Rodney Martin

General Purpose Data-Driven System Monitoring for Space Operations

Modern space propulsion and exploration system designs are becoming increasingly sophisticated and complex. Determining the health state of these systems using traditional methods is becoming more difficult as the number of sensors and component interactions grows. Data-driven monitoring techniques have been developed to address these issues by analyzing system operations data to automatically characterize normal system behavior. The Inductive Monitoring System (IMS) is a data-driven system health monitoring software tool that has been successfully applied to several aerospace applications. IMS uses a data mining technique called clustering to analyze archived system data and characterize normal interactions between parameters. This characterization, or model, of nominal operation is stored in a knowledge base that can be used for real-time system monitoring or for analysis of archived events. Ongoing and developing IMS space operations applications include International Space Station flight control, satellite vehicle system health management, launch vehicle ground operations, and fleet supportability. As a common thread of discussion this paper will employ the evolution of the IMS data-driven technique as related to several Integrated Systems Health Management (ISHM) elements. Thematically, the projects listed will be used as case studies. The maturation of IMS via projects where it has been deployed, or is currently being integrated to aid in fault detection will be described. The paper will also explain how IMS can be used to complement a suite of other ISHM tools, providing initial fault detection support for diagnosis and recovery.

Discovering Anomalous Aviation Safety Events Using Scalable Data Mining Algorithms

The worldwide civilian aviation system is one of the most complex dynamical systems created. Most modern commercial aircraft have onboard flight data recorders that record several hundred discrete ...

Aircraft anomaly detection using performance models trained on fleet data

This paper describes an application of data mining technology called Distributed Fleet Monitoring (DFM) to Flight Operational Quality Assurance (FOQA) data collected from a fleet of commercial aircraft. DFM transforms the data into a list of abnormaly performing aircraft, abnormal flight-to-flight trends, and individual flight anomalies by fitting a large scale multi-level regression model to the entire data set. The model takes into account fixed effects: flight-to-flight and vehicle-to-vehicle variability. The regression parameters include aerodynamic coefficients and other aircraft performance parameters that are usually identified by aircraft manufacturers in flight tests. Using DFM, a multi-terabyte airline data set with a half million flights was processed in a few hours. The anomalies found include wrong values of computed variables such as aircraft weight and angle of attack as well as failures, biases, and trends in flight sensors and actuators. These anomalies were missed by the FOQA data exceedance monitoring currently used by the airline.

Sensor-Based Predictive Modeling for Smart Lighting in Grid-Integrated Buildings

Studies show that if we retrofit all the lighting systems in the buildings of California with dimming ballasts, then it would be possible to obtain a 450 MW of regulation, 2.5 GW of nonspinning reserve, and 380 MW of contingency reserve from participation of lighting loads in the energy market. However, in order to guarantee participation, it will be important to monitor and model lighting demand and supply in buildings. To this end, wireless sensor and actuator networks have proven to bear a great potential for personalized intelligent lighting with reduced energy use at 50%-70%. Closed-loop control of these lighting systems relies upon instantaneous and dense sensing. Such systems can be expensive to install and commission. In this paper, we present a sensor-based intelligent lighting system for future grid-integrated buildings. The system is intended to guarantee participation of lighting loads in the energy market, based on predictive models of indoor light distribution, developed using sparse sensing. We deployed ~92 % fewer sensors compared with state-of-art systems using one photosensor per luminaire. The sensor modules contained small solar panels that were powered by ambient light. Reduction in sensor deployments is achieved using piecewise linear predictive models of indoor light, discretized by clustering for sky conditions and sun positions. Day-ahead daylight is predicted from forecasts of temperature, humidity, and cloud cover. With two weeks of daylight and artificial light training data acquired at the sustainability base at NASA Ames, our model was able to predict the illuminance at seven monitored workstations with 80%-95% accuracy. Moreover, our support vector regression model was able to predict day-ahead daylight at ~92% accuracy.

Supporting Personizable Virtual Internet of Things

This paper reports the design and development of an HTML5-empowered Virtual Sensor Editor (VSE) over Internet of Things cloud. VSE is a scalable tool that allows users to design virtual sensors with user-defined dataflow logic, by visually aggregating existing sensors, either physical sensors or user-defined virtual sensors. VSE supports a real-time and historical visualization of sensor values and analytical studies, and is a cross-platform and customizable tool equipped with ability to support verifiable sensor data service compos ability. A discussion on design decisions is presented. Our preliminary work has been applied to NASA Sustainability Base for Smart Building monitoring. Preliminary performance and scalability study is also reported.

