American Institute of Mathematical Sciences

October  2018, 14(4): 1579-1594. doi: 10.3934/jimo.2018022

Adjoint-based parameter and state estimation in 1-D magnetohydrodynamic (MHD) flow system

 1 School of Automation, Guangdong University of Technology, Guangzhou, Guangdong, China 2 School of Mathematical Sciences, Zhejiang University, Hangzhou, Zhejiang, China 3 School of Control Science and Engineering, Shandong University, Jinan, Shandong, China

* Corresponding author: Zongze Wu

Received  April 2017 Revised  October 2017 Published  January 2018

Fund Project: This work is supported by the National Natural Science Foundation of China grants (61703114,61673126,61473253) and the Open Research Project of the State Key Laboratory of Industrial Control Technology, Zhejiang University, China (ICT170288, ICT170301).

In this paper, an adjoint-based optimization method is employed to estimate the unknown coefficients and states arising in an one-dimensional (1-D) magnetohydrodynamic (MHD) flow, whose dynamics can be modeled by a coupled partial differential equations (PDEs) under some suitable assumptions. In this model, the coefficients of the Reynolds number and initial conditions as well as state variables are supposed to be unknown and need to be estimated. We first employ the Lagrange multiplier method to connect the dynamics of the 1-D MHD system and the cost functional defined as the least square errors between the measurements in the experiment and the numerical simulation values. Then, we use the adjoint-based method to the augmented Lagrangian cost functional to get an adjoint coupled PDEs system, and the exact gradients of the defined cost functional with respect to these unknown parameters and initial states are further derived. The existed gradient-based optimization technique such as sequential quadratic programming (SQP) is employed for minimizing the cost functional in the optimization process. Finally, we illustrate the numerical examples to verify the effectiveness of our adjoint-based estimation approach.

Citation: Zhigang Ren, Shan Guo, Zhipeng Li, Zongze Wu. Adjoint-based parameter and state estimation in 1-D magnetohydrodynamic (MHD) flow system. Journal of Industrial & Management Optimization, 2018, 14 (4) : 1579-1594. doi: 10.3934/jimo.2018022
References:

