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.
Invertibility of linear time invariant dynamical control systems, discussing L-inverses, sequential circuits, inherent integration, reproducibility, pointwise and functional inverses
The concept of a 'best approximating m-dimensional subspace' for a given set of vectors in n-dimensional whole space is introduced. Such a subspace is easily described in terms of the eigenvectors of an associated Gram matrix. This technique is used to approximate an achievable set for a discrete linear time-invariant dynamical system. This approximation characterizes the part of the state space that may be reached using modest levels of control. If the achievable set can be closely approximated by a proper subspace of the whole space then the system is 'essentially uncontrollable'. The notion finds application in studies of failure-tolerant systems, and in decoupling.
In this work, we present an efficient and scalable algorithm for performing matrix-vector multiplications (matvecs) for block Toeplitz matrices. Such matrices, which are shift-invariant with respect to their blocks, arise in the context of solving inverse problems governed by autonomous systems, and time-invariant systems in particular. In this article, we consider inverse problems that infer unknown parameters from observational data of a linear time-invariant dynamical system given in the form of partial differential equations (PDEs). Matrix-free Newton-conjugate-gradient methods are often the gold standard for solving these inverse problems, but they require numerous actions of the Hessian on a vector. Matrix-free adjoint-based Hessian matvecs require solution of a pair of linearized forward/adjoint PDE solves per Hessian action, which may be prohibitive for large-scale inverse problems. Time invariance of the forward PDE problem leads to a block Toeplitz structure of the discretized parameter-to-observable (p2o) map defining the mapping from inputs (parameters) to outputs (observables) of the PDEs. This block Toeplitz structure enables us to exploit two key properties: (1) compact storage of the p2o map and its adjoint, and (2) efficient fast Fourier transform–based Hessian matvecs. The proposed algorithm is mapped onto large multi-GPU clusters and achieves more than 80% of peak bandwidth on NVIDIA A100 GPUs. Excellent weak scaling is shown for up to 48 A100 GPUs. For the targeted problems, the implementation executes Hessian matvecs within fractions of a second, which is orders of magnitude faster than can be achieved by conventional matrix-free Hessian matvecs via forward/adjoint PDE solves.
We present a goal-oriented framework for constructing digital twins with the following properties: (1) they employ discretizations of high-fidelity partial differential equation (PDE) models governed by autonomous dynamical systems, leading to large-scale forward problems; (2) they solve a linear inverse problem to assimilate observational data to infer uncertain model components followed by a forward prediction of the evolving dynamics; and (3) the entire end-to-end, data-to-inference-to-prediction computation is carried out without approximation and in real time through a Bayesian framework that rigorously accounts for uncertainties. Several challenges must be overcome to realize this framework, including the large scale of the forward problem, the high dimensionality of the parameter space, and for a class of problems including those we target, the slow decay of the singular values of the parameter-to-observable map. Here we introduce a methodology to overcome these challenges by exploiting the autonomous structure of the forward model to decompose the solution of the inverse problem into a one-time-only offline phase in which the PDE model is solved a limited number of times (equal to the number of sensors), and an online phase that maps well onto GPUs and computes the parameter inference and prediction of quantities of interest in real time, given observational data. Our ultimate goal is to apply this framework to construct digital twins for subduction zones, including Cascadia, to provide early warning for tsunamis generated by megathrust earthquakes. To this end, we demonstrate how our methodology can be used to employ seafloor pressure observations, along with the coupled acoustic–gravity wave equations, to infer the earthquake-induced spatiotemporal seafloor motion (discretized with $\mathscr{O}$ (10 9 ) parameters) and forward predict the tsunami propagation. We present results of an end-to-end inference, prediction, and uncertainty quantification for a representative test problem with $\mathscr{O}$ (10 8 ) inversion parameters for which goal-oriented Bayesian inference is accomplished exactly and in real time, that is, in a matter of seconds.
Optimal fixed dimensionality dynamic compensator design for linear time-invariant closed-loop system based on quadratic cost and gain criteria
Determining optimal time-invariant output feedback controllers for linear dynamic systems
A method is derived for digital computer simulation of linear time-invariant systems when the insignificant eigenvalues involved in such systems are eliminated by an ALSAP root removal technique. The method is applied to a thirteenth-order dynamic system representing a passive RLC network.
Digital computer program written in FORTRAN 4 has capability of synthesizing low order compensators to stabilize dynamical system in presence of parameter variations. Design scheme is accomplished by means of sensitivity function which is minimized with respect to eigenvalues.
Use is made of linearized error characteristic equation for model-reference adaptive systems to determine a parameter adjustment rule for obtaining time-invariant error dynamics. Theoretical justification of error stability is given and an illustrative example included to demonstrate the utility of the proposed technique.
Conventional observers for linear time-invariant systems are shown to be structurally inadequate from a sensitivity standpoint. It is proved that if a linear dynamic system is to provide observer action despite arbitrary small perturbations in a specified subset of its parameters, it must: (1) be a closed loop system, be driven by the observer error, (2) possess redundancy, the observer must be generating, implicitly or explicitly, at least one linear combination of states that is already contained in the measurements, and (3) contain a perturbation-free model of the portion of the system observable from the external input to the observer. The procedure for design of robust observers possessing the above structural features is established and discussed.
