Tom Brotherton

Development of an Information Fusion System for Engine Diagnostics and Health Management

Aircraft gas-turbine engine data is available from a variety of sources including on-board sensor measurements, maintenance histories, and component models. An ultimate goal of Propulsion Health Management (PHM) is to maximize the amount of meaningful information that can be extracted from disparate data sources to obtain comprehensive diagnostic and prognostic knowledge regarding the health of the engine. Data Fusion is the integration of data or information from multiple sources, to achieve improved accuracy and more specific inferences than can be obtained from the use of a single sensor alone. The basic tenet underlying the data/information fusion concept is to leverage all available information to enhance diagnostic visibility, increase diagnostic reliability and reduce the number of diagnostic false alarms. This paper describes a basic PHM Data Fusion architecture being developed in alignment with the NASA C-17 Propulsion Health Management (PHM) Flight Test program. The challenge of how to maximize the meaningful information extracted from disparate data sources to obtain enhanced diagnostic and prognostic information regarding the health and condition of the engine is the primary goal of this endeavor. To address this challenge, NASA Glenn Research Center (GRC), NASA Dryden Flight Research Center (DFRC) and Pratt & Whitney (P&W) have formed a team with several small innovative technology companies to plan and conduct a research project in the area of data fusion as applied to PHM * . Methodologies being developed and evaluated have been drawn from a wide range of areas including artificial intelligence, pattern recognition, statistical estimation, and fuzzy logic. This paper will provide a broad overview of this work, discuss some of the methodologies employed and give some illustrative examples.

eSTORM: Enhanced self tuning on-board real-time engine model

Abstract : A key to producing reliable engine diagnostics and prognostics resides in the fusion of different processing techniques. Fusion of techniques has been shown to improve diagnostic performance while simultaneously reducing false alarms. Presented here is an approach that fuses a physical model called STORM (Self Tuning Onboard, Real-time engine Model) developed by Pratt & Whitney, with an empirical neural net model to provide a unique hybrid model called enhanced STORM (eSTORM) for engine diagnostics. STORM is a piecewise linear approximation of the engine cycle deck. Though STORM provides significant improvement over existing real-time engine model methods, there are several effects that impact engine performance that STORM does not capture. Integrating an empirical model with STORM accommodates the modeling errors. This paper describes the development of eSTORM for a Pratt & Whitney high bypass turbofan engine. Results of using STORM and eSTORM on simulated engine data are presented and compared. eSTORM is shown to work extremely well in reducing STORM modeling errors and biases for the conditions considered.

Anomaly detector fusion processing for advanced military aircraft

Automated Prognostics and Health Management (PHM) is a requirement for advanced military aircraft. PHM is the key to achieving true condition-based maintenance. PHM processing strategies include modules for the processing of known nominal and fault conditions. However in real operations there will also occur faults and other off-nominal operations that were never anticipated nor ever encountered before. We call these events anomalies. Missing the presence of an anomaly could potentially be catastrophic with the loss of the pilot and aircraft. Several different anomaly detectors (ADs) have been developed for advanced military aircraft to solve this problem. Fusion of these ADs can significantly reduce false alarms while at the same time substantially improving detection performance. Fusion is a way of approaching the goal of perfect detection with zero false alarms. We have developed a neural net approach for performing AD fusion. Presented is a description of that technique and the application to military aircraft subsystem data.

A testbed for data fusion for helicopter diagnostics and prognostics

Abstract : Automated systems to perform aircraft diagnostics and prognostics are of current interest. Development of those systems requires large amounts of data (collection, monitoring, manipulation) to capture and characterize normal, known fault events, and to ensure data is captured early on in a fault progression to support prognostic system development. Continuous data collection is also required to capture relatively rare events. Data collected can then be analyzed to assist in the development of automated systems and for continuous updating of algorithms to improve detection, classification, and prognostic performance. IAC, in collaboration with the Air Force and Army, is developing a testbed on which to perform data collection, and develop diagnostic and prognostic processing techniques using Army helicopter vibration and engine performance data as part of the Armys Vibration Management Enhancement Program (VMEP). VMEP and the testbed being developed for collection and processing of VMEP data are described here.

Generic integrated PHM/controller system

The Air Force is developing a wide variety of technologies for aircraft propulsion system health management (PHM). Unfortunately, many promising new sensor technologies and new algorithms cannot easily be deployed for on-wing demonstration particularly for UAV health monitoring and control applications. We are developing a generic hardware and software system to support the creation and fielding of low cost, integrated engine PHM/control systems. This system is based on an AMD Geode processor, a Xilinx Virtex-II Pro FPGA and the VxWorkstrade real-time operating system. The MathWorks Simulinktrade and Xilinx System Generatortrade are used for algorithm development and validation of the integrated system operation. This system will also facilitate integration of third party PHM algorithms. Presented here is a description of the hardware and software platform being developed to host the PHM/control system and its application to the AIIDED UAV demonstrator

Empirical Tuning of an On-Board Gas Turbine Engine Model for Real-Time Module Performance Estimation

