Search NASA⌕ Search

SEARCH · Search NASA

Results for “extended Kalman”

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.

At least 73 records · Page 4
All sourcesNASA NTRSDOE OSTI
NASA NTRS2002

Observer-Based Magnetic Bearing Controller Developed for Aerospace Flywheels

A prototype of a versatile, observer-based magnetic bearing controller for aerospace flywheels was successfully developed and demonstrated on a magnetic bearing test rig (see the photograph) and an actual flywheel module. The objective of this development included a fast, yet low risk, control development process, and a robust, high-performance controller for a large variety of flywheels. This required a good system model, an efficient development procedure, and a model-based controller that addressed the key problems associated with flywheel and bearing imbalance, sensor error, and vibration. The model used in this control development and tuning procedure included the flexible rotor dynamics and motor-induced vibrations. Such a model was essential for low-risk scheduling of speed-dependent control parameters and for reliable evaluation of novel control strategies. The successfully tested control prototype utilized an extended Kalman filter to estimate the true rotor principal-axis motion from the raw sensor position feedback. For control refinement, the extended Kalman filter also estimated and eliminated the combined effects of mass-imbalance and sensor runouts from the input data. A key advantage of the design based on the extended Kalman filter is its ability to accurately estimate both the rotor's principal-axis position and gyroscopic rates with the least amount of phase lag. This is important for control parameter scheduling to dampen the gyroscopic motions. Because of large uncertainties in the magnetic bearing and imbalance characteristics, this state-estimation scheme alone is insufficient for containing the rotor motion within the desired 1-mil excursion radius. A nonlinear gain adjustment based on an estimation of the principal-axis orbit size was needed to provide a coarse (nonoptimal), but robust, control of the orbit growth. Control current minimization was achieved with a (steepest gradient) search of synchronous errors in the principal-axis position input data. Actual flywheel tests of this observer-based controller (developed entirely in-house) at the NASA Glenn Research Center showed that the model correctly predicted the rotor orbit growth as a function of rotational speed, and it demonstrated the capability of gain adjustments to arrest this growth. Data from these tests on an actual flywheel module spun to 26,000 rpm proved that the controller was able to contain the shaft motion to within much less than 0.5 mils of radial excursion with axis currents less than 300 mA in root-mean-square estimate. The test speed range was limited because of thermal expansion concerns for this particular flywheel unit, not because of any deficiency in the controller. Simulations for this unit indicated that the controller should be robust up to its top operating speed of 60,000 rpm. Aside from these important achievements, and most significantly, it took less than 1 week to adapt this controller from the simple test rig to the actual flywheel and to demonstrate full five-axis levitation and control. This demonstration showed that both the controller and the model-based development and tuning framework are easily adaptable to a wide range of rotors and bearing configurations and, hence, are capable of reducing design risks and costs for many future flywheel technology developments.

Le, Dzu K.↗
DOE OSTI2022

Application of Recursive Estimation to Heat Tracing for Groundwater/Surface‐Water Exchange

Abstract We present and demonstrate a recursive‐estimation framework to infer groundwater/surface‐water exchange based on temperature time series collected at different vertical depths below the sediment/water interface. We formulate the heat‐transport problem as a state‐space model (SSM), in which the spatial derivatives in the convection/conduction equation are approximated using finite differences. The SSM is calibrated to estimate time‐varying specific discharge using the Extended Kalman Filter (EKF) and Extended Rauch‐Tung‐Striebel Smoother (ERTSS). Whereas the EKF is suited to real‐time (“online”) applications and uses only the past and current measurements for estimation (filtering), the ERTSS is intended for near‐real time or batch‐processing (“offline”) applications and uses a window of data for batch estimation (smoothing). The two algorithms are demonstrated with synthetic and field‐experimental data and are shown to be efficient and rapid for the estimation of time‐varying flux over seasonal periods; further, the recursive approaches are effective in the presence of rapidly changing flux and (or) nonperiodic thermal boundary conditions, both of which are problematic for existing approaches to heat tracing of time‐varying groundwater/surface‐water exchange.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Long short-term memory embedded nudging schemes for nonlinear data assimilation of geophysical flows

