Hybrid Online Multi-Sensor Error Detection and Functional Redundancy for Artificial Pancreas Control Systems

Abstract

Sensor errors limit the performance of a supervision and control system. Sensor accuracy can be affected by many factors such as extreme working conditions, sensor deterioration and interferences from other devices. It may be difficult to distinguish sensor errors and real dynamic changes in a system. A hybrid online multi-sensor error detection and functional redundancy (HOMSED&FR) algorithm is developed to monitor the performance of multiple sensors and reconcile the erroneous sensor signals. The algorithm relies on two methods, outlier-robust Kalman filter (ORKF) and a locally-weighted partial least squares (LW-PLS) regression model. The two methods have different way of using data, ORKF is comparing current signal samples with the signal trace indicated by previous samples and LW-PLS is comparing samples in the past window with the samples from a database and uses the samples with the most similarity to build a model to predict the current signal values. The performance of this system is illustrated with a clinical case involving artificial pancreas experiments, which include data from a continuous glucose monitoring (CGM) sensor, and energy expenditure (EE) and Galvanic Skin Response (GSR) information based on wearable sensors that collect data from people with type 1 diabetes. The results indicate that the proposed method can successfully detect most of the erroneous signals and substitute them with reasonably estimated values computed by the functional redundancy system.