Virtual Sensors based on deep-learning models for predicting the NOx emissions of a Diesel Engine under transient conditions were developed and verified. Raw data from laboratory experimental measurements, under marine transient loading cycles, were used for training and evaluation of the developed models. NOx prediction under transient conditions is often inaccurate by implementing conventional methods since they fail to capture the dynamic behavior of internal combustion engines. The proposed model is based on Long Short-Term Memory (LSTM) Networks. A Deep Feed-forward Neural Network (DFNN) was also developed to validate the LSTM. The LSTM input is a time sequence of past measurements of the inputs while the DFNN only uses the most recent measurements. The Bayesian Hyberband Optimization (BOHB) algorithm determined the structure and parameters of each network. Each model uses the same inputs and is directly derived from the engine ECU. The LSTM validation showed that the model can generalize and accurately predict the NOx emissions under transient loading compared to the DFNN.

In the present work, the combustion process of a retrofitted high-speed marine Diesel Methanol Dual Fuel (DMDF) engine is numerically evaluated. This study examines the effects of two important operational parameters, the methanol energy substitution ratio (MESR) and diesel injection timing, with a focus on engine performance and emissions. To perform the analysis, a CFD numerical combustion model was developed, and a mean value model, along with other data-driven models, were employed to estimate the intake cylinder conditions. The numerical models were calibrated and validated using experimental data measured at the DMDF experimental testbed at the Laboratory of Marine Engineering (LME). The models were utilized to conduct a parametric study considering various engine speeds and loads, diesel injection timings, and MESRs up to 75%. The impact of these parameters was quantified with respect to in-cylinder pressure, ignition timing, combustion efficiency, NO_x, soot, and HC emissions. The results revealed that an increased methanol ratio leads to delayed ignition timing, shorter combustion duration, and reduced in-cylinder peak pressure and combustion efficiency. NO_x and soot emissions are also reduced, whereas the concentrations of unburned hydrocarbons in the exhaust gas increase significantly and mainly consist of Volatile Organic Compounds (VOCs). Although advancing injection timing in dual-fuel mode improves combustion efficiency, it increases the maximum in-cylinder pressure and NO_x emissions. The other emissions are either reduced or maintained at the same levels. Moreover, the results suggest that there is a trade-off between NO_x emissions and combustion performance, which must be taken into account when the operational parameters are adjusted for these engines. Finally, the maximum MESRs are estimated to ensure safe combustion within acceptable peak pressure limits and adequate combustion performance.