SEARCH · Engineering Papers
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.
This paper develops a new technique for the identification of parameters in distributed systems. The technique is based on the finite element method. An an illustration, the method is applied to the identification of the mass and stiffness distributions of a space structure, simulated by a nonuniform free-free beam.
This paper examines the use of on-orbit identification based on Maximum Likelihood Estimation (MLE) to provide these high-order, high-accuracy control design models for large space structures (LSS's). First, it outlines a general MLE identification algorithm, together with a covariance-analysis procedure to assess algorithm performance in terms of systematic and stochastic errors. Next, it examines various simplifications appropriate for the LSS identification application. Simplified analytical performance results are presented, as are numerical results to support these analyses. Finally, a graphical interpretation of these results is given.
This paper describes model structures and parameter estimation algorithms suitable for the identification of unsteady aerodynamic models from input-output data. The model structures presented are state space models and include linear time-invariant (LTI) models and linear parameter-varying (LPV) models. They cover a wide range of local and parameter dependent identification problems arising in unsteady aerodynamics and nonlinear flight dynamics. We present a residue algorithm for estimating model parameters from data. The algorithm can incorporate apriori information and is described in detail. The algorithms are evaluated on the F-16XL wind-tunnel test data from NAS Langley Research Center. Results of numerical evaluation are presented. The paper concludes with a discussion major issues and directions for future work.
An approach to system identification is presented which explicitly takes structure errors into account and hence provides a systematic way for answering questions concerning the magnitude of estimated parameter errors resulting from structural errors. It is indicated that, from this point of view, it is possible to define near equivalence between process and model and to obtain meaningful theoretical results on solution error system identification. It remains to apply these results to large realistic problems such as those involving models of complex man machine systems.
Structural modeling is a key part of structural system identification for large space structures. Finite-element structural models are commonly used in practice because of their general applicability and availability. The initial models generated by using a standard computer program such as NASTRAN, ANSYS, SUPERB, STARDYNE, STRUDL, etc., generally contain tens of thousands of degrees of freedom. The models must be reduced for purposes of identification. Not only does the magnitude of the identification effort grow exponentially as a function of the number of degrees of freedom, but numerical procedures may also break down because of accumulated round-off errors. Guyan reduction is usually applied after a static condensation. Misapplication of Guyan reduction can lead to serious modeling errors. It is quite unfortunate and disappointing, since the accuracy of the original detailed finite-element model one tries very hard to achieve is lost by the reduction. First, why and how Guyan reduction always causes loss of accuracy is examined. An alternative approach is then introduced. The alternative can be thought of as an improvement of Guyan reduction, the Rayleigh-Ritz method, and in particular the recent algorithm of Wilson, Yuan, and Dickens. Unlike Guyan reduction, the use of the alternative does not need any special insight, experience, or skill for partitioning the structural degrees of freedom. In addition to model condensation, this alternative approach can also be used for predicting analytically, quickly, and economically, what are those structural modes that are excitable by a force actuator at a given trial location. That is, in the excitation of the structural modes for identification, it can be used for guiding the placement of the force actuators.
The problem of identifying linear dynamical systems is studied by considering structural and deterministic properties of linear systems that have an impact on stochastic identification algorithms. In particular considered is parametrization of linear systems so that there is a unique solution and all systems in appropriate class can be represented. It is assumed that a parametrization of system matrices has been established from a priori knowledge of the system, and the question is considered of when the unknown parameters of this system can be identified from input/output observations. It is assumed that the transfer function can be asymptotically identified, and the conditions are derived for the local, global and partial identifiability of the parametrization. Then it is shown that, with the right formulation, identifiability in the presence of feedback can be treated in the same way. Similarly the identifiability of parametrizations of systems driven by unobserved white noise is considered using the results from the theory of spectral factorization.
This paper describes the use of a combined model and parameter identification approach, based on modal analysis and artificial intelligence (AI) techniques, for identifying damage or flaws in a rotating truss structure incorporating embedded piezoceramic sensors. This smart structure example is representative of a class of structures commonly found in aerospace systems and next generation space structures. Artificial intelligence techniques of classification, heuristic search, and an object-oriented knowledge base are used in an AI-based model identification approach. A finite model space is classified into a search tree, over which a variant of best-first search is used to identify the model whose stored response most closely matches that of the input. Newly-encountered models can be incorporated into the model space. This adaptativeness demonstrates the potential for learning control. Following this output-error model identification, numerical parameter identification is used to further refine the identified model. Given the rotating truss example in this paper, noisy data corresponding to various damage configurations are input to both this approach and a conventional parameter identification method. The combination of the AI-based model identification with parameter identification is shown to lead to smaller parameter corrections than required by the use of parameter identification alone.
