Adaptive sliding mode observer for updating maps with an application to mass air flow sensors in diesel engines

Abstract

A novel method for mass air flow (MAF) sensor bias compensation and error map (or look-up table) adaptation with model error correction is proposed. A key feature of the approach is its method of handling and storing operating-point-dependent MAF sensor errors due to installation and ageing in diesel engines; such errors lead to adverse impacts on emission performance. The model of the MAF sensor error depending on the engine operating point is represented as a two-dimensional (2D) map, which is described as a piecewise bilinear interpolation model in the form of a vector–vector dot product. The mean-value engine model of a diesel engine with additional model biases is analysed and employed to improve the estimation precision of the 2D map. Based on the combination of the 2D map regression model and diesel engine mean-value engine model with additional model biases, a linear parameter varying adaptive sliding mode observer is designed, which achieves the disturbance suppression for the nonlinear model errors, as well as the simultaneous estimation of the system state, linear model errors and map parameters. The convergence of the proposed algorithm is proven under the conditions of the persistent excitation and given inequalities. The observer is validated against simulation data from the engine software enDYNA provided by TESIS. The results demonstrate that the estimation precision of the MAF sensor error map can be improved using the proposed method.