Reduced rank nonlinear filters are increasingly utilized in data assimilation of geophysical flows, but often require a set of ensemble forward simulations to estimate forecast covariance. On the other hand, predictor-corrector type nudging approaches are still attractive due to their simplicity of implementation when more complex methods need to be avoided. However, optimal estimate of nudging gain matrix might be cumbersome. In this paper, we put forth a fully nonintrusive recurrent neural network approach based on a long short-term memory (LSTM) embedding architecture to estimate the nudging term, which plays a role not only to force the state trajectories to the observations but also acts as a stabilizer. Furthermore, our approach relies on the power of archival data and the trained model can be retrained effectively due to power of transfer learning in any neural network applications. In order to verify the feasibility of the proposed approach, we perform twin experiments using Lorenz 96 system. Our results demonstrate that the proposed LSTM nudging approach yields more accurate estimates than both extended Kalman filter (EKF) and ensemble Kalman filter (EnKF) when only sparse observations are available. With the availability of emerging AI-friendly and modular hardware technologies and heterogeneous computing platforms, we articulate that our simplistic nudging framework turns out to be computationally more efficient than either the EKF or EnKF approaches.

42 ENGINEERING↗
NASA NTRS1992

Subsonic flight test evaluation of a propulsion system parameter estimation process for the F100 engine

Integrated engine-airframe optimal control technology may significantly improve aircraft performance. This technology requires a reliable and accurate parameter estimator to predict unmeasured variables. To develop this technology base, NASA Dryden Flight Research Facility (Edwards, CA), McDonnell Aircraft Company (St. Louis, MO), and Pratt & Whitney (West Palm Beach, FL) have developed and flight-tested an adaptive performance seeking control system which optimizes the quasi-steady-state performance of the F-15 propulsion system. This paper presents flight and ground test evaluations of the propulsion system parameter estimation process used by the performance seeking control system. The estimator consists of a compact propulsion system model and an extended Kalman filter. The extended Laman filter estimates five engine component deviation parameters from measured inputs. The compact model uses measurements and Kalman-filter estimates as inputs to predict unmeasured propulsion parameters such as net propulsive force and fan stall margin. The ability to track trends and estimate absolute values of propulsion system parameters was demonstrated. For example, thrust stand results show a good correlation, especially in trends, between the performance seeking control estimated and measured thrust.

Orme, John S.↗
DOE OSTI2021

Dynamic parameter estimation of generators

Various examples are provided extended Kalman filtering estimators for generators. In one example, among others an extended Kalman filtering estimator includes a dynamics estimator configured to calculate an estimated variable such as an estimated dynamic state or an estimated parameter of a generator, a geometry estimator configured to estimate generator values, and a Kalman filter component configured to determine a correction to the estimated state variable.

Miao, Zhixin↗
NASA NTRS1973

Filtering theory applied to orbit determination

Modifications to the extended Kalman filter and the Jazwinski filter are made and compared with the classical extended Kalman filter in applications to orbit determination using real data. The results show that with the kind of data available today, the application of filtering theories in this field presents many problems.

Torroglosa, V.↗
NASA NTRS2005

Generic Kalman Filter Software

The Generic Kalman Filter (GKF) software provides a standard basis for the development of application-specific Kalman-filter programs. Historically, Kalman filters have been implemented by customized programs that must be written, coded, and debugged anew for each unique application, then tested and tuned with simulated or actual measurement data. Total development times for typical Kalman-filter application programs have ranged from months to weeks. The GKF software can simplify the development process and reduce the development time by eliminating the need to re-create the fundamental implementation of the Kalman filter for each new application. The GKF software is written in the ANSI C programming language. It contains a generic Kalman-filter-development directory that, in turn, contains a code for a generic Kalman filter function; more specifically, it contains a generically designed and generically coded implementation of linear, linearized, and extended Kalman filtering algorithms, including algorithms for state- and covariance-update and -propagation functions. The mathematical theory that underlies the algorithms is well known and has been reported extensively in the open technical literature. Also contained in the directory are a header file that defines generic Kalman-filter data structures and prototype functions and template versions of application-specific subfunction and calling navigation/estimation routine code and headers. Once the user has provided a calling routine and the required application-specific subfunctions, the application-specific Kalman-filter software can be compiled and executed immediately. During execution, the generic Kalman-filter function is called from a higher-level navigation or estimation routine that preprocesses measurement data and post-processes output data. The generic Kalman-filter function uses the aforementioned data structures and five implementation- specific subfunctions, which have been developed by the user on the basis of the aforementioned templates. The GKF software can be used to develop many different types of unfactorized Kalman filters. A developer can choose to implement either a linearized or an extended Kalman filter algorithm, without having to modify the GKF software. Control dynamics can be taken into account or neglected in the filter-dynamics model. Filter programs developed by use of the GKF software can be made to propagate equations of motion for linear or nonlinear dynamical systems that are deterministic or stochastic. In addition, filter programs can be made to operate in user-selectable "covariance analysis" and "propagation-only" modes that are useful in design and development stages.