Two extensions to an earlier work on system identification for large flexible structures are presented. The first extension applies an integrated frequency-domain ID approach to experiments utilizing rib root actuators for full system excitation of both 'boom-dish' and 'dish' modes of the structure; while the second extension employs a time-domain identification, utilizing frequency-domain results obtained for initialization of the parameter estimates. For the first extension, the results show that in the presence of closely packed modes, the curve-fit algorithm employed can distinguish modes with frequency separation as small as 0.04 Hz. The maximum likelihood estimation used in the second extension produce estimates close to the parametric modal values of frequencies and damping.
This paper reviews recent results in learning control and learning system identification, with particular emphasis on discrete-time formulation, and their relation to adaptive theory. Related continuous-time results are also discussed. Among the topics presented are proportional, derivative, and integral learning controllers, time-domain formulation of discrete learning algorithms. Newly developed techniques are described including the concept of the repetition domain, and the repetition domain formulation of learning control by linear feedback, model reference learning control, indirect learning control with parameter estimation, as well as related basic concepts, recursive and non-recursive methods for learning identification.
This paper presents two approaches that will improve the accuracy of explicit approximations of frequency responses. They are important components of structural design optimization and system identification. The first is a product form approximation that improves approximated frequency responses near resonant conditions. This approach is applicable to both design optimization and system identification problems. The second is aimed at improving the computed sensitivity coefficients that are used to solve system identification problems.
An outline of this work is presented. It begins with a schematic flow diagram and a logical flow diagram of the identification process for large space structures (LSS). Next, the task is defined by a structure model definition. A matrix polynomial formulation with a node displacement equation and a state variable formulation with node displacement and velocities are outlined. Further outlined is the identification of LSS on orbit; modeling errors and uncertainties; verification and validation of model; and noise, computations, and data collection.
The paper examines the on-orbit modal identification of large space structures, stressing the importance of planning and experience, in preparation for the Space Station Structural Characterization Experiment (SSSCE) for the Space Station Freedom. The necessary information to foresee and overcome practical difficulties is considered in connection with seven key factors, including test objectives, dynamic complexity of the structure, data quality, extent of exploratory studies, availability and understanding of software tools, experience with similar problems, and pretest analytical conditions. These factors affect identification success in ground tests. Comparisons with similar ground tests of assembled systems are discussed, showing that the constraints of space tests make these factors more significant. The absence of data and experiences relating to on-orbit modal identification testing is shown to make identification a uniquely mathematical problem, although all spacecraft are constructed and verified by proven engineering methods.
The paper deals with the problem of system identifiability for a linear dynamical system. Two theorems are given relating the sufficient condition of system identifiability for certain linear structures to the total number of inputs and outputs. The principle of least squares, that seeks the minimization of a cost function is employed to carry out the system identification process. To illustrate the concept of the paper, a structural model of a beam with point masses is examined. Parameter identification methods are studied and a random search technique is introduced.
A methodology is demonstrated for frequency domain identification of large space structures which systematically transforms experimental raw data into a form required for synthesizing H(infinity) controllers using modern robust control design software (e.g., Matlab Toolboxes). A unique feature of this approach is that the additive uncertainty is characterized to a specified statistic confidence rather than with hard bounds. In this study, the difference in robust performance is minimal between the two levels of confidence. In general cases, the present methodology provides a tool for performance/confidence level tradeoff studies. For simplicity, the additive uncertainty on a frequency grid is considered and the interpolation error in between grid points is neglected.
Three major goals were delineated: (1) to develop a general method for determining the response of a structure to combined base and acoustic random excitation: (2) to develop parametric relationships to aid in the design of plates which are subjected to random force or random base excitation: (3) to develop a method to identify the individual acoustic and base input to a structure with only a limited number of measurement channels, when both types of excitation act simultaneously.
It is shown that identification of modes with practically the same frequency but with different mode shapes can be carried out in experimental settings. Using multiple-input multiple-output data, distinct modes whose FRFs (frequency response functions) appear as single peaks can be distinguished from their input and output weights. When the peak is modeled with a single mode, some input/output channels may have good peaks, but others will not. Addition of more modes and use of multiple input data result in good fits for all channels if the system behaves in a linear modal manner.
Recent experiences in the field of flexible structure control in space have indicated a need for on-orbit system identification to support robust control redesign to avoid in-flight instabilities and maintain high spacecraft performance. This paper highlights an automated frequency domain system identification methodology recently developed to fulfill this need. The methodology is focused to support (1) the estimation of system quantities useful for robust control analysis and design; (2) experiment design tailored to performing system identification in a typically constrained on-orbit environment; and (3) the automation of operations to reduce 'human in the loop' requirements.
The authors present a novel method called time-varying transfer function estimation (TTFE) in which time-domain parameters are computed through identification in the frequency domain. The method is particularly well suited for flexible structure control problems. An example of a flexible manipulator system is presented for which a self-tuning control law with frequency shaping is derived and demonstrated.