Adaptive polynomial filter applied to in-cylinder pressure signal in internal combustion engines

• Determination of continuous and derivable functions from discrete data. • ICE pressure signal analysis using convolution integration. • ICE pressure signals adaptive filtering. • Quantitative estimators of filter quality. Pressure record processing is a powerful tool for research and diagnosis of internal combustion engines. From the pressure record, using the energy conservation equation, it is possible to calculate the rate of heat release ( RoHR ) and thus quantify the evolution of the combustion process. In order to obtain accurate RoHR results, it is necessary to remove as much noise as possible from the pressure signal, without removing relevant information from the studied phenomenon. In reciprocating internal combustion engines, pressure data are recorded discretely and synchronized with the crankshaft angle. Both pressure signal and its derivative are used in RoHR calculation. For the integration of differential equations using data as a boundary condition, it is necessary to have no discontinuities in the function and its derivative. This work presents a novel methodology for filtering discrete data adaptively and converting it into a continuous and derivable function. For this, polynomial fits are used in each interval between two experimental data. The polynomial order and the number of points used to make the fit are chosen depending on the value taken by a convolution of the signal. With the adaptive filter it is possible to reduce noise significantly in parts of the cycle where signal-to-noise ratio is low without affecting the parts where signal-to-noise ratio is high. The novelty and main advantage of this filtering methodology is that it is configured with only one parameter independent of the operating conditions of the engine while preserving the information of cycle-to-cycle variations.