Lisano, Michael E., II↗
NASA NTRS2005

Adaptive Filtering Using Recurrent Neural Networks

A method for adaptive (or, optionally, nonadaptive) filtering has been developed for estimating the states of complex process systems (e.g., chemical plants, factories, or manufacturing processes at some level of abstraction) from time series of measurements of system inputs and outputs. The method is based partly on the fundamental principles of the Kalman filter and partly on the use of recurrent neural networks. The standard Kalman filter involves an assumption of linearity of the mathematical model used to describe a process system. The extended Kalman filter accommodates a nonlinear process model but still requires linearization about the state estimate. Both the standard and extended Kalman filters involve the often unrealistic assumption that process and measurement noise are zero-mean, Gaussian, and white. In contrast, the present method does not involve any assumptions of linearity of process models or of the nature of process noise; on the contrary, few (if any) assumptions are made about process models, noise models, or the parameters of such models. In this regard, the method can be characterized as one of nonlinear, nonparametric filtering. The method exploits the unique ability of neural networks to approximate nonlinear functions. In a given case, the process model is limited mainly by limitations of the approximation ability of the neural networks chosen for that case. Moreover, despite the lack of assumptions regarding process noise, the method yields minimum- variance filters. In that they do not require statistical models of noise, the neural- network-based state filters of this method are comparable to conventional nonlinear least-squares estimators.

Parlos, Alexander G.↗
NASA NTRS2002

Attitude Error Representations for Kalman Filtering

The quaternion has the lowest dimensionality possible for a globally nonsingular attitude representation. The quaternion must obey a unit norm constraint, though, which has led to the development of an extended Kalman filter using a quaternion for the global attitude estimate and a three-component representation for attitude errors. We consider various attitude error representations for this Multiplicative Extended Kalman Filter and its second-order extension.

Markley, F. Landis↗
NASA NTRS2005

Attitude Estimation for Large Field-of-View Sensors

The QUEST measurement noise model for unit vector observations has been widely used in spacecraft attitude estimation for more than twenty years. It was derived under the approximation that the noise lies in the tangent plane of the respective unit vector and is axially symmetrically distributed about the vector. For large field-of-view sensors, however, this approximation may be poor, especially when the measurement falls near the edge of the field of view. In this paper a new measurement noise model is derived based on a realistic noise distribution in the focal-plane of a large field-of-view sensor, which shows significant differences from the QUEST model for unit vector observations far away from the sensor boresight. An extended Kalman filter for attitude estimation is then designed with the new measurement noise model. Simulation results show that with the new measurement model the extended Kalman filter achieves better estimation performance using large field-of-view sensor observations.

Cheng, Yang↗
NASA NTRS2025

Pose Estimation in the Geometric Algebra G (3, 0, 1)

Using techniques from Geometric Algebra, methods for solving attitude determination problems are extended to the problem of simultaneously estimating attitude and location, i.e. pose, from direct or indirect measurements of geometric objects including points, lines, and planes. Wahba’s problem is generalized for the solution of initial pose. The multiplicative extended Kalman filter is generalized for the maintenance of pose solutions. Farrankopf’s model for biases in rotation rate sensors is generalized to biases in twist rate measurements. A multiplicative extended Kalman filter for pose in the presence of such biases is developed.

