Predicting the output error of the suboptimal state estimator to improve the performance of the MPC-based artificial pancreas

Abstract

The error of single step-ahead output prediction is the information traditionally used to correct the state estimate while exploiting the new measurement of the system output. However, its dynamics and statistical properties can be further studied and exploited in other ways. It is known that in the case of suboptimal state estimation, this output prediction error forms a correlated sequence, hence it can be effectively predicted in real time. Such a suboptimal scenario is typical in applications where the process noise model is not known or it is uncertain. Therefore, the paper deals with the problems of analytical and empirical modeling, identification, and prediction of the output error of the suboptimal state estimator for the sake of improving the output prediction accuracy and ultimately the performance of the model predictive control. The improvements are validated on an empirical model of type 1 diabetes within an in-silico experiment focused on glycemia prediction and implementation of the MPC-based artificial pancreas.

Automated blood glucose regulation for nonlinear model of type-1 diabetic patient under uncertainties: GWOCS type-2 fuzzy approach

“…For mentioned control approaches in [ 13 , 23 – 26 ] it provides satisfactory performance in regulating BGL of T1DPs against parametric uncertainty and meal disturbance however it considered in the design the possibility of measuring all state variables which practically is not true for reasons of high cost and reliability. In [ 27 – 29 ] model predictive control (MPC) was utilized to regulate BGL of T1DP, it has the ability to deal with parametric uncertainty of glucose-insulin physiological system however the computational of iterative online optimization process is complex and its efficiency depends on how the accuracy of the predicted output is, in [ 30 ] nonlinear explicit MPC (NEMPC) was used to avoid the computational complexity of iterative online optimization, in [ 31 ] the output error state observer was used with MPC to correct the output variable prediction and hence improve the performance of the MPC, also in [ 32 ] extended kalman filter was used with NEMPC to estimate unavailable states for T1DP to improve the accuracy of the predicted BGL. To deal with unmeasured state variables, observer was used in conjunction with nonlinear adaptive controller in [ 33 ], with backstepping controller in [ 9 ], and with predictor feedback controller in [ 34 ] to estimate unmeasured state variables.…”

“…To deal with unmeasured state variables, observer was used in conjunction with nonlinear adaptive controller in [ 33 ], with backstepping controller in [ 9 ], and with predictor feedback controller in [ 34 ] to estimate unmeasured state variables. For mentioned control approaches in [ 9 , 31 – 34 ] better regulation of BGL was achieved based only on BGL measurements without the need to measure other status variables of remote insulin and plasma insulin level, however the desired performance of the system depends on the accuracy of the estimated state variables which the observer model must be close to the real system. Accordingly, there is a research gap in the literature, is there a control approach that can effectively regulate the BGL of T1DPs to counteract the uncertainties introduced into the system based only on BGL measurements without having to use the system model, nor does it need to measure or estimate other state variables of remote insulin and plasma level.…”