A State-Space Approach to Optimal Level-Crossing Prediction for Linear Gaussian Processes

In this paper, approximations of an optimal level-crossing predictor for a zero-mean stationary linear dynamical system driven by Gaussian noise in state-space form are investigated. The study of this problem is motivated by the practical implications for design of an optimal alarm system, which will elicit the fewest false alarms for a fixed detection probability in this context. This work introduces the use of Kalman filtering in tandem with the optimal level-crossing prediction problem. It is shown that there is a negligible loss in overall accuracy when using approximations to the theoretically optimal predictor, at the advantage of greatly reduced computational complexity.

Unsupervised Anomaly Detection and Diagnosis for Liquid Rocket Engine Propulsion

The results of a comprehensive array of unsupervised anomaly detection algorithms applied to Space Shuttle main engine (SSME) data are presented. Most of the algorithms are based upon variants of the well-known unconditional Gaussian mixture model (GMM). One goal of the paper is to demonstrate the maximum utility of these algorithms by the exhaustive development of a very simple GMM. Selected variants will provide us with the added benefit of diagnostic capability. Another algorithm that shares a common technique for detection with the GMM is presented, but instead uses a different modeling paradigm. The model provides a more rich description of the dynamics of the data, however the data requirements are quite modest. We will show that this very simple and straightforward method finds an event that characterizes a departure from nominal operation. We show that further diagnostic investigation with the GMM-based method can be used as a means to gain insight into operational idiosyncrasies for this nominally categorized test. Therefore, by using both modeling paradigms we can corroborate planned operational commands or provide warnings for unexpected operational commands.

Near Real-Time Optimal Prediction of Adverse Events in Aviation Data

The prediction of anomalies or adverse events is a challenging task, and there are a variety of methods which can be used to address the problem. In this paper, we demonstrate how to recast the anomaly prediction problem into a form whose solution is accessible as a level-crossing prediction problem. The level-crossing prediction problem has an elegant, optimal, yet untested solution under certain technical constraints, and only when the appropriate modeling assumptions are made. As such, we will thoroughly investigate the resilience of these modeling assumptions, and show how they aect nal performance. Finally, the predictive capability of this method will be assessed by quantitative means, using both validation and test data containing anomalies or adverse events from real aviation data sets that have previously been identied as operationally signicant by domain experts. It will be shown that the formulation proposed yields a lower false alarm rate on average than competing methods based on similarly advanced concepts, and a higher correct detection rate than a standard method based upon exceedances that is commonly used for prediction.

Neural network forecasting of solar power for NASA Ames sustainability base

Solar power prediction remains an important challenge for renewable energy integration primarily due to its inherent variability and intermittency. In this work, a neural network based solar power forecasting framework is developed for the NASA Ames Sustainability Base (SB) solar array using the publicly available National Oceanic and Atmospheric Administration (NOAA) weather data forecasts. The prediction inputs include temperature, irradiance and wind speed obtained through the NOAA NOMADS server in real-time. The neural network (ANN) is trained and tested on input-output data from on-site sensors. The NOAA archived forecast data is then input to the trained ANN model to predict power output spanning over nine months (June 2013-March 2014). The efficacy of the model is determined by comparing predicted power output against on-site sensor data.

Optimal level-crossing prediction for jump linear MIMO dynamical systems

In this article, the theoretically optimal prediction of level-crossings for a jump linear MIMO (multi-input/multi-output) dynamical system driven by a control input is investigated. The study of this problem is motivated by the practical implications for design of an optimal alarm system as applied to the advance prediction of adverse events, which will elicit the fewest false alarms for a fixed detection probability. It was found that using an additional control input term results in increased uncertainty due to the associated model for the control. However, using the appropriate condition for optimality provides an accommodation for this uncertainty which does not translate into compromised predictive capability. For the given application, it was observed that modeling the control input as a linear dynamical system results in better prediction performance and qualitatively has a higher model fidelity than when using a hidden Markov model for the same function.