Dennis Assanis

Nonlinear identification of a gasoline HCCI engine using neural networks coupled with principal component analysis

Homogeneous charge compression ignition (HCCI) is a futuristic combustion technology that operates with high efficiency and reduced emissions. HCCI combustion is characterized by complex nonlinear dynamics which necessitates the use of a predictive model in controller design. Developing a physics based model for HCCI involves significant development times and associated costs arising from developing simulation models and calibration. In this paper, a neural networks (NN) based methodology is reported where black box type models are developed to predict HCCI combustion behavior during transient operation. The NN based approach can be considered a low cost and quick alternative to the traditional physics based modeling. A multi-input single-output model was developed each for indicated net mean effective pressure, combustion phasing, maximum in-cylinder pressure rise rate and equivalent air-fuel ratio. The two popular architectures namely multi-layer perceptron (MLP) and radial basis network (RBN) models were compared with respect to design, prediction performance and overall applicability to the transient HCCI modeling problem. A principal component analysis (PCA) is done as a pre-processing step to reduce input dimension thereby reducing memory requirements of the models. Also, PCA reduces the cross-validation time required to identify optimal model hyper-parameters. On comparing the model predictions with the experimental data, it was shown that neural networks can be a powerful approach for non-linear identification of a complex combustion system like the HCCI engine.

An accelerated multi-zone model for engine cycle simulation of homogeneous charge compression ignition combustion

We have developed an accelerated multi-zone model for engine cycle simulation (AMECS) of homogeneous charge compression ignition (HCCI) combustion. This model incorporates chemical kinetics and is intended for use in system-level simulation software. A novel methodology to capture thermal stratification in the multi-zone model is proposed. The methodology calculates thermal stratification inside the cylinder based on a single computational fluid dynamics (CFD) calculation for motored conditions. CFD results are used for tuning zone heat loss multipliers that characterize wall heat loss from each individual engine zone based on the assumption that these heat loss multipliers can then be used at operating conditions different from those used in the single CFD run because the functional form of thermal stratification is more dependent on engine geometry than on operating conditions. The model is benchmarked against detailed CFD calculations and fully coupled HCCI CFD chemical kinetics calculations. The results indicate that the heat loss multiplier approach accurately predicts thermal stratification during the compression stroke and (therefore) HCCI combustion. The AMECS model with the thermal stratification methodology and reduced gasoline chemical kinetics shows good agreement with boosted gasoline HCCI experiments over a range of operating conditions, in terms of in-cylinder pressure and heat release rate predictions. The computational advantage of this method derives from the need for only a single motoring CFD run for a given engine, which makes the method very well suited for rapid HCCI calculations in system-level codes such as GT-Power, where it is often desirable to evaluate consecutive engine cycles.

An ELM based predictive control method for HCCI engines

We formulate and develop a control method for homogeneous charge compression ignition (HCCI) engines using model predictive control (MPC) and models learned from operational data. An HCCI engine is a highly efficient but complex combustion system that operates with a high fuel efficiency and reduced emissions compared to the present technology. HCCI control is a nonlinear, multi-input multi-output problem with state and actuator constraints which makes controller design a challenging task. In this paper, we propose an MPC approach where the constraints are elegantly included in the control problem along with optimality in control. We develop the engine models using experimental data so that the complexity and time involved in the modeling process can be reduced. An Extreme Learning Machine (ELM) is used to capture the engine dynamic behavior and is used by the MPC controller to evaluate control actions. We also used a simplified quadratic programming making use of the convexity of the MPC problem so that the algorithm can be implemented on the engine control unit that is limited in memory. The working and effectiveness of the proposed MPC methodology has been analyzed in simulation using a nonlinear HCCI engine model. The controller tracks several reference signals taking into account the constraints defined by HCCI states, actuators and operational limits.