Sensor Bias Modeling↗
NASA NTRS1987

Spacecraft attitude determination using a second-order nonlinear filter

The stringent attitude determination accuracy and faster slew maneuver requirements demanded by present-day spacecraft control systems motivate the development of recursive nonlinear filters for attitude estimation. This paper presents the second-order filter development for the estimation of attitude quaternion using three-axis gyro and star tracker measurement data. Performance comparisons have been made by computer simulation of system models and filter mechanization. It is shown that the second-order filter consistently performs better than the extended Kalman filter when the performance index of the root sum square estimation error of the quaternion vector is compared. The second-order filter identifies the gyro drift rates faster than the extended Kalman filter. The uniqueness of this algorithm is the online generation of the time-varying process and measurement noise covariance matrices, derived as a function or the process and measurement nonlinearity, respectively.

Vathsal, S.↗
NASA NTRS2014

Orbit Determination Accuracy Analysis of the Magnetospheric Multiscale Mission During Perigee Raise

The Goddard Space Flight Center (GSFC) Flight Dynamics Facility (FDF) will provide orbit determination and prediction support for the Magnetospheric Multiscale (MMS) mission during the missions commissioning period. The spacecraft will launch into a highly elliptical Earth orbit in 2015. Starting approximately four days after launch, a series of five large perigee-raising maneuvers will be executed near apogee on a nearly every-other-orbit cadence. This perigee-raise operations concept requires a high-accuracy estimate of the orbital state within one orbit following the maneuver for performance evaluation and a high-accuracy orbit prediction to correctly plan and execute the next maneuver in the sequence. During early mission design, a linear covariance analysis method was used to study orbit determination and prediction accuracy for this perigee-raising campaign. This paper provides a higher fidelity Monte Carlo analysis using the operational COTS extended Kalman filter implementation that was performed to validate the linear covariance analysis estimates and to better characterize orbit determination performance for actively maneuvering spacecraft in a highly elliptical orbit. The study finds that the COTS extended Kalman filter tool converges on accurate definitive orbit solutions quickly, but prediction accuracy through orbits with very low altitude perigees is degraded by the unpredictability of atmospheric density variation.

Determination↗
NASA NTRS2014

Orbit Determination Accuracy Analysis of the Magnetospheric Multiscale Mission During Perigee Raise

The Goddard Space Flight Center (GSFC) Flight Dynamics Facility (FDF) will provide orbit determination and prediction support for the Magnetospheric Multiscale (MMS) mission during the mission's commissioning period. The spacecraft will launch into a highly elliptical Earth orbit in 2015. Starting approximately four days after launch, a series of five large perigee-raising maneuvers will be executed near apogee on a nearly every-other-orbit cadence. This perigee-raise operations concept requires a high-accuracy estimate of the orbital state within one orbit following the maneuver for performance evaluation and a high-accuracy orbit prediction to correctly plan and execute the next maneuver in the sequence. During early mission design, a linear covariance analysis method was used to study orbit determination and prediction accuracy for this perigee-raising campaign. This paper provides a higher fidelity Monte Carlo analysis using the operational COTS extended Kalman filter implementation that was performed to validate the linear covariance analysis estimates and to better characterize orbit determination performance for actively maneuvering spacecraft in a highly elliptical orbit. The study finds that the COTS extended Kalman filter tool converges on accurate definitive orbit solutions quickly, but prediction accuracy through orbits with very low altitude perigees is degraded by the unpredictability of atmospheric density variation.

Multiscale↗
NASA NTRS1977

Failure detection and correction for turbofan engines

In this paper, a failure detection and correction strategy for turbofan engines is discussed. This strategy allows continuing control of the engines in the event of a sensor failure. An extended Kalman filter is used to provide the best estimate of the state of the engine based on currently available sensor outputs. Should a sensor failure occur the control is based on the best estimate rather than the sensor output. The extended Kalman filter consists of essentially two parts, a nonlinear model of the engine and up-date logic which causes the model to track the actual engine. Details on the model and up-date logic are presented. To allow implementation, approximations are made to the feedback gain matrix which result in a single feedback matrix which is suitable for use over the entire flight envelope. The effect of these approximations on stability and response is discussed. Results from a detailed nonlinear simulation indicate that good control can be maintained even under multiple failures.

