[Paper Review] Nonlinear Model Predictive Control of A Gasoline HCCI Engine Using Extreme Learning Machines

Vijay Manikandan Janakiraman, XuanLong Nguyen+1 more|arXiv (Cornell University)|Jan 16, 2015

Advanced Combustion Engine Technologies36 references3 citations

TL;DR

This paper proposes a nonlinear Model Predictive Control (MPC) framework for a gasoline HCCI engine using Extreme Learning Machines (ELM) to learn nonlinear engine dynamics from experimental data. The ELM model enables fast, accurate multi-step-ahead predictions, and the MPC controller efficiently handles multi-input, multi-output control with state and actuator constraints in real time, demonstrating robust tracking and constraint satisfaction in simulations.

ABSTRACT

Homogeneous charge compression ignition (HCCI) is a futuristic combustion technology that operates with a high fuel efficiency and reduced emissions. HCCI combustion is characterized by complex nonlinear dynamics which necessitates a model based control approach for automotive application. HCCI engine control is a nonlinear, multi-input multi-output problem with state and actuator constraints which makes controller design a challenging task. Typical HCCI controllers make use of a first principles based model which involves a long development time and cost associated with expert labor and calibration. In this paper, an alternative approach based on machine learning is presented using extreme learning machines (ELM) and nonlinear model predictive control (MPC). A recurrent ELM is used to learn the nonlinear dynamics of HCCI engine using experimental data and is shown to accurately predict the engine behavior several steps ahead in time, suitable for predictive control. Using the ELM engine models, an MPC based control algorithm with a simplified quadratic program update is derived for real time implementation. The working and effectiveness of the MPC approach has been analyzed on a nonlinear HCCI engine model for tracking multiple reference quantities along with constraints defined by HCCI states, actuators and operational limits.

Motivation & Objective

• Address the challenge of controlling highly nonlinear, multi-input, multi-output HCCI engines with strict state and actuator constraints.
• Overcome the long development time and high calibration cost of first-principles models in HCCI control system design.
• Develop a data-driven, computationally efficient control framework suitable for real-time implementation on engine ECUs.
• Enable predictive control using machine learning models that capture complex, nonlinear engine dynamics without simplifying assumptions.
• Demonstrate the feasibility of using ELM-based models within an MPC framework for real-time HCCI engine control with stability and performance constraints.

Proposed method

• Employ a recurrent Extreme Learning Machine (ELM) to identify and model the nonlinear dynamics of an HCCI engine using experimental in-cylinder pressure and control input data.
• Use the analytical derivative structure of ELM to compute exact gradients for optimization, avoiding numerical differentiation issues.
• Formulate the MPC problem as a convex quadratic program to enable fast, real-time solution using efficient QP solvers.
• Incorporate state constraints (e.g., maximum rate of pressure rise $R_{ ext{max}}$) and actuator limits directly into the MPC optimization framework.
• Design a simplified MPC update strategy to reduce computational load while maintaining control performance.
• Validate the controller using a nonlinear HCCI engine model with simulated noise and varying reference trajectories (e.g., sinusoidal IMEP and CA50 commands).

Experimental results

Research questions

• RQ1 Can ELM-based models accurately capture the nonlinear, multi-step-ahead dynamics of an HCCI engine from experimental data?
• RQ2 Can an MPC controller using ELM models achieve real-time control performance with acceptable computational complexity?
• RQ3 How effectively can the ELM-MPC framework handle multiple constraints, including stability limits like $R_{ ext{max}}$?
• RQ4 How does the inclusion of active constraints (e.g., $R_{ ext{max}} < 3.5$ bar/deg CA) affect tracking performance and control authority?
• RQ5 Can the ELM-MPC approach maintain accurate tracking of complex, time-varying reference signals (e.g., sinusoidal power and combustion phasing commands) under realistic operating conditions?

Key findings

• The ELM model accurately predicted HCCI engine behavior several steps ahead in time, enabling effective predictive control without relying on first-principles models.
• The MPC controller successfully tracked sinusoidal reference commands for IMEP and CA50 with minimal steady-state error, demonstrating robust setpoint tracking.
• When the $R_{ ext{max}}$ constraint was active (set to 3.5 bar/deg CA), the MPC reduced the start of injection (SOI) command to avoid excessive pressure rise rates, ensuring stability.
• Constraint handling significantly reduced control freedom, resulting in slightly degraded tracking performance, but maintained safe operation within stability limits.
• The computational complexity of the MPC was manageable, with the convex QP formulation enabling real-time implementation suitable for onboard ECU deployment.
• The ELM-MPC framework effectively balanced performance, constraint satisfaction, and computational efficiency, demonstrating strong potential for real-world HCCI engine control.