show all references

References:
Estimation of initial condition $u(x,0)$ (scenario 1).
Estimation of initial condition $B(x,0)$ (scenario 1).
Iteration process of cost function value (scenario 1).
Estimation of state $u(x,t)$ (scenario 1).
Estimation of state $B(x,t)$ (scenario 1).
Estimation of initial condition $u(x,0)$ (scenario 2).
Estimation of initial condition $B(x,0)$ (scenario 2).
Iteration process of cost function value (scenario 2).
Estimation of state $u(x,t)$ (scenario 2).
Estimation of state $B(x,t)$ (scenario 2).
 Algorithm 1. Gradient-based optimization procedure for solving Problem P$_0$. Step 1: Choose the initial guess $\nu$, $\nu_m$ and $u_{0}^{in}(x)$, $B_{0}^{in}(x)$. Step 2: Solve the 1-D MHD model (1) forward in time corresponding to initial guess values from $t=0$ to $t=T$, and solve the adjoint coupled PDE systems (7) backward in time corresponding to initial guess values from $t=T$ to $t=0$ to obtain $u(x, t)$, $B(x, t)$, $\lambda_1(x, t)$, $\lambda_2(x, t)$. Step 3: Compute cost functional and its gradients according to (4) and (5)-(6). Step 4: Use the gradients information obtained in $\mathbf{Step 3}$ to perform an optimality test. If $\nu$, $\nu_m$ and $u_{0}^{in}(x), B_0^{in}(x)$ are optimal, then exit; Otherwise, go to $\mathbf{Step 5}$. Step 5: Use the gradient information obtained in $\mathbf{Step 3}$ to calculate a searching direction. Step 6: Perform a line search to determine the optimal step length. Step 7: Compute the new points $\nu$, $\nu_m$ and $u_{0}^{in}(x)$, $B_{0}^{in}(x)$ and return to $\mathbf{Step 2}$.
 Algorithm 1. Gradient-based optimization procedure for solving Problem P$_0$. Step 1: Choose the initial guess $\nu$, $\nu_m$ and $u_{0}^{in}(x)$, $B_{0}^{in}(x)$. Step 2: Solve the 1-D MHD model (1) forward in time corresponding to initial guess values from $t=0$ to $t=T$, and solve the adjoint coupled PDE systems (7) backward in time corresponding to initial guess values from $t=T$ to $t=0$ to obtain $u(x, t)$, $B(x, t)$, $\lambda_1(x, t)$, $\lambda_2(x, t)$. Step 3: Compute cost functional and its gradients according to (4) and (5)-(6). Step 4: Use the gradients information obtained in $\mathbf{Step 3}$ to perform an optimality test. If $\nu$, $\nu_m$ and $u_{0}^{in}(x), B_0^{in}(x)$ are optimal, then exit; Otherwise, go to $\mathbf{Step 5}$. Step 5: Use the gradient information obtained in $\mathbf{Step 3}$ to calculate a searching direction. Step 6: Perform a line search to determine the optimal step length. Step 7: Compute the new points $\nu$, $\nu_m$ and $u_{0}^{in}(x)$, $B_{0}^{in}(x)$ and return to $\mathbf{Step 2}$.
 [1] Armin Lechleiter, Tobias Rienmüller. Factorization method for the inverse Stokes problem. Inverse Problems & Imaging, 2013, 7 (4) : 1271-1293. doi: 10.3934/ipi.2013.7.1271 [2] Marion Darbas, Jérémy Heleine, Stephanie Lohrengel. Numerical resolution by the quasi-reversibility method of a data completion problem for Maxwell's equations. Inverse Problems & Imaging, 2020, 14 (6) : 1107-1133. doi: 10.3934/ipi.2020056 [3] Alexandr Mikhaylov, Victor Mikhaylov. Dynamic inverse problem for Jacobi matrices. Inverse Problems & Imaging, 2019, 13 (3) : 431-447. doi: 10.3934/ipi.2019021 [4] Deren Han, Zehui Jia, Yongzhong Song, David Z. W. Wang. An efficient projection method for nonlinear inverse problems with sparsity constraints. Inverse Problems & Imaging, 2016, 10 (3) : 689-709. doi: 10.3934/ipi.2016017 [5] Habib Ammari, Josselin Garnier, Vincent Jugnon. Detection, reconstruction, and characterization algorithms from noisy data in multistatic wave imaging. Discrete & Continuous Dynamical Systems - S, 2015, 8 (3) : 389-417. doi: 10.3934/dcdss.2015.8.389 [6] Guirong Jiang, Qishao Lu. The dynamics of a Prey-Predator model with impulsive state feedback control. Discrete & Continuous Dynamical Systems - B, 2006, 6 (6) : 1301-1320. doi: 10.3934/dcdsb.2006.6.1301 [7] Yunfei Lv, Rong Yuan, Yuan He. Wavefronts of a stage structured model with state--dependent delay. Discrete & Continuous Dynamical Systems - A, 2015, 35 (10) : 4931-4954. doi: 10.3934/dcds.2015.35.4931 [8] Daoyuan Fang, Ting Zhang. Compressible Navier-Stokes equations with vacuum state in one dimension. Communications on Pure & Applied Analysis, 2004, 3 (4) : 675-694. doi: 10.3934/cpaa.2004.3.675 [9] Hildeberto E. Cabral, Zhihong Xia. Subharmonic solutions in the restricted three-body problem. Discrete & Continuous Dynamical Systems - A, 1995, 1 (4) : 463-474. doi: 10.3934/dcds.1995.1.463 [10] Qiang Guo, Dong Liang. An adaptive wavelet method and its analysis for parabolic equations. Numerical Algebra, Control & Optimization, 2013, 3 (2) : 327-345. doi: 10.3934/naco.2013.3.327 [11] Michel Chipot, Mingmin Zhang. On some model problem for the propagation of interacting species in a special environment. Discrete & Continuous Dynamical Systems - A, 2020  doi: 10.3934/dcds.2020401 [12] Fritz Gesztesy, Helge Holden, Johanna Michor, Gerald Teschl. The algebro-geometric initial value problem for the Ablowitz-Ladik hierarchy. Discrete & Continuous Dynamical Systems - A, 2010, 26 (1) : 151-196. doi: 10.3934/dcds.2010.26.151 [13] Gloria Paoli, Gianpaolo Piscitelli, Rossanno Sannipoli. A stability result for the Steklov Laplacian Eigenvalue Problem with a spherical obstacle. Communications on Pure & Applied Analysis, 2021, 20 (1) : 145-158. doi: 10.3934/cpaa.2020261 [14] Tao Wu, Yu Lei, Jiao Shi, Maoguo Gong. An evolutionary multiobjective method for low-rank and sparse matrix decomposition. Big Data & Information Analytics, 2017, 2 (1) : 23-37. doi: 10.3934/bdia.2017006 [15] Boris Kramer, John R. Singler. A POD projection method for large-scale algebraic Riccati equations. Numerical Algebra, Control & Optimization, 2016, 6 (4) : 413-435. doi: 10.3934/naco.2016018 [16] Petra Csomós, Hermann Mena. Fourier-splitting method for solving hyperbolic LQR problems. Numerical Algebra, Control & Optimization, 2018, 8 (1) : 17-46. doi: 10.3934/naco.2018002 [17] Christina Surulescu, Nicolae Surulescu. Modeling and simulation of some cell dispersion problems by a nonparametric method. Mathematical Biosciences & Engineering, 2011, 8 (2) : 263-277. doi: 10.3934/mbe.2011.8.263 [18] Min Li. A three term Polak-Ribière-Polyak conjugate gradient method close to the memoryless BFGS quasi-Newton method. Journal of Industrial & Management Optimization, 2020, 16 (1) : 245-260. doi: 10.3934/jimo.2018149 [19] Rongchang Liu, Jiangyuan Li, Duokui Yan. New periodic orbits in the planar equal-mass three-body problem. Discrete & Continuous Dynamical Systems - A, 2018, 38 (4) : 2187-2206. doi: 10.3934/dcds.2018090 [20] Manfred Einsiedler, Elon Lindenstrauss. On measures invariant under diagonalizable actions: the Rank-One case and the general Low-Entropy method. Journal of Modern Dynamics, 2008, 2 (1) : 83-128. doi: 10.3934/jmd.2008.2.83

2019 Impact Factor: 1.366