Search NASA⌕ Search

SEARCH · Search NASA

Results for “Model reduction”

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 19 records
All sourcesNASA NTRSDOE OSTI
NASA NTRS1988

Modal model reduction or model reduction of large space structures in frequency domain

Large space structures are characterized by a large number of modes, grouped frequencies, and small inherent damping. Model reduction techniques in time domain may not be effective due to small damping. The model truncation method is generally used. This method can not solve the problem of grouped frequencies, and will lose all the information about the higher order modes. A new method developed in this paper, which tries to minimize the error of interested transfer functions, makes use of all the information of the original system, and achieves improvement not only from a smaller error of transfer functions but also from better frequency distribution.

Mifang, Ruan↗
DOE OSTI2026

Toward real-time optimization through model reduction and model discrepancy sensitivities

Optimization problems arise in a range of scenarios, from optimal control to model parameter estimation. In many applications, such as the development of digital twins, it is essential to solve these optimization problems within wall-clock-time limitations. However, this is often unattainable for complex systems, such as those modeled by nonlinear partial differential equations. One strategy for mitigating this issue is to construct a reduced-order model (ROM) that enables more rapid optimization. In particular, the use of nonintrusive ROMs—those that do not require access to the full-order model at evaluation time—is popular because they facilitate the computation of optimization solutions within the wall-clock time requirements. However, the optimization solution will be unreliable if the iterates move outside the ROM training data. This article proposes the use of hyper-differential sensitivity analysis with respect to model discrepancy (HDSA-MD) as a computationally efficient tool to augment ROM-constrained optimization and improve its reliability. The proposed approach consists of two phases: (i) an offline phase where several full-order model evaluations are computed to train the ROM, and (ii) an online phase where a ROM-constrained optimization problem is solved, a limited number of full-order model evaluations are computed, and HDSA-MD is used to enhance the optimization solution. Numerical results are demonstrated for two examples, atmospheric contaminant control and wildfire ignition location estimation, in which a ROM is trained offline using inaccurate atmospheric data. In conclusion, the HDSA-MD update yields a significant improvement in the ROM-constrained optimization solution using only one full-order model evaluation online with corrected atmospheric data.

PDE-constrained optimization↗
NASA NTRS1986

On model reduction

Three model reduction methods are described. These are the discrete balanced realizations of Mullis and Roberts (1976) where a characterization of the reduction error is given and a previously unknown L(infinity) norm bound on the reduction error, is obtained. Another method is a new model reduction technique for discrete time systems which has the advantage that the reduced order model is balanced and has an L(infinity) norm bound on the reduction error. The last method derived is a frequency weighting technique for continuous and discrete systems where it is possible to specify the approximation accuracy with frequency and also, for this method, an L(infinity) norm on the weighted reduction error is obtained.

Al-Saggaf, Ubaid M.↗
DOE OSTI2025

Online learning of quadratic manifolds from streaming data for nonlinear dimensionality reduction and nonlinear model reduction

Here, this work introduces an online greedy method for constructing quadratic manifolds from streaming data, designed to enable in situ analysis of numerical simulation data on the Petabyte scale. Unlike traditional batch methods, which require all data to be available upfront and take multiple passes over the data, the proposed online greedy method incrementally updates quadratic manifolds in one pass as data points are received, eliminating the need for expensive disk input/output operations as well as storing and loading data points once they have been processed. A range of numerical examples demonstrate that the online greedy method learns accurate quadratic manifold embeddings while being capable of processing data that far exceed common disk input/output capabilities and volumes as well as main-memory sizes.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1991

Model reduction by trimming for a class of semi-Markov reliability models and the corresponding error bound

Semi-Markov processes have proved to be an effective and convenient tool to construct models of systems that achieve reliability by redundancy and reconfiguration. These models are able to depict complex system architectures and to capture the dynamics of fault arrival and system recovery. A disadvantage of this approach is that the models can be extremely large, which poses both a model and a computational problem. Techniques are needed to reduce the model size. Because these systems are used in critical applications where failure can be expensive, there must be an analytically derived bound for the error produced by the model reduction technique. A model reduction technique called trimming is presented that can be applied to a popular class of systems. Automatic model generation programs were written to help the reliability analyst produce models of complex systems. This method, trimming, is easy to implement and the error bound easy to compute. Hence, the method lends itself to inclusion in an automatic model generator.

