TY - GEN
T1 - Physics informed stochastic grey-box model of the flow-front in a vacuum assisted resin transfer moulding process with missing data
AU - Relan, Rishi
AU - Junker, Rune Grønborg
AU - Nauheimer, Michael
AU - Thygesen, Uffe Høgsbro
AU - Lindström, Erik
AU - Madsen, Henrik
PY - 2021
Y1 - 2021
N2 - Real-time fault monitoring and control of the Vacuum Assisted Resin Transfer Moulding production process requires knowledge of the position of the epoxy flow-front inside the mould. Therefore, a fast and accurate flow-front tracking system is highly prized. Physics-informed grey-box models deliver a good trade-off between high fidelity and data-driven black-box models for designing such a flow-front tracking system. In this paper, we propose stochastic differential equations (SDEs) based grey-box model of the flow-front dynamics in the case of missing sensor information. The proposed method uses the finite difference approximation of the spatial domain of the flow-front for estimating the spatial flow pattern of the epoxy. To accommodate for the missing sensor data, we utilize a modified version of the continuous-discrete extended Kalman filter based estimation framework for SDEs that takes into consideration the effective dimension of the measurement space during the identification process. The performance of the method is evaluated for common fault scenarios.
AB - Real-time fault monitoring and control of the Vacuum Assisted Resin Transfer Moulding production process requires knowledge of the position of the epoxy flow-front inside the mould. Therefore, a fast and accurate flow-front tracking system is highly prized. Physics-informed grey-box models deliver a good trade-off between high fidelity and data-driven black-box models for designing such a flow-front tracking system. In this paper, we propose stochastic differential equations (SDEs) based grey-box model of the flow-front dynamics in the case of missing sensor information. The proposed method uses the finite difference approximation of the spatial domain of the flow-front for estimating the spatial flow pattern of the epoxy. To accommodate for the missing sensor data, we utilize a modified version of the continuous-discrete extended Kalman filter based estimation framework for SDEs that takes into consideration the effective dimension of the measurement space during the identification process. The performance of the method is evaluated for common fault scenarios.
KW - Continuous-discrete Kalman filter
KW - Missing data
KW - Physics informed grey-box model
KW - Stochastic differential equations
U2 - 10.1016/j.ifacol.2021.08.459
DO - 10.1016/j.ifacol.2021.08.459
M3 - Conference article
SN - 2405-8963
VL - 54
SP - 797
EP - 802
JO - IFAC-PapersOnLine
JF - IFAC-PapersOnLine
IS - 7
ER -