Heiko Sequenz

Control Oriented NOx and Soot Models for Diesel Engines

Abstract The further improvement of Diesel engine control and diagnosis requires information on the dynamical soot and NO x formation. Investigations of the air and exhaust path show that process and sensor dynamics play an important role for emission modelling. The simplified combustion process and the emission formation is modelled as a stationary batch process. The air path states and the location of mass fraction burned 50% introduce external dynamics to the stationary model structure. It is shown that the main dynamic influences in emission formation come from the air path states at the engine intake. Additional dynamics result from the gas propagation in the exhaust system and the emission sensors. The presented emission models overcome the measurement dynamics leading to relatively fast dynamic models.

Identification of Nonlinear Static Processes with Local Polynomial Regression and Subset Selection

Abstract The presented method for nonlinear system identification is based on the LOLIMOT algorithm introduced by Nelles and Isermann [1996]. The LOLIMOT algorithm divides the input space by a tree-construction algorithm and interpolates the local linear models by local membership functions. Instead of assuming local linear models, the presented algorithm utilizes general local nonlinear functions, which make the algorithm more flexible. These are approximated by a multidimensional Taylor series. Since the amount of regressors grows fast with the number of inputs and the expansion order, a subset selection procedure is introduced. It reveals significant regressors and gives information about the local functional behavior. The local subset selection is implemented as a stepwise regression with replacement of regressors. Mallows’ C p -statistic is used for the subset selection algorithm and is also implemented for final model selection. The benefit of the extended algorithm lies in the higher flexibility in the local models, which results in less partitions of the input space by a similar approximation quality.

Emission Model Structures for an Implementation on Engine Control Units

Abstract The number of model-based approaches in modern engine control units (ECUs) increases permanently with the increase of engine complexity. Therefore the efficient storage and the fast computation of these models is a challenging task. Due to their fast computation properties most ECUs utilize 2-dimensional grid map structures. Higher dimensional relations need then to be mapped with nested structures of several 2-dimensional grid maps. The determination of suited structures is a time-consuming task and the number of utilized 2-dimensional grid maps may be large for higher dimensional relations. Therefore in the following a grid map structure, a LOLIMOT structure, a Kernel method and an adaptive polynomial approach are compared with respect to model accuracy, computation time and required memory. Conclusions about the implementation on state of the art ECUs are given. As model output the NO x emissions of a DTH Z19 Opel CR-Diesel engine with VGT-turbocharger and exhaust gas recirculation are regarded. Model structures are presented for a global model approach comprising the engine operation point as input, and for a global-local approach predicting the output by weighting local models.

Identification of Emission Measurement Dynamics for Diesel Engines

Abstract Dynamic emission models for Diesel engines are important for future engine development. Emission formation takes place in the cylinder, while the emissions are usually measured downstream the exhaust pipe. Besides operation point dependent gas transportation delays the emission sensors themselves possess a significant dead time and sensor time constant varying for different measured quantities. These sensor individual dynamic measurement characteristics complicate the chronological assignment of cause and effect in emission modelling. Therefore the dynamic characteristics of a NO x , a micro soot sensor and an opacimeter are investigated. Models for the measurement dynamics are derived to separate the measurement dynamics from the engine dynamics. Hence excitation signals which strongly excite the emission formation and only weakly excite the intake and exhaust system are presented. Model results for the measurement dynamics are shown and validated with testbed data.

Stationary Global-Local Emission Models of a CR-Diesel Engine with Adaptive Regressor Selection for Measurements of Airpath and Combustion

Abstract A model structure for stationary emission models is presented. The emissions NO x , soot and opacity are separately modeled. Model inputs are the air mass flow rate, charge air pressure, intake temperature, the location of mass fraction burned 50% and the engine operation point. Local polynomials are trained for each operation point, defined by engine speed and injection quantity. To increase accuracy a selection algorithm of the significant regressors is presented. The local models are composed by means of weighting functions, depending on the engine operation points, to global emission models. For a rasterized operation range, the weighting can be interpreted as a linear interpolation of local models. This enables an easy implementation to common electronic control units (ECUs). Measurements from a CR-Diesel engine show the quality of the models.