White, Allan L.↗
NASA NTRS1990

Projection and assembly method for multibody component model reduction

The problem addressed is that of obtaining reduced-order component models for use in simulating the dynamics of a multibody system. In certain cases, nonlinear system models may be constructed using linear dynamic models for each component, but allowing large angle motion between components. Without some form of model reduction, system models constructed in this manner may be too large for use in control system design and simulation trades. This paper analyzes one method of component model reduction that allows systems level requirements (e.g., capturing the effect of body 1 reaction wheel noise on body 2 camera pointing) to aid in the selection of the reduced-order component models. Briefly stated, important modes are selected at the system level and projected onto the components, and reduced-order components are then assembled into a reduced-order system model that retains the projected modes.

Bernard, Douglas E.↗
NASA NTRS1992

A new algorithm for L2 optimal model reduction

In this paper the quadratically optimal model reduction problem for single-input, single-output systems is considered. The reduced order model is determined by minimizing the integral of the magnitude-squared of the transfer function error. It is shown that the numerator coefficients of the optimal approximant satisfy a weighted least squares problem and, on this basis, a two-step iterative algorithm is developed combining a least squares solver with a gradient minimizer. Convergence of the proposed algorithm to stationary values of the quadratic cost function is proved. The formulation is extended to handle the frequency-weighted optimal model reduction problem. Three examples demonstrate the optimization algorithm.

Spanos, J. T.↗
NASA NTRS1988

Model reduction methods for control design

Several different model reduction methods are developed and detailed implementation information is provided for those methods. Command files to implement the model reduction methods in a proprietary control law analysis and design package are presented. A comparison and discussion of the various reduction techniques is included.

Dunipace, K. R.↗
NASA NTRS1992

A component modes projection and assembly model reduction methodology for articulated, multi-flexible body structures

A two-stage model reduction methodology, combining the classical Component Mode Synthesis (CMS) method and the newly developed Enhanced Projection and Assembly (EP&A) method is proposed in this research. The first stage of this methodology, called COmponent Mode Projection and Assembly model REduction (COMPARE) method, involves the generations of CMS mode sets, such as the MacNeal-Rubin mode sets. These mode sets are then used to reduce the order of each component model in the Rayleigh-Ritz sense. The resultant component models are then combined to generate reduced-order system models at various system configurations. A composite mode set which retains important system modes at all system configurations is then selected from these reduced-order system models. In the second stage, the EP&A model reduction method is employed to reduce further the order of the system model generated in the first stage. The effectiveness of the COMPARE methodology has been successfully demonstrated on a high-order, finite-element model of the cruise-configured Galileo spacecraft.

Lee, Allan Y.↗
NASA NTRS1993

A component modes projection and assembly model reduction methodology for articulated, multi-flexible body structures

A two-stage model reduction methodology, combining the classical Component Mode Synthesis (CMS) method and the newly developed Enhanced Projection and Assembly (EP&A) method, is proposed in this research. The first stage of this methodology, called the COmponent Modes Projection and Assembly model REduction (COMPARE) method, involves the generation of CMS mode sets, such as the MacNeal-Rubin mode sets. These mode sets are then used to reduce the order of each component model in the Rayleigh-Ritz sense. The resultant component models are then combined to generate reduced-order system models at various system configurations. A composite mode set which retains important system modes at all system configurations is then selected from these reduced-order system models. In the second stage, the EP&A model reduction method is employed to reduce further the order of the system model generated in the first stage. The effectiveness of the COMPARE methodology has been successfully demonstrated on a high-order, finite-element model of the cruise-configured Galileo spacecraft.

Lee, Allan Y.↗
NASA NTRS1992

Model reduction for Space Station Freedom

Model reduction is an important practical problem in the control of flexible spacecraft, and a considerable amount of work has been carried out on this topic. Two of the best known methods developed are modal truncation and internal balancing. Modal truncation is simple to implement but can give poor results when the structure possesses clustered natural frequencies, as often occurs in practice. Balancing avoids this problem but has the disadvantages of high computational cost, possible numerical sensitivity problems, and no physical interpretation for the resulting balanced 'modes'. The purpose of this work is to examine the performance of the subsystem balancing technique developed by the investigator when tested on a realistic flexible space structure, in this case a model of the Permanently Manned Configuration (PMC) of Space Station Freedom. This method retains the desirable properties of standard balancing while overcoming the three difficulties listed above. It achieves this by first decomposing the structural model into subsystems of highly correlated modes. Each subsystem is approximately uncorrelated from all others, so balancing them separately and then combining yields comparable results to balancing the entire structure directly. The operation count reduction obtained by the new technique is considerable: a factor of roughly r(exp 2) if the system decomposes into r equal subsystems. Numerical accuracy is also improved significantly, as the matrices being operated on are of reduced dimension, and the modes of the reduced-order model now have a clear physical interpretation; they are, to first order, linear combinations of repeated-frequency modes.

