Automated and Risk-Aware Engine Control Calibration Using Constrained Bayesian Optimization
Authors:
Maarten Vlaswinkel,
Duarte Antunes,
Frank Willems
Abstract:
Decarbonization of the transport sector sets increasingly strict demands to maximize thermal efficiency and minimize greenhouse gas emissions of Internal Combustion Engines. This has led to complex engines with a surge in the number of corresponding tunable parameters in actuator set points and control settings. Automated calibration is therefore essential to keep development time and costs at acc…
▽ More
Decarbonization of the transport sector sets increasingly strict demands to maximize thermal efficiency and minimize greenhouse gas emissions of Internal Combustion Engines. This has led to complex engines with a surge in the number of corresponding tunable parameters in actuator set points and control settings. Automated calibration is therefore essential to keep development time and costs at acceptable levels. In this work, an innovative self-learning calibration method is presented based on in-cylinder pressure curve shaping. This method combines Principal Component Decomposition with constrained Bayesian Optimization. To realize maximal thermal engine efficiency, the optimization problem aims at minimizing the difference between the actual in-cylinder pressure curve and an Idealized Thermodynamic Cycle. By continuously updating a Gaussian Process Regression model of the pressure's Principal Components weights using measurements of the actual operating conditions, the mean in-cylinder pressure curve as well as its uncertainty bounds are learned. This information drives the optimization of calibration parameters, which are automatically adapted while dealing with the risks and uncertainties associated with operational safety and combustion stability. This data-driven method does not require prior knowledge of the system. The proposed method is successfully demonstrated in simulation using a Reactivity Controlled Compression Ignition engine model. The difference between the Gross Indicated Efficiency of the optimal solution found and the true optimum is 0.017%. For this complex engine, the optimal solution was found after 64.4s, which is relatively fast compared to conventional calibration methods.
△ Less
Submitted 26 March, 2025;
originally announced March 2025.
Data-Based In-Cylinder Pressure Model with Cyclic Variations for Combustion Control: A RCCI Engine Application
Authors:
Maarten Vlaswinkel,
Frank Willems
Abstract:
Cylinder pressure-based control is a key enabler for advanced pre-mixed combustion concepts. Besides guaranteeing robust and safe operation, it allows for cylinder pressure and heat release shaping. This requires fast control-oriented combustion models. Over the years, mean-value models have been proposed that can predict combustion measures (e.g., Gross Indicated Mean Effective Pressure, or the c…
▽ More
Cylinder pressure-based control is a key enabler for advanced pre-mixed combustion concepts. Besides guaranteeing robust and safe operation, it allows for cylinder pressure and heat release shaping. This requires fast control-oriented combustion models. Over the years, mean-value models have been proposed that can predict combustion measures (e.g., Gross Indicated Mean Effective Pressure, or the crank angle where 50% of the total heat is released) or models that predict the full in-cylinder pressure. However, these models are not able to capture cyclic variations. This is important in the control design for combustion concepts, like Reactivity Controlled Compression Ignition, that can suffer from large cyclic variations. In this study, the in-cylinder pressure and cyclic variation are modelled using a data-based approach. The model combines Principle Component Decomposition and Gaussian Process Regression. A detailed study is performed on the effects of the different hyperparameters and kernel choices. The approach is applicable to any combustion concept, but most valuable for advance combustion concepts with large cyclic variation. The potential of the proposed approach is demonstrated for an Reactivity Controlled Compression Ignition engine running on Diesel and E85. The prediction quality of the evaluated combustion measures has an overall accuracy of 13.5% and 65.5% in mean behaviour and standard deviation, respectively. The peak-pressure rise-rate is traditionally hard to predict, in the proposed model it has an accuracy of 22.7% and 96.4% in mean behaviour and standard deviation, respectively. This Principle Component Decomposition-based approach is an important step towards in-cylinder pressure shaping. The use of Gaussian Process Regression provides important information on cyclic variation and provides next-cycle controls information on safety and performance criteria.
△ Less
Submitted 7 March, 2024; v1 submitted 6 March, 2024;
originally announced March 2024.