The control of linear continuous dynamical systems is investigated as a problem of limited state feedback control. The equations which describe the structure of an observer are developed constrained to time-invarient systems. The optimal control problem is formulated, accounting for the uncertainty in the design parameters. Expressions for bounds on closed loop stability are also developed. The results indicate that very little uncertainty may be tolerated before divergence occurs in the recursive computation algorithms, and the derived stability bound yields extremely conservative estimates of regions of allowable parameter variations.
Attitude control concepts for use onboard structurally nonrigid spacecraft that must be pointed with great precision are examined. The task of determining the eigenproperties of a system of linear time-invariant equations (in terms of hybrid coordinates) representing the attitude motion of a flexible spacecraft is discussed. Literal characteristics are developed for the associated eigenvalues and eigenvectors of the system. A method is presented for determining the poles and zeros of the transfer function describing the attitude dynamics of a flexible spacecraft characterized by hybrid coordinate equations. Alterations are made to linear regulator and observer theory to accommodate modeling errors. The results show that a model error vector, which evolves from an error system, can be added to a reduced system model, estimated by an observer, and used by the control law to render the system less sensitive to uncertain magnitudes and phase relations of truncated modes and external disturbance effects. A hybrid coordinate formulation using the provided assumed mode shapes, rather than incorporating the usual finite element approach is provided.
The development of an active, momentum-exchange system to be used for attitude stabilization of a class of cable-connected artificial-g space stations is studied. A system which employs a single movable control mass is examined for the control of a space station which has the physical appearance of two cylinders connected axially by cables. The dynamic model for the space station includes its aggregate rigid body rotation and relative torsional rotation between the bodies. A zero torsional stiffness design (one cable) and a maximum torsional stiffness design (eight cables) are examined in various stages of deployment, for selected spin velocities ranging from 4 rpm upwards. A linear, time-invariant, feed-back control system is employed, with gains calculated via a root-specification procedure. The movable mass controller provides critical wobble-damping capability for the crew quarters for all configurations and spin velocity.
A special class of decentralized control problem is discussed in which the objectives of the control agents are to steer the state of the system to desired levels. Each agent is concerned about certain aspects of the state of the entire system. The state and control equations are given for linear time-invariant systems. Stability and coordination, and the optimization of decentralized control are analyzed, and the information structure design is presented.
The properties of reachable sets for linear dynamical systems for specified control sets are discussed. Iterative procedures for determining numerical approximations of the reachable set are suggested and methods of obtaining an admissible control function which transfers an initial state to as near a prescribed target as possible is described. The problem of reachability with multiple control constraints is discussed and certain aspects of reachability for time-invariant systems with adjustable parameters is considered.
The relations characterizing the optimal minimal-order observer-based compensators for a linear time-invariant multivariable system in a certain canonical form, with a random initial state (or equivalently, known initial state with white driving noise) have been reported by Miller and independently by others. In this note, we establish in general that the plant transfer function uniquely determines the optimal compensator transfer function, and that this characterizes precisely the degrees of freedom in the compensator design. Any two realizations of the optimal compensator dynamics yield the same performance.
One of the more significant technological problems associated with the orbital operation of large astronomical telescope's is the fabrication and maintenance of the primary mirror surface to the tolerance required for diffraction-limited performance. An interesting approach to the solution of this problem involves continuously measuring and automatically correcting the optical surface of a thin deformable mirror by means of discrete actuators located on its rear surface: The realization of diffraction-limited performance from a telescope in space by this method rests on the ability of the designer to achieve extremely accurate control of a highly complex, interacting, multivariable system. This paper presents the results of a detailed study of the discrete control of linear distributed systems with specific application to the design of a practical controller for a plant representative of a telescope primary mirror for an orbiting astronomical observatory. The problem of controlling the distributed plant is treated by employing modal techniques to represent variations in the optical figure. Distortion of the mirror surface, which arises primarily from thermal gradients, is countered by actuators working against a backing structure to apply a corrective force distribution to the controlled surface. Each displacement actuator is in series with a spring attached to the mirror by means of a pad intentionally introduced to restrict the excitation of high-order modes. Control is then exerted over a finite number (equal.to the number of actuators) of the most significant modes. Through the application of the modal expansion technique the mirror equation of motion is transformed tb a set of uncoupled, linear, time-invariant, ordinary differential equations. The desired dynamic response and static accuracy may then be achieved by the application of classical single-variable design techniques. The formulation of a quadratic performance index which incorporates a measure of image quality permits determination of the trade-off between the-number of actuators and optical purity. A criterion for defining actuator placement and pad size is presented which minimizes the tendency of the controller to excite the unmonitored modes.
Optimized static and dynamic linear system compensator designs with a priori fixed dimensions are discussed. Expressions are derived to determine the parameters of these designs, including the fixed-gain output feedback, the time-varying output feedback, and time-invariant compensation. A relevant deterministic off-line optimization problem is formulated, and the properties of a solution to the problem are analyzed. A unified design philosophy for limited state-variable control problems of essentially deterministic systems is developed. The philosophy incorporates design constraints, a careful selection of performance criteria, and interpretation of the results.