An official website of the United States government
Abstract Many real-world systems modeled using differential equations involve unknown or uncertain parameters. Standard approaches to address parameter estimation inverse problems in this setting typically focus on estimating constants; yet some unobservable system parameters may vary with time without known evolution models. In this work, we propose a novel approximation method inspired by the Fourier series to estimate time-varying parameters in deterministic dynamical systems modeled with ordinary differential equations. Using ensemble Kalman filtering in conjunction with Fourier series-based approximation models, we detail two possible implementation schemes for sequentially updating the time-varying parameter estimates given noisy observations of the system states. We demonstrate the capabilities of the proposed approach in estimating periodic parameters, both when the period is known and unknown, as well as non-periodic time-varying parameters of different forms with several computed examples using a forced harmonic oscillator. Results emphasize the importance of the frequencies and number of approximation model terms on the time-varying parameter estimates and corresponding dynamical system predictions.
more »
« less
Using Monte Carlo Particle Methods to Estimate and Quantify Uncertainty in Periodic Parameters
Estimating and quantifying uncertainty in system parameters remains a big challenge in applied and computational mathematics. A subset of these problems includes estimating periodic parameters that have unknown dynamics. Along with their time series, the period of these parameters may also be unknown and need to be estimated. The aim of this paper is to address the periodic parameter estimation problem, with particular focus on exploring the associated uncertainty, using Monte Carlo particle methods, such as the ensemble Kalman filter. Both parameter tracking and piecewise function approximations of periodic parameters are considered, highlighting aspects of parameter uncertainty in each approach when considering factors such as the frequency of available data and the number of piecewise segments used in the approximation. Estimation of the period of the periodic parameters and related uncertainty is also analyzed in the piecewise formulation. The pros and cons of each approach are discussed relative to a numerical example estimating the external voltage parameter in the FitzHugh-Nagumo system for modeling the spiking dynamics of neurons.
more »
« less
- Award ID(s):
- 1819203
- PAR ID:
- 10174863
- Date Published:
- Journal Name:
- Advances in Mathematical Sciences: AWM Research Symposium, Houston, TX, April 2019
- Volume:
- 21
- Page Range / eLocation ID:
- 213-226
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Estimating and quantifying uncertainty in unknown system parameters from limited data remains a challenging inverse problem in a variety of real-world applications. While many approaches focus on estimating constant parameters, a subset of these problems includes time-varying parameters with unknown evolution models that often cannot be directly observed. This work develops a systematic particle filtering approach that reframes the idea behind artificial parameter evolution to estimate time-varying parameters in nonstationary inverse problems arising from deterministic dynamical systems. Focusing on systems modeled by ordinary differential equations, we present two particle filter algorithms for time-varying parameter estimation: one that relies on a fixed value for the noise variance of a parameter random walk; another that employs online estimation of the parameter evolution noise variance along with the time-varying parameter of interest. Several computed examples demonstrate the capability of the proposed algorithms in estimating time-varying parameters with different underlying functional forms and different relationships with the system states (i.e. additive vs. multiplicative).more » « less
-
Abstract Soil moisture is an important driver of growth in boreal Alaska, but estimating soil hydraulic parameters can be challenging in this data‐sparse region. Parameter estimation is further complicated in regions with rapidly warming climate, where there is a need to minimize model error dependence on interannual climate variations. To better identify soil hydraulic parameters and quantify energy and water balance and soil moisture dynamics, we applied the physically based, one‐dimensional ecohydrological Simultaneous Heat and Water (SHAW) model, loosely coupled with the Geophysical Institute of Permafrost Laboratory (GIPL) model, to an upland deciduous forest stand in interior Alaska over a 13‐year period. Using a Generalized Likelihood Uncertainty Estimation parameterisation, SHAW reproduced interannual and vertical spatial variability of soil moisture during a five‐year validation period quite well, with root mean squared error (RMSE) of volumetric water content at 0.5 m as low as 0.020 cm3/cm3. Many parameter sets reproduced reasonable soil moisture dynamics, suggesting considerable equifinality. Model performance generally declined in the eight‐year validation period, indicating some overfitting and demonstrating the importance of interannual variability in model evaluation. We compared the performance of parameter sets selected based on traditional performance measures such as the RMSE that minimize error in soil moisture simulation, with one that is designed to minimize the dependence of model error on interannual climate variability using a new diagnostic approach we call CSMP, which stands for Climate Sensitivity of Model Performance. Use of the CSMP approach moderately decreases traditional model performance but may be more suitable for climate change applications, for which it is important that model error is independent from climate variability. These findings illustrate (1) that the SHAW model, coupled with GIPL, can adequately simulate soil moisture dynamics in this boreal deciduous region, (2) the importance of interannual variability in model parameterisation, and (3) a novel objective function for parameter selection to improve applicability in non‐stationary climates.more » « less
-
Algorithms often have tunable parameters that impact performance metrics such as runtime and solution quality. For many algorithms used in practice, no parameter settings admit meaningful worst-case bounds, so the parameters are made available for the user to tune. Alternatively, parameters may be tuned implicitly within the proof of a worst-case approximation ratio or runtime bound. Worst-case instances, however, may be rare or nonexistent in practice. A growing body of research has demonstrated that a data-driven approach to parameter tuning can lead to significant improvements in performance. This approach uses atraining setof problem instances sampled from an unknown, application-specific distribution and returns a parameter setting with strong average performance on the training set. We provide techniques for derivinggeneralization guaranteesthat bound the difference between the algorithm’s average performance over the training set and its expected performance on the unknown distribution. Our results apply no matter how the parameters are tuned, be it via an automated or manual approach. The challenge is that for many types of algorithms, performance is a volatile function of the parameters: slightly perturbing the parameters can cause a large change in behavior. Prior research [e.g.,12,16,20,62] has proved generalization bounds by employing case-by-case analyses of greedy algorithms, clustering algorithms, integer programming algorithms, and selling mechanisms. We streamline these analyses with a general theorem that applies whenever an algorithm’s performance is a piecewise-constant, piecewise-linear, or—more generally—piecewise-structuredfunction of its parameters. Our results, which are tight up to logarithmic factors in the worst case, also imply novel bounds for configuring dynamic programming algorithms from computational biology.more » « less
-
Tethered capture of space debris is a promising method of Active Debris Removal, but has numerous well-known challenges in the post-capture phase. Control of the system in this phase is complicated by nonlinear dynamics with the potential of chaotic motion, unknown debris parameters, and debris tumbling. The inherent uncertainty present in the system and the need for control often necessitate the estimation of debris states and parameters such that the post-capture system can remain controlled. To this end, a relative distance Proportional-Integral-Derivative control and an Unscented Kalman Filter are implemented during the post-capture phase of an ADR mission. It is assumed that some debris (target) states and properties are unknown to the chaser, requiring the estimation of the attitude, angular rates, and principal moments of inertia of the target. For estimating these states, the UKF uses measurements of the tether tension magnitude and pixel coordinates of feature points on the target provided by a camera mounted on the chaser. Both estimation and control are done simultaneously, simulating online estimation and control of an ADR mission. The PID control was found to maintain safe conditions when using the estimated states in two separate Monte-Carlo simulations, differing in the measurement frequency of the pixel coordinates, while the estimation of the principal moments of inertia of the debris was satisfactory.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1007/978-3-030-42687-3_14
