Robert Prucka

Control-oriented residual gas mass prediction for spark ignition engines

Trapped residual gas mass is an important physical factor that influences combustion phasing and variation, fuel consumption, and air mass prediction for fuel control. There are currently no mass-production sensors available to directly measure in-cylinder residual gas mass, so prediction models must be utilized for control. Residual gas content of the cylinder can be difficult to directly model for control purposes because it involves complex flows during gas exchange that are driven by fluctuating pressures in the intake and exhaust systems. Capturing these effects in an accurate manner generally requires high model complexity and computational effort outside of the capability of most production intent engine controllers. This paper presents a semi-physics-based control oriented residual gas mass (RGM) prediction method. The RGM model is based on Bernoulli’s principle and considers engine operating conditions, valve timing and geometry, and piston motion effects. Moreover, to more accurately estimate the burned gas back flow, this model captures gas wave dynamic effects in intake and exhaust manifold pressures. The model is described in detail and its prediction accuracy is compared to that of a high fidelity simulation that utilizes experimentally measured crank angle resolved intake, exhaust, and cylinder pressures as boundary conditions. The new model is incorporated into a rapid-prototype control system for real-time operation during transient and steady-state engine operation. The results show that the proposed RGM model provides real-time predictions within 1.9-2.3% RGF, creating relative estimation errors in the range of 10-24%, and is capable of running real-time for engine control.

Nonlinear economic Model Predictive Control for SI engines based on Sequential Quadratic Programming

This paper proposes a model predictive torque control strategy for Spark Ignition (SI) engines with external Exhaust Gas Recirculation (EGR). The proposed Nonlinear (economic) Model Predictive Controller (NMPC) tries to minimize fuel consumption with given Indicate Mean Effective Pressure (IMEP) tracking reference and abnormal combustion constraints like knock and combustion stability. A Nonlinear Programming (NLP) problem is formulated and solved using Sequential Quadratic Programming (SQP) to obtain the desired control actions. The SQP exploits the Gauss-Newton like structure of the real time NLP problem to simplify computation of Hessian matrix. Simulation results demonstrate that the proposed model predictive IMEP control can track the IMEP reference for engine cycles without active constraints (with a RMS tracking error of 1.1%). When the IMEP reference conflicts with constraints, the SQP MPC can efficiently find close to optimal control actions that are similar to those from off-line feed forward calibration.

Pattern recognition technique based active set QP strategy applied to MPC for a driving cycle test

Application of constrained Model Predictive Control (MPC) to systems with fast dynamics is limited by the time consuming iterative optimization solvers. This paper proposes a fast and reliable Quadratic Programming (QP) strategy to solve MPC problems. While the optimal control action is calculated with a fast online dual QP algorithm, a “warm start” technique is adopted to reduce iterations of the online search process. The warm start solution is calculated from a predicted active constraint set generated by a pattern recognition function (Artificial Neural Network, ANN, is discussed). This function is calibrated with data from Monte Carlo simulation of the MPC controller over finite sampling points of the state-space. The proposed MPC strategy can adapt to applications with long prediction/control horizons, Linear Parameter Varying (LPV) dynamics and time varying constraints with balance between computation time, memory requirement and calibration effort. This MPC approach is applied to control vehicle speed for a HIL driving cycle test on an engine dynamometer. Simulation results demonstrate the speed profile tracking error of the MPC “driver” can be 67% less than a PID “driver”. Furthermore, smooth throttle/brake actuations, similar to human drivers are achieved with the MPC controller.

Model-based automotive system integration: Using vehicle hardware in-the-loop simulation for an integration of advanced hybrid electric powertrain

This paper presents a new method for the design and validation of advanced propulsion systems using a new approach called vehicle hardware–in–the–loop (VHiL); where the development process, and mainly the validation phase, of adding hybrid electric powertrain modules to an existing platform is carried out faster, in a cost effective and controllable manner. In the VHiL laboratory a complete real vehicle equipped with a conventional powertrain is set up in a hardware–in–the–loop simulation environment, where a chassis dynamometer is used to simulate the road load. The working principle and the added value of VHiL are demonstrated with test results of the performance of the system. This new concept assists validating the inclusion of hybrid electric module to an existing platform during the design/concept stage.

