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.
Parameter identification of the mathematical model of spiral-drill wear in deep drilling is considered. Wear is simulated by piecewise-linear approximation of the growth of drag torque in drilling. The diagnostic index selected is the constant component of the drag torque after the onset of drilling. The significance of the diagnostic index in determining the wear is demonstrated.
In many mechanistic medical, biological, physical, and engineered spatiotemporal dynamic models the numerical solution of partial differential equations (PDEs), especially for diffusion, fluid flow and mechanical relaxation, can make simulations impractically slow. Biological models of tissues and organs often require the simultaneous calculation of the spatial variation of concentration of dozens of diffusing chemical species. One clinical example where rapid calculation of a diffusing field is of use is the estimation of oxygen gradients in the retina, based on imaging of the retinal vasculature, to guide surgical interventions in diabetic retinopathy. Furthermore, the ability to predict blood perfusion and oxygenation may one day guide clinical interventions in diverse settings, i.e., from stent placement in treating heart disease to BOLD fMRI interpretation in evaluating cognitive function (Xie et al., 2019; Lee et al., 2020). Since the quasi-steady-state solutions required for fast-diffusing chemical species like oxygen are particularly computationally costly, we consider the use of a neural network to provide an approximate solution to the steady-state diffusion equation. Machine learning surrogates, neural networks trained to provide approximate solutions to such complicated numerical problems, can often provide speed-ups of several orders of magnitude compared to direct calculation. Surrogates of PDEs could enable use of larger and more detailed models than are possible with direct calculation and can make including such simulations in real-time or near-real time workflows practical. Creating a surrogate requires running the direct calculation tens of thousands of times to generate training data and then training the neural network, both of which are computationally expensive. Often the practical applications of such models require thousands to millions of replica simulations, for example for parameter identification and uncertainty quantification, each of which gains speed from surrogate use and rapidly recovers the up-front costs of surrogate generation. We use a Convolutional Neural Network to approximate the stationary solution to the diffusion equation in the case of two equal-diameter, circular, constant-value sources located at random positions in a two-dimensional square domain with absorbing boundary conditions. Such a configuration caricatures the chemical concentration field of a fast-diffusing species like oxygen in a tissue with two parallel blood vessels in a cross section perpendicular to the two blood vessels. To improve convergence during training, we apply a training approach that uses roll-back to reject stochastic changes to the network that increase the loss function. The trained neural network approximation is about 1000 times faster than the direct calculation for individual replicas. Because different applications will have different criteria for acceptable approximation accuracy, we discuss a variety of loss functions and accuracy estimators that can help select the best network for a particular application. We briefly discuss some of the issues we encountered with overfitting, mismapping of the field values and the geometrical conditions that lead to large absolute and relative errors in the approximate solution.
Recursive method for the optimal smoothing of data has numerical superiority and is more easily understood in terms of physical reasoning than earlier methods. Using a Kalman filter, the smoothing technique, applied to a nonlinear parameter identification problem, is useful in those situations where linearization about a reference solution is valid.
A procedure for estimating noise statistics as well as system model parameters, based on a maximum likelihood criterion, has been successfully applied to a number of aircraft parameter-identification problems. In the present investigation, an a posteriori criterion is examined for use in such an application, when it is desired to model process noise. The criteria lead to identical results when there is no process noise.
A sequential estimator is presented and demonstrated which successfully tracks the system state and model errors in the presence of significant and unpredictable system or environmental variations. This adaptive estimation concept is shown to lead to a new and attractive approach to parameter identification problems. Simulation results are presented in an orbit determination problem, where the estimator tracks the orbit and unmodeled accelerations due to errors in the geopotential model. Some results in a re-entry trajectory estimation problem are also summarized.
A sequential estimator is presented and demonstrated which successfully tracks the system state and model errors in the presence of significant and unpredictable system or environmental variations. This adaptive estimation concept is shown to lead to a new and attractive approach to parameter identification problems. Simulation results are presented in an orbit determination problem and in a reentry trajectory estimation problem.
Techniques for the measurement of the dynamic stability of linear systems are discussed. Particular attention is given to an analysis of the errors in the procedures, and to methods for calculating the system damping from the data. The techniques discussed include: transient decay, moving block analysis, spectral analysis, random decrement signatures, transfer function analysis, and parameter identification methods. The special problems of rotorcraft dynamic stability testing are discussed.
Specific test requirements related to the wind tunnel testing of the XV-15 advanced tilt rotor research aircraft were determined. The following analytical tools were developed: (1) digital simulation of the XV-15, incorporating a simplified tunnel support model, control system loop, measurement lags, gust disturbances, and sensor noise, (2) specialization of existing data analysis programs to the high order XV-15 dynamical model (transfer function program, a time series analysis program, an advanced maximum likelihood parameter identification program), (3) several auxiliary programs to provide estimates of damping from transfer functions as well as calculations of model decomposition of system response. The following results were discussed: (1) modelling of the aircraft, instrumentation, and controls, (2) results of the rotor/cantilever wing model and coupled wing, (3) examples of data prediction with system identification techniques, and (4) detailed conclusions and recommendations.
