Search Results

This paper presents a method for identifying misfiring cylinders using crankshaft angular velocity measurements as input. The measured angular velocity is used in conjunction with a system model to estimate the indicated torque developed by each cylinder. The method is valid for engines with overlapping firing pulses, and also under conditions in which there is large torsional deflection in the crankshaft. The method is suitable for implementation for on-board misfire diagnostics. Experimental validation is provided using a 19 liter, 6 cylinder diesel engine. The experimental data shows that accurate detection and identification of misfiring cylinders is possible over the full speed and load range of the engine.

An overview of analog and digital torsional vibration measurement methods is presented. Significant sources of digital measurement error are discussed in detail. Methods to minimize measurement errors are explored. A method to use multi-bit A/D data acquisition of a magnetic pickup signal to accurately measure tooth passing frequency with low sample rates is proposed. A frequency domain correction window is proposed which improves the frequency response of the standard digital demodulation technique. The proposed improvements enable accurate torsional vibration measurement with inexpensive A/D data acquisition equipment.

The technical literature for engine misfire detection and faulty cylinder identification using crankshaft angular velocity measurements is reviewed. Current methods are categorized into three main approaches: threshold criteria, pattern recognition, and model based deconvolution. Real time detection of engine misfire using measured speed fluctuations is currently practical for a variety of engine and road conditions. Techniques to identify the location of specific faulty cylinders are less mature. An overview and comparison of current methods is presented. The implications of various assumptions made for the location of misfiring cylinders are explored. Remaining challenges are identified. A summary of relevant literature by author is presented. Finally, the current methodology is summarized.

Order based torsional vibration analysis codes require a separate, non-linear model to compute single cylinder excitation torque for input. The slider crank mechanism model is typically derived using a crankshaft constant angular velocity assumption. Additionally, the equation describing this constant speed mechanism is derived in an approximate form. An exact model for the slider crank mechanism turning at non-constant speed is used in an iterative procedure with an order based impedance model torsional analysis code to quantify the effects of these assumptions. Results are presented for an in-line six diesel engine, and generalized where possible.