Stochastic gradient based extreme learning machines for stable online learning of advanced combustion engines

We propose and develop SG-ELM, a stable online learning algorithm based on stochastic gradients and Extreme Learning Machines (ELM). We propose SG-ELM particularly for systems that are required to be stable during learning; i.e., the estimated model parameters remain bounded during learning. We use a Lyapunov approach to prove both asymptotic stability of estimation error and boundedness in the model parameters suitable for identification of nonlinear dynamic systems. Using the Lyapunov approach, we determine an upper bound for the learning rate of SG-ELM. The SG-ELM algorithm not only guarantees a stable learning but also reduces the computational demand compared to the recursive least squares based OS-ELM algorithm (Liang et al., 2006). In order to demonstrate the working of SG-ELM on a real-world problem, an advanced combustion engine identification is considered. The algorithm is applied to two case studies: An online regression learning for system identification of a Homogeneous Charge Compression Ignition (HCCI) Engine and an online classification learning (with class imbalance) for identifying the dynamic operating envelope. The case studies demonstrate that the accuracy of the proposed SG-ELM is comparable to that of the OS-ELM approach but adds stability and a reduction in computational effort.

Identification of the Dynamic Operating Envelope of HCCI Engines Using Class Imbalance Learning

Homogeneous charge compression ignition (HCCI) is a futuristic automotive engine technology that can significantly improve fuel economy and reduce emissions. HCCI engine operation is constrained by combustion instabilities, such as knock, ringing, misfires, high-variability combustion, and so on, and it becomes important to identify the operating envelope defined by these constraints for use in engine diagnostics and controller design. HCCI combustion is dominated by complex nonlinear dynamics, and a first-principle-based dynamic modeling of the operating envelope becomes intractable. In this paper, a machine learning approach is presented to identify the stable operating envelope of HCCI combustion, by learning directly from the experimental data. Stability is defined using thresholds on combustion features obtained from engine in-cylinder pressure measurements. This paper considers instabilities arising from engine misfire and high-variability combustion. A gasoline HCCI engine is used for generating stable and unstable data observations. Owing to an imbalance in class proportions in the data set, the models are developed both based on resampling the data set (by undersampling and oversampling) and based on a cost-sensitive learning method (by overweighting the minority class relative to the majority class observations). Support vector machines (SVMs) and recently developed extreme learning machines (ELM) are utilized for developing dynamic classifiers. The results compared against linear classification methods show that cost-sensitive nonlinear ELM and SVM classification algorithms are well suited for the problem. However, the SVM envelope model requires about 80% more parameters for an accuracy improvement of 3% compared with the ELM envelope model indicating that ELM models may be computationally suitable for the engine application. The proposed modeling approach shows that HCCI engine misfires and high-variability combustion can be predicted ahead of time, given the present values of available sensor measurements, making the models suitable for engine diagnostics and control applications.

A System Identification Framework for Modeling Complex Combustion Dynamics Using Support Vector Machines

Machine Learning is being widely applied to problems that are difficult to model using fundamental building blocks. However, the application of machine learning in powertrain modeling is not common because existing powertrain systems have been simple enough to model using simple physics. Also, black box models are yet to demonstrate sufficient robustness and stability features for widespread powertrain applications. However, with emergence of advanced technologies and complex systems in the automotive industry, obtaining a good physical model in a short time becomes a challenge and it becomes important to study alternatives. In this chapter, support vector machines (SVM) are used to obtain identification models for a gasoline homogeneous charge compression ignition (HCCI) engine. A machine learning framework is discussed that addresses several challenges for identification of the considered system that is nonlinear and whose region of stable operation is very narrow.

Effects of fuel injection parameters on the performance of homogeneous charge compression ignition at low-load conditions

With the objective of enhancing the effectiveness of late fuel injection strategy in extending the low-load limit of homogeneous charge compression ignition engines, a numerical study is conducted to investigate the effects of fuel injection parameters, such as the injection pressure and spray cone angle, on the overall combustion efficiency and CO/NOx emissions. Closed-cycle engine simulations are performed incorporating detailed iso-octane reaction kinetics and combustion submodel based on the spray-interactive flamelet approach. Extensive parametric studies are conducted to provide a detailed map of the combustion efficiency and emission performance. In general, it is found that the in-cylinder charge stratification can be reduced by both an increased injection pressure and a wider spray cone angle, resulting in substantially lower NOx emissions and reasonably high combustion efficiency simultaneously. The present study demonstrates that an optimal adjustment of the two fuel injection parameters can result in significant extension of the low-load limit of homogeneous charge compression ignition through delayed fuel injection strategy.

