Dynamic skip fire is a control method for internal combustion engines in which engine cylinders are selectively fired or skipped to meet driver torque demand. In this type of engine operation, fueling, and possibly intake and exhaust valves of each cylinder are actuated on an individual firing opportunity basis. The ability to operate each cylinder at or near its best thermal efficiency, and to achieve flexible control of acoustic and vibrational excitations has been described in previous publications. Due to intermittent induction and exhaust events, air induction and torque production in a DSF engine can vary more than conventional engines on a cycle-to-cycle basis. This paper describes engine thermofluid modeling for this type of operation for purposes of air flow and torque prediction. Development of a one-dimensional model of medium complexity is described, along with solutions for practical issues encountered with the standard configuration of onedimensional simulation packages such as GT-SUITE. Airflow dynamic and thermodynamic simulation results for skip fire engine operation are presented and compared with experimental data under several different firing sequences. The dependence of air charge and net indicated mean effective pressure on skip fire sequence is illustrated. Finally, a method of air estimation compensation is described via characterization of each induction event by skip history, both of the particular cylinder as well as previous cylinders in the firing order.

No abstract

Misfire detection and monitoring on US passenger vehicles are required to comply with detailed and specific requirements contained in the OBD-II regulations. Numerous technical papers and patents discuss various methods and metrics for detecting misfire in conventional all-cylinder firing engines. However, the current methods are generally not suitable for detecting misfires in a dynamic skip fire engine. For example, a detection approach based on peak crankshaft angular acceleration may work well in conventional, all-cylinder firing engine operation, since it is expected that crankshaft acceleration will remain generally consistent for a given operating condition. In a skip fire engine, any cylinder or cycle may be skipped. As a result, the crankshaft acceleration peaks and profiles may change abruptly as the firing sequence changes. This paper presents two approaches for detecting misfires in a dynamic skip fire engine.The first method utilizes crankshaft angular acceleration with the addition of cylinder skip or fire status, which is used to recognize a firing sequence in order to ignore skips and apply a separate threshold to various sequences. For the second approach, a torque model based on multi-cylinder pressure modeling is employed. The paper describes the details in modeling cylinder pressure, indicated torque and crankshaft angular acceleration, and proposes a new metric for misfire detection. Validation tests are carried out both on an engine dynamometer and a vehicle under steady state and transient conditions. The results indicate a very promising approach for detecting misfires in a dynamic skip fire engine.

No abstract

<div class="section abstract"><div class="htmlview paragraph">Engines equipped with Dynamic Skip Fire (DSF) technology generate low frequency and high amplitude excitations that could reduce vehicles drive quality if not properly calibrated. The excitation frequency of each firing pattern depends on its length and on the rotational speed of the engine. Excitation amplitude mainly depends on the requested engine torque by the driver. During the calibration process, the torque characteristics that results in production level of noise, vibration, and harshness (NVH), must be identified, for each firing pattern and engine speed. This process is quite time consuming but necessary.</div><div class="htmlview paragraph">To improve our process, a novel machine learning technique is utilized to accelerate the calibration effort. The idea is to automate the vibration rating procedure such that given the relevant power-train parameters, a vibration rating associated with that driving condition can be predicted. This process is divided into two <span class="xref">(2)</span> prediction models. The first model is a multiple additive regression trees that predicts the seat accelerometer data based on the various engine and vehicle parameters. The predicted seat accelerometer data is used as an input to the second machine learning model which correlates, along with other relevant engine and vehicle parameters, to a final vibration rating score. The results indicate that using this machine learning approach can significantly improve capability of automating the DSF calibration process delivering commercial NVH performance.</div></div>