Williams, Trevor↗
NASA NTRS1992

Model reduction in a subset of the original states

A model reduction method is investigated to provide a smaller structural dynamic model for subsequent structural control design. A structural dynamic model is assumed to be derived from finite element analysis. It is first converted into the state space form, and is further reduced by the internal balancing method. Through the co-ordinate transformation derived from the states that are deleted during reduction, the reduced model is finally expressed with the states that are members of the original states. Therefore, the states in the final reduced model represent the degrees of freedom of the nodes that are selected by the designer. The procedure provides a more practical implementation of model reduction for applications in which specific nodes, such as sensor and/or actuator attachment points, are to be retained in the reduced model. Thus, it ensures that the reduced model is under the same input and output condition as the original physical model. The procedure is applied to two simple examples and comparisons are made between the full and reduced order models. The method can be applied to a linear, continuous and time-invariant model of structural dynamics with nonproportional viscous damping.

Yae, K. H.↗
DOE OSTI2024

Optimization-Based Model Reduction Scheme for Renewable Energy Power Plants Using Standardized Testing Scenarios

This paper presents an optimization-based model reduction scheme for renewable energy (RE) power plants consisting of inverter-based resources (IBRs) operating in grid-following (GFL) or grid-forming (GFM) modes. More importantly, the datasets feeding the optimization-based model reduction scheme are generated and re-used through the standardized grid-interactive testing scenarios. Particularly, the proposed scheme makes use of the power plant point of common coupling (PCC) measurements of various quantities specified by standardized tests (e.g., voltage and frequency ride through) as per IEEE 2800, to estimate the parameters of the reduced-order model such that its dynamic performance aligns with the original detailed power plant model. The proposed model reduction approach does not require the parameters of individual IBRs and using standardized test data as input to the formulated optimization problem simplifies the reduced-order modelling scheme. Extensive case studies following standardized test scenarios verified the remarkable accuracy of the proposed approach.

Yallamilli, Ram S. [Purdue University]↗
NASA NTRS1988

Grammians and model reduction in limited time and frequency intervals

In this paper the controllability and observability grammians in limited time and frequency intervals are studied, and used for model reduction. In balanced and modal coordinates a near-optimal reduction procedure is developed, yielding the reduction error (norm of the difference between the output of the original system and the reduced model) almost minimal. Several examples are given to illustrate the concept for model reduction of continuous- and discrete-time systems, stable and unstable systems, in limited tim or/and frequency interval. Finally, the model reduction of a flexible truss structure is presented.

Gawronski, Wodek↗
NASA NTRS1990

Optimal model reduction and frequency-weighted extension

In this paper the quadratically optimal model reduction problem for single-input, single-output systems is considered. The reduced order model is determined by minimizing the integral of the magnitude-squared of the transfer function error. It is shown that the numerator coefficients of the optimal approximant satisfy a weighted least squares problem and, on this basis, a two-step iterative algorithm is developed combining a least squares solver with a gradient minimizer. The existence of globally optimal stable solutions to the optimization problem is established, and convergence of the algorithm to stationary values of the cost function is proved. The formulation is extended to handle the frequency-weighted optimal model reduction problem. Three examples demonstrate the optimization algorithm.

Spanos, J. T.↗
NASA NTRS1987

Model reduction for discrete bilinear systems

A model reduction method for discrete bilinear systems is developed which matches q sets of Volterra and covariance parameters. These parameters are shown to represent both deterministic and stochastic attributes of the discrete bilinear system. A reduced order model which matches these q sets of parameters is defined to be a q-Volterra covariance equivalent realization (q-Volterra COVER). An algorithm is presented which constructs a class of q-Volterra COVERs parameterized by solutions to a Hermitian, quadratic, matrix equation. The algorithm is applied to a bilinear model of a robot manipulator.

King, A. M.↗
NASA NTRS1993

Model reduction for the dynamics and control of large structural systems via neutral network processing direct numerical optimization

Three neural network processing approaches in a direct numerical optimization model reduction scheme are proposed and investigated. Large structural systems, such as large space structures, offer new challenges to both structural dynamicists and control engineers. One such challenge is that of dimensionality. Indeed these distributed parameter systems can be modeled either by infinite dimensional mathematical models (typically partial differential equations) or by high dimensional discrete models (typically finite element models) often exhibiting thousands of vibrational modes usually closely spaced and with little, if any, damping. Clearly, some form of model reduction is in order, especially for the control engineer who can actively control but a few of the modes using system identification based on a limited number of sensors. Inasmuch as the amount of 'control spillover' (in which the control inputs excite the neglected dynamics) and/or 'observation spillover' (where neglected dynamics affect system identification) is to a large extent determined by the choice of particular reduced model (RM), the way in which this model reduction is carried out is often critical.

Becus, Georges A.↗