A design study of adaptive control logic suitable for implementation in modern airborne digital flight computers was conducted. Both explicit controllers which directly utilize parameter identification and implicit controllers which do not require identification were considered. Extensive analytical and simulation efforts resulted in the recommendation of two explicit digital adaptive flight controllers. Interface weighted least squares estimation procedures with control logic were developed using either optimal regulator theory or with control logic based upon single stage performance indices.
This paper considers the design of flight control inputs which result in a minimization of the nonuniqueness problem of parameter identification due to statistical correlation. The performance index for the control design is taken as a linear weighted sum of the squares of the correlation coefficients of the aerodynamic stability and control derivatives as computed from the Cramer-Rao lower bound matrix. An optimal input design is demonstrated for an F-8 aircraft with supercritical wing. Reduced correlations are noted and corresponding estimates and confidence levels presented.
A number of current research directions in the fields of digital signal processing and modern control and estimation theory were studied. Topics such as stability theory, linear prediction and parameter identification, system analysis and implementation, two-dimensional filtering, decentralized control and estimation, image processing, and nonlinear system theory were examined in order to uncover some of the basic similarities and differences in the goals, techniques, and philosophy of the two disciplines. An extensive bibliography is included.
The relative merits of sinusoidal excitation versus random atmospheric turbulence was investigated. The randomdec and autocorrelation methods were used to analyze data from a Learjet flight test. A parameter identification digital program, using least squares approach, was developed to determine the aeroelastic characteristics of a two mode system. The flight test program, computer program, and data reduction procedure is presented. Final results of the two modes of excitation obtained by Randomdec method are discussed.
A simplified aerodynamic force model based on the physical principle of Prandtl's lifting line theory and trailing vortex concept has been developed to account for unsteadiness in the aircraft dynamics. The wake is assumed to be compressed to a single shed vortex element of appropriate strength moving downstream at a speed sufficient to approximate the Wagner function. Results are presented illustrating the ability of the simplified theory to duplicate exact solutions in unsteady aerodynamics. Further, consideration is given to the utility of the model in a parameter identification application.
Research directions in the fields of digital signal processing and modern control and estimation theory are discussed. Stability theory, linear prediction and parameter identification, system synthesis and implementation, two-dimensional filtering, decentralized control and estimation, and image processing are considered in order to uncover some of the basic similarities and differences in the goals, techniques, and philosophy of the disciplines.
A technique was developed which permits simultaneous extraction of complete lift, drag, and thrust power curves from time histories of a single aircraft maneuver such as a pullup (from V sub max to V sub stall) and pushover (to sub V max for level flight.) The technique is an extension to non-linear equations of motion of the parameter identification methods of lliff and Taylor and includes provisions for internal data compatibility improvement as well. The technique was show to be capable of correcting random errors in the most sensitive data channel and yielding highly accurate results. This technique was applied to flight data taken on the ATLIT aircraft. The drag and power values obtained from the initial least squares estimate are about 15% less than the 'true' values. If one takes into account the rather dirty wing and fuselage existing at the time of the tests, however, the predictions are reasonably accurate. The steady state lift measurements agree well with the extracted values only for small values of alpha. The predicted value of the lift at alpha = 0 is about 33% below that found in steady state tests while the predicted lift slope is 13% below the steady state value.
This paper is a review of the current state of aerodynamic mathematical modeling for aircraft motions at high angles of attack. The mathematical model serves to define a set of characteristic motions from whose known aerodynamic responses the aerodynamic response to an arbitrary high angle-of-attack flight maneuver can be predicted. Means are explored of obtaining stability parameter information in terms of the characteristic motions, whether by wind-tunnel experiments, computational methods, or by parameter-identification methods applied to flight-test data. A rationale is presented for selecting and verifying the aerodynamic mathematical model at the lowest necessary level of complexity. Experimental results describing the wing-rock phenomenon are shown to be accommodated within the most recent mathematical model by admitting the existence of aerodynamic hysteresis in the steady-state variation of the rolling moment with roll angle. Interpretation of the experimental results in terms of bifurcation theory reveals the general conditions under which aerodynamic hysteresis must exist.
A technique was developed which permits simultaneous extraction of complete lift, drag, and thrust power curves from time histories of a single aircraft maneuver such as a pull up (from V max to V stall) and pushover (to V max for level flight). The technique, which is an extension of nonlinear equations of motion of the parameter identification methods of Iliff and Taylor and includes provisions for internal data compatibility improvement as well, was shown to be capable of correcting random errors in the most sensitive data channel and yielding highly accurate results. Flow charts, listings, sample inputs and outputs for the relevant routines are provided as appendices. This technique was applied to flight data taken on the ATLIT aircraft. Lack of adequate knowledge of the correct full throttle thrust horsepower true airspeed variation and considerable internal data inconsistency made it impossible to apply the trajectory matching features of the technique.
The overall remotely piloted drop model operation, descriptions, instrumentation, launch and recovery operations, piloting concept, and parameter identification methods are discussed. Static and dynamic stability derivatives were obtained for an angle attack range from -20 deg to 53 deg. It is indicated that the variations of the estimates with angle of attack are consistent for most of the static derivatives, and the effects of configuration modifications to the model were apparent in the static derivative estimates.