A practical consideration for implementing a real-time on-board engine component performance tracking system is the development of high fidelity engine models capable of providing a reference level from which performance changes can be trended. Real-time engine models made their advent as state variable models in the mid-1980s, which utilized a piecewise linear model that granted a reasonable representation of the engine during steady state operation and mild transients. Increased processor speeds over the next decade allowed more complex models to be considered, that were a combination of linear and nonlinear physics-based elements. While the latter provided greater fidelity over both transient operation and the engine operational flight envelope, these models could be further improved to provide the high level of accuracy required for long-term performance tracking, as well as address the issue of engine-to-engine variation. Over time, these models may deviate enough from the actual engine being monitored, as a result of improvements made during an engine’s life cycle such as hardware modifications, bleed and stator vane schedule alterations, cooling flow adjustments, and the like, that the module performance estimations are inaccurate and often misleading. The process described in this paper will address these shortcomings while maintaining the execution speed required for real-time implementation.

Ontogenetic reasoning system for autonomic logistics

Joint Strike Fighter Autonomic Logistics minimize operational and support costs by increasing system reliability, while reducing maintenance requirements to essential levels. Using Prognostics and Health Management, parts and service are ordered or performed only when needed, obviating costly routine scheduled maintenance, and reducing aircraft downtime. Realizing this vision requires communication between the aircraft, industrial contractors and suppliers, and the maintenance and support team. Management of interactions between these entities is challenging, characterized by uncertain information, and conflicting demands for resources. A system is needed to encode the knowledge of maintenance personnel, suggest corrective actions in fault conditions, and learn from previous decisions. Integration with the supply chain would free human resources for more critical decision making tasks. IAC is developing an intelligent software infrastructure to manage these complexities. This ontogenetic reasoning system features an adaptive knowledge base of maintenance information, and autonomous software agents which: (1) analyze on-board sensor and model data, and past behaviors; (2) recommended actions under dynamic and uncertain conditions; (3) manage knowledge base evolution; (4) connect maintenance activities to the supply chain; and (5) perform various communications, security and support functions. This paper presents the architectural design of the system, describes an example application scenario, and concludes with an assessment of the technical challenges in developing such a system using the multi-agent systems approach

On false alarm mitigation

Condition-based maintenance (CBM) of complex military vehicles or industrial machines presumes the capability to correctly detect faults in components or subsystems. Faults are malfunctions that are observed in the monitoring system. Two types of errors can occur during automated fault detection: (1) missed detections or (2) false alarms. The practical consequence of either type of error is that a failed component may not be replaced when necessary, or alternatively, may be unnecessarily serviced due to a false alarm. In the most catastrophic cases of incorrect fault detection, the outcome may include the loss of an expensive military asset, or human life. The false alarm problem is particularly troublesome for condition based maintenance and prognostics problems where fault signatures necessarily need to be detected at lower levels. Here, we present an overview of some of the vagaries of fault and anomaly detection in the framework of reducing false alarms. New algorithmic approaches to user-controllable false alarm rates are presented, followed by a dose of pragmatism, as real-world false alarm mitigation on a currently-deployed military aircraft is described

Adaptive On-Wing Gas Turbine Engine Performance Estimation

A key technological concept for producing reliable engine diagnostics and prognostics exploits the benefits of fusing sensor data, information, and/or processing algorithms. In this paper, we consider a real-time physics based model of a commercial turbofan engine called STORM: self tuning on-board, real-time engine model. The STORM system provides a means for tracking engine module performance changes in real-time. However, modeling error can have a corruptive effect on STORMs estimation of performance changes. Fusing an empirical neural network based model with STORM forms a unique hybrid model of the engine called enhanced STORM (eSTORM). This approach can eliminate the STORM engine diagnostic errors. A practical consideration for implementing the hybrid engine model, involves the application of some form of sequential model building to construct and specify the empirical elements. A methodology for constructing the empirical model (EM) in a sequential manner without the requirement for storing all of the original data has been developed. This paper describes the development of the adaptive hybrid model scheme for a commercial turbofan engine. This adaptive hybrid-modeling scheme has been implemented in real-time on an intelligent automation corporation (IAC) computational platform. Model performance achieved with the automated update algorithm using real on-wing commercial aircraft engine data will be presented.

aTEDS: airborne turbine engine diagnostic system

There is a significant need for technologies that improve flight safety and reduce maintenance and support costs for aircraft turbine engines. The C17 aircraft is being used extensively in the fight against terror in Iraq and other parts of the world. Engines are being stressed with both high operation temperatures and a severe short take off and landing environment. The current C17/F117 engines do not have an engine vibration monitoring system. If high vibration conditions occur the flight crew cannot easily identify the problem engine. IAC is developing an airborne turbine engine diagnostics system (aTEDS) for the C17/F117 engine. aTEDS is designed to automatically collect, process, and monitor vibration data collected from the F117 engine. In addition to the on-board system aTEDS includes a laptop PC ground based diagnostic system to aid in data collection, visualization, analysis and fault isolation. The description of the aTEDS system and preliminary results from collection and processing of T1 C17 flight data is presented in this paper.