Research Areas


  • Multivariable control design for engine and exhaust aftertreatment systems

  • Robust and adaptive control for variability reduction and performance optimization

  • Engine and catalyst fault-detection and real-time diagnostics/prognostics

  • Control adaptation and optimization to variable operating conditions, aging and biofuel content

  • Engineering desktop and data driven tools for powertrain monitoring and control


Selected Past Project Descriptions


Learning Controller Based Diagnostics

Demonstrated is the use of robust control for precision tracking applied to system subjected to periodic sequences. The objective is to achieve robust tracking of the periodic inputs/disturbances for engine diagnostics and controller parameter tuning. The control solution is systematic where the controller design process can be executed in a desktop environment using a data driven approach.

Lean NOx Trap Gain Scheduled Control

Demonstrated is the linear parameter varying (LPV) control for Lean Operation in SI engines. The objective is to maximize A/F ratio regulating performance despite variations in engine speed and transport delay. The control solution provides the systematic design of a low complexity gain scheduled controller using LPV methods.

Model Based Diagnostics

Information Synthesis is a knowledge basis that integrates online system identification techniques with first principle models to realize accurate prognostics of engines. This approach compresses large amounts of data into a minimal realization, significantly reduces false detections caused by system variability and sensor noise, and addresses transient operation as well as steady state operation.

Data Driven Controller Identification

Demonstrated is the use of data for the purpose of controller identification as opposed to controller design. The objective is to identify controllers using nonparametric and parametric modeling techniques. The control solution reveals controller structure and calibration in a systematic design process that can be executed in a desktop environment.

Three Way Catalyst Control & Diagnostics

Demonstrated is the adaptive control of a Three-Way Catalyst (TWC). The objective is to maximize the TWC conversion efficiencies despite TWC health and drive cycle. The control solution requires the calibration of only 2 parameters. Since on-line TWC parameter estimation is part of the control solution, TWC health estimation is also achieved.

Multivariable Control-Loop Cooperation

Demonstrated is the use of loop interaction in multivariable systems. The objective is to exploit multivariable control to achieve multi-objectives in engine transient control, air handling, torque/speed quality and combustion management. The control solution is systematic where the loop interactions are displayed controller design process can be executed in a desktop environment.

Self-Tuning 2DOF Adaptive Controllers

Demonstrated is the online identification of a model from measured inputs/outputs (I/O) and the integration of this adapting model into a robust feedback controller design solution. The objective is to automatically adapt controller parameters based on the adapted I/O model. The control solution directly addresses meaningful performance specifications, saturation and time delays.

Automating Governor Calibration

The objective of this research is to develop a systematic diesel engine PI governor design methodology independent of the application. The solution approach is to integrate online modeling and robust controller design in two steps: first, a system identification is performed in the four-step instrumental variable (IV4) method, and second, an adaptive controller is designed based on Quantitative Feedback Theory (QFT) and is executed on a Nichols Chart.

EGR System Real-Time Fault Diagnostics in Diesel Engines Using Least-Squares Methods

A model-based real-time fault detection and estimation methodology is proposed to detect high flow or low flow in the EGR system of Diesel engines. An advantage of the proposed method is its capability for estimating the magnitude of a fault. The method was successfully validated to diagnose low flow and high flow faults in Diesel engines using experimental data