Multicylinder engine pressure reconstruction using NARX neural networks and crank kinematics

Abstract

A NARX neural network is adapted for cylinder pressure trace reconstruction on a multicylinder engine. Following a systematic study to establish the required NARX input information (using measured pressure traces and simulated crank kinematics), two fully recurrent training algorithms are developed and applied to real engine data. These include a back-propagation-through-time algorithm (BPTT) and an extended Kalman filter (EKF). For multi-cylinder engines, two cases are examined, both assuming crank kinematics is obtained from a single shaft-encoder fitted at the forward end of the crankshaft. In one case, a NARX model is constructed to provide an inverse relationship between the kinematics at the encoder location and the pressure trace in an arbitrary cylinder. In the second case, by transforming the kinematics (to emulate a local encoder), a different NARX model is constructed to relate the kinematics at the crank location of a particular cylinder to the corresponding pressure trace. The accuracy and efficiency of both NARX models is examined for application to a three-cylinder in-line DISI engine (in which pressure traces are measured on all cylinders). The paper shows that the computational requirements of training are substantial and, although the efficiency of the EKF algorithm is better than the BPTT, the fitting accuracies are similarly good. For generalization, however (to unseen data), neither method is yet sufficiently accurate (even for steady state engine operation) unless substantially more training data are used to achieve the target accuracy of ± 4 per cent. The overall conclusion of the paper is that the NARX model has the correct architecture for multicylinder pressure reconstruction.

Engine cylinder pressure reconstruction using crank kinematics and recurrently-trained neural networks

“…Optimal inverse linear filter and averaging using engine condition monitoring principles of 'cyclostationary processes' were adopted in [21] to process data from a cylinder head-bolt accelerometer for a 4 cylinder 2-stroke diesel engine running at 900 rpm and various load conditions to produce Pmax within 5-10%, and In this paper recurrent neural network based reconstruction is re-addressed. Previous attempts to train a fully-recurrent network have been severely restricted by the amount of data that can be used owing to the serious inefficiency of the methods, for example in [14].…”

“…In addition, certain features required for On Board Diagnostics could also be improved were pressure traces available for misfire detection, and for in-vehicle calibration by using the cylinder pressure trace for torque estimation [3]. Previous approaches using other on-board sensors, include the use of engine block vibration [4][5], crank kinematics [6] [7], and spark ignition ionisation current [8].…”

“…RBFs trained by recursive hybrid learning of crank or block response data, were used in [13] to predict Pmax to within 2.9%, and θmax within ± 1.5° on a 9 litre 6-cylinder diesel engine at 39 conditions with speeds varying between 800 rpm and 2000 rpm, from 10% to 90% full load. A NARX ANN was trained in [14] using crank acceleration and fully-recurrent training via the BackPropagation Through Time (BPTT) algorithm and the Extended Kalman Filter (EKF), to reconstruct Pmax within 2% for an I3 DISI engine at 1500 rpm and 25.5 Nm. BPTT training was found to be unacceptably slow, and the nominally much faster EKF training was still too slow.…”

“…Optimal inverse linear filter and averaging using engine condition monitoring principles of 'cyclostationary processes' were adopted in [21] to process data from a cylinder head-bolt accelerometer for a 4 cylinder 2-stroke diesel engine running at 900 rpm and various load conditions to produce Pmax within 5-10%, and In this paper recurrent neural network based reconstruction is re-addressed. Previous attempts to train a fully-recurrent network have been severely restricted by the amount of data that can be used owing to the serious inefficiency of the methods, for example in [14].A faster, more robust, recurrent training method is needed so that more data can be used.Here an adaptive gradient descent method proposed in [25] is applied to gasoline engine cylinder pressure reconstruction. The method combines a weight-update algorithm using standard back propagation (BP) with real-time recurrent learning (RTRL) according to required convergence and stability conditions.…”

The Applications of Artificial Neural Networks to Engines

“…Nevertheless, the application is also used for virtual sensing such as emissions (Hanzevack, 1997;Atkinson, 2002) or as described in Prokhorov (Prokhorov, 2005) for misfire detection, torque monitoring or tyre pressure change detection. The combustion process itself has been investigated and parameters been modelled with neural networks by different authors (Potenza et al, 2007;He et al, 2004). Potenza et al developed a model estimating Air-to-Fuel Ratio (AFR) or in-cylinder pressure and temperature on the basis of crankshaft kinematics and its vibrations.…”