaches to prognostics including data-driven, physics-based, and statistical approaches.  We use a variety of advanced real-time data mining techniques that incorporate model-based information with sensor data to identify potential precursors of failure.  Using these techniques, we can forecast trends and potentially anomalous behavior based on real-time information. 
POC: Ashok Srivastava

Damage Metric and Signature Modeling:  Damage metrics are being developed and evaluated using empirical data derived from operational flights and ground test rigs.  Difficult problems concerning  sampling requirements, data preparation, and signal separation are being addressed.  We also use first principles, finite-difference modeling to simulate normal and damage signature trajectories of mechanical components.  These methods make us of unique automated grid generation techniques, and may be applied to damage propagation of non-rotating components.
POC: Ed Huff, 650-604-4870
PDF: Structural Vibration Modeling

Prognostic tools for complex dynamical systems: We are developing a method for statistical inference on multidimensional nonlinear dynamical systems and dynamical networks. The inference method allows us to learn, and track in time, the structure of a dynamical network from a set of time-series measurements coming from the individual subsystems (nodes). Potential failures and undesirable trends reveal themselves in changes of the network structure. By tracking the structure in time such changes can be detected and mitigated.
POC: Vadim Smelyanskiy, 650-604-2261
PDF: Prognostics for Complex Systems

Mitigation of potential catastrophic failures: Our recent results on the statistical inference of fully nonlinear multidimensional dynamical models from the sensor data allows us to estimate the degrading stability thresholds, provide early warning of hidden system faults, and develop a set of optimal  controllers that will help mitigate the catastrophic events before they occur.
POC: Vadim Smelyanskiy, 650-604-2261

Multi-level Immune Learning Detection (MILD): The MILD software tool implements an immunity-based technique for anomaly and fault detection.  The technique uses a real-valued Negative Selection Algorithm (NSA) algorithm inspired by the biological immune system for detection of faults in system operation. The detection algorithm uses sensory data exhibiting the normal behavior patterns to generate probabilistically a set of fault detectors that can detect any abnormalities (including faults and damages) in the behavior pattern of the system response. Different versions of the tool have been tested with datasets (collected under normal and various simulated failure conditions) using the NASA Ames human-in-the-loop high-fidelity C-17 flight simulator.
POC: K. Krishnakumar, 650-604-2417
PDF: Multi-Level Immune Learning

Verification and Validation of Autonomous Controllers: We have developed verification methods and tools for model-based reasoning, and particularly diagnosis and IVHM, over the last five years.  The tools apply to the Livingstone diagnosis system as a target application (with extensions towards MIT’s Titan), but the scope of the underlying methods is potentially much broader.  We have been exploring three main lines of work: 1) Model-based verification, 2) Simulation-based verification, 3) Verification of diagnosability.  We have demonstrated these three approaches to various degrees on two main applications: first, the In-Situ Propellant Production system developed at KSC; second, the X-34 propulsion feed sub-system and its PITEX project sequel.
POC: Charles Pecheur, 650-604-3588
PDF: Verification and Validation for ISHM

Automated software synthesis for monitors and classifiers: We have developed methods to automatically generate efficient code for advanced analysis (e.g., statistical analysis, filters (Kalman filters, particle filters, etc) of sensors and other data sources) as monitoring components in an ISHM system.  The system generates highly documented and compact code for customizable platforms (e.g., Matlab, VxWorks). Automatic certification provides a verifiable guarantee that important software properties (e.g., memory safety) are not violated.
POC: Johann Schumann, 650-604-0941
PDF: Automated Software Synthesis

Intelligent Vehicle Automation: Spacecraft, airplanes, launch vehicles and other complex systems use ISHM methods to gather information about the health of the system and understand the implications of failures. We focus on testability and diagnosability during design to optimize sensor selection.  We also work on runtime algorithms to monitor and diagnose failures of systems during operation.  
POC: Ann Patterson-Hine, 650-604-4178
PDF: Autonomous Space Nuclear

Advanced Diagnostic Systems for ISS: NASA Ames and Johnson Space Center collaborate to develop next-generation capabilities for enhanced space vehicle management. The focus is on real-time subsystem monitoring, fault detection, advanced diagnosis and adaptive recovery for International Space Station (ISS) flight systems. The objective is to quantify the effort needed to develop such tools, and to demonstrate specific benefits by evaluating prototype tool suites during actual mission operations.
POC: Richard Alena, 650-604-0262
PDF: Advanced Diagnostics for ISS

ISStrider: Model-based Mission Operations: The ISStrider project is developing model-based reasoning, visualization and document retrieval capabilities for mission and spacecraft operations. These include the Caution and Warning Fusion (CWF) capability for root-cause determination, the Caution and Warning Cube (CW^3) for visualization and the Real-time Knowledge Management (RKM) capability for document retrieval. We are applying these capabilities to the International Space Station (ISS) domain. The lessons we learn from the ISS legacy spacecraft will allow us to develop approaches applicable to future spacecraft and mission operations concepts.
POC: Peter Robinson, 650-604-35132
POC: Mark Shirley, 650- 604-3389
PDF: ISStrider-Model Mission Ops