Zur Identifikation mehrdimensionaler Kennfelder für Verbrennungsmotoren

Zusammenfassung Es werden verschiedene Kennfeldstrukturen zur Identifikation des stationären Verhaltens multidimensionaler nichtlinearer Prozesse betrachtet. Zur experimentellen Bestimmung der Kennfelder werden die Methode der kleinsten Quadrate in expliziter und rekursiver Form und eine Methode mit Zwischenmodell beschrieben. Es wird eine Kennfeldregularisierung eingeführt, welche die Identifikation von nicht angeregten Stützstellen ermöglicht und eine Glättung erlaubt. Die beschriebenen Methoden sind besonders für Verbrennungsmotoren mit vielen Ein- und Ausgangssignalen geeignet. Als eine Anwendung wird die Identifikation und Invertierung einer Kennfeldstruktur für ein Momentenmodell gezeigt. Abstract Different look-up table structures are regarded for the identification of the stationary behaviour of higher dimensional nonlinear processes. For the experimental determination of look-up tables the methods of least squares in explicit and recursive formulation and an approach with an intermediate model are described. A regularisation of look-up tables is introduced, which enables the identification of not excited grid points and allows a smoothing. The described methods are in particular suited for combustion engines with numerous inputs and outputs. As an application the identification and inversion of a look-up table structure for a torque model is shown.

Model-Based Optimisation of a Step in Acceleration for a CR-Diesel Engine

Abstract Due to more and more stringent emission regulations the optimisation of transient engine operation becomes necessary for future Diesel engine developments. Exemplary for the transient operation a step in acceleration pedal is investigated and optimised with respect to the engine emissions. This gives insights about the emission formation at transient driving and a benchmark for the design of open and closed-loop controls. A model structure to simulate the engine emissions and the engine torque is presented. It consists of a dynamical mean value air path model and a stationary combustion model. Thus the relevant outputs NO x , soot and torque are simulated with respect to the engine actuators. For optimisation the calibrated emission values for steady state are regarded. The actuators of the air path and the crank angle of the main injection are optimised such that the emissions follow a desired trajectory. The results are compared to a non-optimised step in acceleration and to a nonlinear closed loop control of the air path.

A Global-Local Emission-Model for NOx and Soot Emissions of Turbocharged CR-Diesel Engines

Stationary and dynamic models for the emissions of a CR-Diesel engine are developed using a global-local model approach. Results for the NOx and soot emissions are presented. All model inputs are measurable air path states and combustion parameters. They determine the emission formation before the combustion takes place. Therefore the model can be used for emission prediction and simulation. The combustion process is regarded as a batch process, such that the dynamics are introduced as external dynamics to the model via the inputs. Thus a stationary model structure can be applied. As the space of possible air path states varies widely for different engine operation points, several input and output transformations are given that linearize the input and output space. This improves the model quality and extends the operation range of the model. Modeling results are shown for stationary and dynamic data as well as for local and global model operation.Copyright

Stationäre Motorvermessung mit verschiedenen Methoden und Modellen

Die kontinuierliche Verscharfung der gesetzlichen Vorgaben bezuglich Emissionen stellt die Motorenentwickler vor neue Herausforderungen (Bild 6-1, [1, 2]). Gleichzeitig ist zu erwarten, dass sich der langfristige Trend der letzten Jahren bei den Kraftstoffpreisen [3] trotz kurzzeitiger Schwankungen fortsetzen wird (Bild 6-1 rechts, Jahresmittelwerte). Auf die steigenden Kraftstoffpreise sowie die Verpflichtungen bezuglich der CO2- Emissionswerte kann dauerhaft nur mit Kraftstoff sparenden Motoren reagiert werden.