Separable Least Squares Identification of a Wiener Model with Application to Powered Air Purifying Respirators

Abstract

Powered Air Purifying Respirator (PAPRs) are respirators that utilise a small centrifugal blower to aid in drawing air through a filter. If the dynamic behaviour of PAPRs with filters can be modelled, adaptive controllers can be made so that air is only supplied through the filter when a user inhales. This paper presents the identification of a Wiener model using a separable least squares method with experimental application to a PAPR. The linear dynamics are shown to fit to an output error model structure, with the nonlinearity being a quadratic function. The model structure is then validated on a PAPR constructed in open loop formation. The validation is furthered by altering the filter resistance to check the performance of the model.