Deep learning procedure for knock, performance and emission prediction at steady-state condition of a gasoline engine

Abstract

<jats:p> Recently, deep learning has played an important role in the rise of artificial intelligence, and its accuracy has gained recognition in various research fields. Although engine phenomena are very complicated, they can be predicted with high accuracy using deep learning because they are based on the fundamentals of physics and chemistry. In this research, models were built with deep neural networks for gasoline engine prediction. The model consists of two sub-models. The first predicts the knock occurrence, and the second predicts performance, combustion, and emissions. This includes maximum cylinder pressure, crank angle at maximum cylinder pressure, maximum pressure rise rate, and brake mean effective pressure, brake-specific fuel consumption, brake-specific nitrogen oxides, and brake-specific carbon oxide, which are representative results of the engine (for normal combustion cases without knock). Model input parameters were selected considering engine operating conditions, and physically measurable sensor values. For test cases, the accuracy of the first model for knock classification is 99.0%, and the coefficient of determination (R<jats:sup>2</jats:sup>) values for the second model are all above 0.99. Test times of both models were approximately 2 ms. The robustness of all the models was verified using K-fold cross-validation. A sensitivity study of accuracy, according to the amount of training utilized, was also conducted to determine how many data points are required to effectively train the deep learning model. Accordingly, a deep learning approach was applied to predict the steady-state conditions of a gasoline engine. Achieved model accuracies and robustness proved deep learning to be an effective modeling approach, and test time was recognized to be able to apply for the real-time prediction. The sensitivity analysis can be applied for the preliminary study to define the number of experimental points for the deep learning model. </jats:p>

Obstacle avoidance trajectory planning strategy considering network communication constraints

“…The greater the punishment needs to be imposed on the controlled vehicle to force the control away from the obstacle vehicle. According to the above analysis, the reward function of the repulsion potential field of the obstacle vehicle is established, as shown in equation (8).…”

“…In the existing research, deep neural networks are widely used to describe the nonlinear random uncertainty characteristics that cannot be accurately modelled. [5][6][7][8] Lusch et al 9 used deep learning to find the representation of the Koopman characteristic function from the data, which can effectively model and analyse the nonlinear uncertainty. In addition, the existing obstacle avoidance strategies are generally based on rules in terms of decision-making methods.…”

Optimization of a Spark Ignition Engine Knock and Performance Using the Epsilon-Constrained Differential Evolution Algorithm and Multi-Objective Differential Evolution Algorithm

“…In order to capture the different chemistry of knock lover a wide range of temperatures, the kinetics-fit model uses three different induction times to describe the knock in the low-, intermediate-, and high-temperature regions, respectively. 46 , 47 The overall induction time of this model comprises three different induction times. 48 − 50 Each induction time is defined by eq 5 .…”

“…In order to capture the different chemistry of knock lover a wide range of temperatures, the kinetics-fit model uses three different induction times to describe the knock in the low-, intermediate-, and high-temperature regions, respectively. , The overall induction time of this model comprises three different induction times. − Each induction time is defined by eq . where M 1 is the induction time multiplier, a i through f i are the constant parameters of the model, ON represents the octane number of the fuel used to run the engine, [Diluent] means the mass fraction of the residuals in the unburned zone, mainly including N 2 , CO 2 , and H 2 O, and M 2 is the activation energy multiplier …”

Predicting transient diesel engine NOx emissions using time-series data preprocessing with deep-learning models

“…The knock, performance, combustion, and emissions of a gasoline engine were simultaneously predicted by adopting the deep neural network (DNN) structure. 13 The single model covered combustion results, such as the maximum cylinder pressure, maximum pressure rise rate, and crank angle at the maximum pressure rise rate as well as performance results such as the brake mean effective pressure, BSFC, and emissions, for example, brake-specific nitrogen oxides and brake-specific carbon oxide. However, the results were based on steady-state conditions, rather than transient conditions reflecting real-driving situations.…”