Reaction-space analysis of homogeneous charge compression ignition combustion with varying levels of fuel stratification under positive and negative valve overlap conditions

Full-cycle computational fluid dynamics simulations with gasoline chemical kinetics were performed to determine the impact of breathing and fuel injection strategies on thermal and compositional stratification, combustion and emissions during homogeneous charge compression ignition combustion. The simulations examined positive valve overlap and negative valve overlap strategies, along with fueling by port fuel injection and direct injection. The resulting charge mass distributions were analyzed prior to ignition using ignition delay as a reactivity metric. The reactivity stratification arising from differences in the distributions of fuel–oxygen equivalence ratio ( ϕ FO ), oxygen molar fraction ( χ O 2 ) and temperature (T) was determined for three parametric studies. In the first study, the reactivity stratification and burn duration for positive valve overlap valve events with port fuel injection and early direct injection were nearly identical and were dominated by wall-driven thermal stratification. nitrogen oxide (NO) and carbon monoxide (CO) emissions were negligible for both injection strategies. In the second study, which examined negative valve overlap valve events with direct injection and port fuel injection, reactivity stratification increased for direct injection as the ϕ FO and T distributions associated with direct fuel injection into the hot residual gas were positively correlated; however, the latent heat absorbed from the hot residual gas by the evaporating direct injection fuel jet reduced the overall thermal and reactivity stratification. These stratification effects were offsetting, resulting in similar reactivity stratification and burn durations for the two injection strategies. The higher local burned gas temperatures with direct injection resulted in an order of magnitude increase in NO, while incomplete combustion of locally over-lean regions led to a sevenfold increase in CO emissions compared to port fuel injection. The final study evaluated positive valve overlap and negative valve overlap valve events with direct injection. Relative to positive valve overlap, the negative valve overlap condition had a wider reactivity stratification, a longer burn duration and higher NO and CO emissions associated with reduced fuel–air mixing.

Defeat of the Soot/NOx Trade-off Using Biodiesel-Ethanol in a Moderate Exhaust Gas Recirculation Premixed Low-Temperature Combustion Mode

The soot/nitric oxides (NOx) trade-off of diesel, biodiesel, and biodiesel–ethanol in a moderate exhaust gas recirculation (EGR) premixed low temperature combustion (LTC) mode is investigated in this study. Compared to diesel, biodiesel demonstrates poorer spray behavior and shorter ignition delay, but its oxygen content results in less soot. Blending ethanol into biodiesel enhances spray behavior, prolongs ignition delay, and further increases fuel oxygen fraction, resulting in a larger reduction in soot. In the moderate EGR premixed low temperature combustion mode, an obvious soot/NOx trade-off is demonstrated with diesel fuel. The soot/NOx trade-off is improved by biodiesel fuel and defeated by the biodiesel–ethanol blend. Low soot, low NOx, and high combustion efficiency are achieved with the biodiesel–ethanol blend and proper EGR rate. [DOI: 10.1115/1.4024380]

Scaling and dimensional methods to incorporate knock and flammability limits in models of high-efficiency gasoline and ethanol engines

Recent work has shown the utility of using simplified models with prescribed burn rates to assess the potential of advanced combustion strategies to increase engine efficiency. However, this approach can be improved by incorporating knock and flammability limits. This work incorporates such limits using a combination of simplified conceptual models that are based on theoretical understanding of knock and flame phenomenon and calibration with experimental results. Using this method, the ideal (unconstrained) and feasible (constrained by knock and flammability) potential of a high efficiency gasoline and E85 engine are compared against a baseline naturally aspirated gasoline engine. Turbocharging, dilution with EGR, and higher compression ratios are used to increase the efficiency potential of the high efficiency gasoline and E85 engines. Results demonstrate the benefit of using this simplified approach in modeling high efficiency engines: the high efficiency gasoline engine is most limited by knock while the E85 engine is limited much less; also increased EGR can be used for the E85 engine due to the higher flame speeds of ethanol. Fuel economy maps are created for each engine/fuel strategy and evaluated in a vehicle model to obtain fuel economy results. Results comparing feasible engines show that peak brake thermal efficiency (BTE) is increased by 11.4% for the high efficiency gasoline engine and 17.8% for the E85 engine, as compared to the baseline gasoline engine. Projected vehicle fuel economy (energy equivalent) improvements are 30.1% for the high efficiency gasoline, and 40.9% for the E85 engine relative to baseline.