Abstract:A supervisory Model Predictive Control (MPC) approach is developed for an air path system for multi-mode operation in a diesel engine. MPC is a control method based on a predictive dynamic model of system and determines actuator control positions through the optimization of various factors such as tracking performances of target setpoints, moving speed of actuators, limits, etc. Previously, linear MPC has been successfully applied on the air path control problem of a diesel engine, however, most of these applications were developed for a single operation mode which has only one set of control target setpoint values. In reality, a single operation mode cannot cover all requirements of current diesel engines and this complicates practical implementations of linear MPC. The high priority targets for the development of diesel engines are low emissions, high thermal efficiency and robustness. These objectives require multi-mode operations such as a HP EGR (High pressure exhaust gas recirculation) mode in cold coolant condition, a Double EGR mode for sufficient EGR rates, a Diesel Particulate Filter (DPF) regeneration mode for the heat-up of exhaust gas temperature and a rich mode of exhaust gas for Lean NOx Trap (LNT) regeneration. Each engine operation mode requires different target setpoints from the air path, such as air mass flow rate and oxygen concentration. In the multi-mode operation, simple linearization based MPC is limited in the practical application because a linearization point changes depending on target setpoints of each operation mode. This means that a linear model for a specific operation mode is not valid in other operation modes, which have different target setpoints. Moreover, the modern diesel air path system is highly nonlinear and would require a significant number of linear models to adequately represent the entire behavior not only at all operation points of engine speed and load, but also at the various operation modes. In this study, a new scheme is proposed and tested on a diesel engine utilizing combined benefits of a supervisory MPC and component level nonlinear compensators. The developed nonlinear compensators are based on dynamic real-time inversions of individual component models. The control scheme comprises of three parts: a supervisory MPC as the coordination of target setpoints in high-level to achieve the objectives, a component level control of nonlinear compensator, state observation either by virtual sensor or by Electronic Control Unit (ECU) sensor. The test results from an engine test bench and a chassis dynamometer demonstrates that the proposed method works well in multi-mode operation and can be applied with the significant benefit that no mode specific control strategies and calibration are needed. A valuable benefit of this approach can be seen in the calibration effort, particularly during development when the setpoints are not necessarily fixed.

Introducing a Deep Neural Network-based Model Predictive Control Framework for Rapid Controller Implementation

End-to-End Deep Neural Network Based Nonlinear Model Predictive Control: Experimental Implementation on Diesel Engine Emission Control

Get the design right, the rest is automatic: Towards an adaptive MPC controller

Machine Learning Integrated with Model Predictive Control for Imitative Optimal Control of Compression Ignition Engines

Embedded Implementation of Deep Learning-based Linear Model Predictive Control

Deep learning based model predictive control for compression ignition engines

Neural-Network-Based Predictive Control for PMSM Drives

Model Predictive Control of Internal Combustion Engines: A Review and Future Directions

Nonlinear Model Predictive Control of A Gasoline HCCI Engine Using Extreme Learning Machines

Development of a Model Predictive Airpath Controller for a Diesel Engine on a High-Fidelity Engine Model with Transient Thermal Dynamics

Model Predictive Control of Diesel Engine Emissions Based on Neural Network Modeling

A Neural Network Architecture to Learn Explicit MPC Controllers from Data

Real-time Neural-MPC: Deep Learning Model Predictive Control for Quadrotors and Agile Robotic Platforms

Learning-Based Neural Dynamic Surface Predictive Control for MMC

Data-Enabled Finite State Predictive Control for Power Converters Via Adaline Neural Network

Fast Calibration of a Robust Model Predictive Controller for Diesel Engine Airpath

Hierarchical multi-time-scale predictive thermal management and fuel optimization for heavy-duty compression ignition engines

Fast Explicit Machine Learning-Based Model Predictive Control of Nonlinear Processes Using Input Convex Neural Networks

PlanNetX: Learning an Efficient Neural Network Planner from MPC for Longitudinal Control

Model Predictive Control of an Air Path System for Multi-Mode Operation in a Diesel Engine

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