Engine load torque estimation using nonlinear observers

An important element in developing dynamic internal combustion diagnostic algorithms utilizing crankshaft speed measurements is to have a relatively accurate estimate of the load torque imposed on the engine. Without this estimate crankshaft effects from the load torque can be incorrectly attributed to cylinder pressure variations, and costly false detections can result. Cylinder pressure estimates improve considerably in accuracy when load torque estimation errors are reduced. This paper describes the formulation of nonlinear engine load torque observers that can be used in conjunction with engine diagnostic algorithms. Simulations of these observers with and without integrators are presented to show the tracking ability during typical vehicle upshifts. Sensor noise, modeling errors, and sensor discretization effects are included in the simulations to show the robustness of these algorithms.