An investigation of semiphysical artificial neural networks for multi-fuel combustion phasing control of spark ignition engines:

The number of engine control actuators and potential fuel sources are constantly increasing to meet fuel economy targets and global energy demand. The increased engine control complexity resulting from new actuators and fuels motivates the use of model-based control methodologies over map-based empirical approaches. Purely physics-based control techniques have the potential to decrease calibration burdens but must be complex to represent nonlinear engine behavior with low computational requirements. Artificial neural networks are recognized as powerful tools for modeling systems which exhibit nonlinear relationships, but they lack physical significance. Combining these two techniques to produce semiphysical artificial neural network models which provide acceptable accuracy while minimizing the artificial neural network size, the calibration effort and the computational intensity is the focus of this research. To minimize the size of the neural network, sensitivity analyses are carried out on the critical inputs and the minimum number of required neurons. The most critical physical parameters are selected as follows: the laminar flame speed; the turbulence intensity; the total in-cylinder mass. The control algorithm derivation is described, and the process validated in real time using an engine dynamometer. The real-time experimental results demonstrate that the semiphysical artificial neural network approach can produce accurate ignition timing control for both gasoline and E85. Robustness of the semiphysical neural network approach is also discussed on the basis of the real-time experimental results.

Short-term and long-term adaptation algorithm for low-pressure exhaust gas recirculation estimation in spark-ignition engines

Low-pressure exhaust gas recirculation systems are capable of increasing fuel efficiency of spark-ignition engines; however, they introduce control challenges. The low available pressure differential that drives exhaust gas recirculation flow, along with the significant pressure pulsations in the exhaust environment of a turbocharged engine hamper the accuracy of feed-forward estimation models. For that reason, feedback measurements are required in an effort to increase prediction accuracy. Additionally, the accumulation of deposits in the exhaust gas recirculation system and the aging of the valve, change the flow characteristics over time. Under these considerations, an adaptation algorithm is developed which handles both short-term (operating-point-dependent errors) and long-term (system aging) corrections for exhaust gas recirculation flow estimation. The algorithm is based on an extended Kalman filter for joint state and parameter estimation and uses the output of an intake oxygen sensor to adjust the feed-forward prediction by creating an online adaptation map. Two different exhaust gas recirculation estimation models are developed and coupled with the adaptation algorithm. The performance of the algorithm for both estimation models is evaluated in real-time through transient experiments with a turbocharged spark-ignition engine. It is demonstrated that this methodology is capable of creating an adaptation map which captures system aging, while also reduces the estimation bias by more than four times resulting in a prediction error of less than 1%. Finally, this approach proves to be a valuable tool that can significantly reduce offline calibration efforts for such models.

Nonlinear model predictive air path control for turbocharged SI engines with low pressure EGR and a continuous surge valve

This paper proposes a model predictive strategy for air path control turbocharged Spark Ignition (SI) engines with low pressure Exhaust Gas Recirculation (EGR). The proposed Nonlinear Model Predictive Controller (NMPC) is designed to track manifold pressure and EGR concentration reference, by manipulating throttle, EGR valve, continuous surge valve and waste gate. The NMPC is solved in real time using Sequential Quadratic Programming (SQP) to obtain the desired control actions. Simulation results demonstrate that the proposed model predictive air path control can coordinate all the actuators to track manifold absolute pressure (MAP) and EGR concentration demand with minimal response time.

An Economic Nonlinear Model Predictive Control Strategy for SI Engines: Model-Based Design and Real-Time Experimental Validation

This paper proposes a model predictive torque control strategy for spark-ignition engines with external exhaust gas recirculation. The proposed economic nonlinear model predictive controller (E-NMPC) tries to minimize fuel consumption, given an indicated mean effective pressure (IMEP) tracking reference and abnormal combustion constraints such as knock and combustion variabilities. A nonlinear optimization problem is formulated and solved in real time using sequential quadratic programming (SQP) to obtain the desired control actions. The SQP utilizes active set quadratic programming (QP) algorithms, with warm-start techniques that exploit the structural similarities between successive sub-QPs along the SQP sequence. This process reduces QP iterations by approximately 60% for each SQP update. Simulation results demonstrate that the proposed model predictive controller can track the IMEP reference with an rms error of 1.1% for engine cycles without active combustion constraints. When the IMEP reference conflicts with constraints, the SQP E-NMPC can efficiently find close-to-optimal control actions that are similar to those from off-line feed-forward calibration. The proposed algorithm is validated on an engine dynamometer. The algorithm executes in a prototype engine controller with a mean computation time of 1.07 ms, proving its feasibility for future engine control unit implementation.