Corley, R. C.↗
NASA NTRS2002

Recent Flight Results of the TRMM Kalman Filter

The Tropical Rainfall Measuring Mission (TRMM) spacecraft is a nadir pointing spacecraft that nominally controls the roll and pitch attitude based on the Earth Sensor Assembly (ESA) output. TRMM's nominal orbit altitude was 350 km, until raised to 402 km to prolong mission life. During the boost, the ESA experienced a decreasing signal to noise ratio, until sun interference at 393 km altitude made the ESA data unreliable for attitude determination. At that point, the backup attitude determination algorithm, an extended Kalman filter, was enabled. After the boost finished, TRMM reacquired its nadir-pointing attitude, and continued its mission. This paper will briefly discuss the boost and the decision to turn on the backup attitude determination algorithm. A description of the extended Kalman filter algorithm will be given. In addition, flight results from analyzing attitude data and the results of software changes made onboard TRMM will be discussed. Some lessons learned are presented.

Andrews, Stephen F.↗
NASA NTRS1994

Improved solution accuracy for TDRSS-based TOPEX/Poseidon orbit determination

Orbit determination results are obtained by the Goddard Space Flight Center (GSFC) Flight Dynamics Division (FDD) using a batch-least-squares estimator available in the Goddard Trajectory Determination System (GTDS) and an extended Kalman filter estimation system to process Tracking and Data Relay Satellite (TDRS) System (TDRSS) measurements. GTDS is the operational orbit determination system used by the FDD in support of the Ocean Topography Experiment (TOPEX)/Poseidon spacecraft navigation and health and safety operations. The extended Kalman filter was implemented in an orbit determination analysis prototype system, closely related to the Real-Time Orbit Determination System/Enhanced (RTOD/E) system. In addition, the Precision Orbit Determination (POD) team within the GSFC Space Geodesy Branch generated an independent set of high-accuracy trajectories to support the TOPEX/Poseidon scientific data. These latter solutions use the geodynamics (GEODYN) orbit determination system with laser ranging and Doppler Orbitography and Radiopositioning integrated by satellite (DORIS) tracking measurements. The TOPEX/Poseidon trajectories were estimated for November 7 through November 11, 1992, the timeframe under study. Independent assessments were made of the consistencies of solutions produced by the batch and sequential methods. The batch-least-squares solutions were assessed based on the solution residuals, while the sequential solutions were assessed based on primarily the estimated covariances. The batch-least-squares and sequential orbit solutions were compared with the definitive POD orbit solutions. The solution differences were generally less than 2 meters for the batch-least-squares and less than 13 meters for the sequential estimation solutions. After the sequential estimation solutions were processed with a smoother algorithm, position differences with POD orbit solutions of less than 7 meters were obtained. The differences among the POD, GTDS, and filter/smoother solutions can be traced to differences in modeling and tracking data types, which are being analyzed in detail.

Doll, C. E.↗
NASA NTRS1993

Comparison of TOPEX/Poseidon orbit determination solutions obtained by the Goddard Space Flight Center Flight Dynamics Division and Precision Orbit Determination Teams

Orbit determination results are obtained by the Goddard Space Flight Center (GSFC) Flight Dynamics Division (FDD) using the Goddard Trajectory Determination System (GTDS) and a real-time extended Kalman filter estimation system to process Tracking Data and Relay Satellite (TDRS) System (TDRSS) measurements in support of the Ocean Topography Experiment (TOPEX)/Poseidon spacecraft navigation and health and safety operations. GTDS is the operational orbit determination system used by the FDD, and the extended Kalman fliter was implemented in an analysis prototype system, the Real-Time Orbit Determination System/Enhanced (RTOD/E). The Precision Orbit Determination (POD) team within the GSFC Space Geodesy Branch generates an independent set of high-accuracy trajectories to support the TOPEX/Poseidon scientific data. These latter solutions use the Geodynamics (GEODYN) orbit determination system with laser ranging tracking data. The TOPEX/Poseidon trajectories were estimated for the October 22 - November 1, 1992, timeframe, for which the latest preliminary POD results were available. Independent assessments were made of the consistencies of solutions produced by the batch and sequential methods. The batch cases were assessed using overlap comparisons, while the sequential cases were assessed with covariances and the first measurement residuals. The batch least-squares and forward-filtered RTOD/E orbit solutions were compared with the definitive POD orbit solutions. The solution differences were generally less than 10 meters (m) for the batch least squares and less than 18 m for the sequential estimation solutions. The differences among the POD, GTDS, and RTOD/E solutions can be traced to differences in modeling and tracking data types, which are being analyzed in detail.

Doll, C.↗