Engine Exhaust Gas Temperature Research Articles

Creating a neural network-based model to predict the exhaust gas temperature of the internal combustion engine

The regulation and improvement of the performance of internal combustion engines is a continual primary focus of research and development activities conducted within the automobile industry and other relevant sectors. To succeed in reaching this objective, it is necessary to have an accurate and complete model of these engines. However, due to internal combustion engines' complex and nonlinear nature, accurately replicating their behavior may be challenging and time-consuming. Neural networks are a potentially useful strategy for simulating these engines successfully since they offer a solution that strikes a healthy balance between speed and precision. This research investigates the process of building a model of an internal combustion engine by using not one but two separate kinds of neural networks: multilayer perceptrons and radial basis functions. These neural networks aim to simulate and make predictions about the temperature of the engine's exhaust gas. They are especially useful for modeling nonlinear systems because of their incredible convergence speed and excellent accuracy levels.

The marine diesel engine exhaust gas temperature baseline model based on particle swarm optimised generalised regression neural network

ABSTRACT This study addresses limitations in traditional condition monitoring for marine diesel engine (MDE) reliability and stability by proposing a hybrid machine learning and deep learning (DL) model called Particle Swarm Optimization-Generalized Regression Neural Network (PSO-GRNN) for real-time exhaust gas temperature (EGT) monitoring. The model integrates Mahalanobis distance (MD) calibration, wavelet packet denoising, and Pearson correlation analysis for comprehensive data preprocessing. Validation is conducted using historical '6L34DF' diesel engine data. Particle Swarm Optimization (PSO) optimizes the spread value (σ) of the Generalized Regression Neural Network (GRNN) to establish an optimal baseline model for EGT prediction. Experimental results show that PSO-GRNN outperforms Backpropagation Neural Network (BPNN), GRNN, Bidirectional Gated Recurrent Unit (BiGRU), Genetic Algorithm- Generalized Regression Neural Network (GA-GRNN), and Deep Belief Network-Support Vector Regression (DBN-SVR) in terms of training time and accuracy, demonstrating its suitability for MDE baseline modeling.

Influence of oxyhydrogen gas retrofit into two-stroke engine on emissions and exhaust gas temperature variations

The generation of power and fuel sustainability that contributes to a cleaner output of exhaust gases is one of the most important objectives the world seeks. In this paper, oxyhydrogen gas is used to retrofit into a two-stroke engine. The water was electrolysed and generated a mixture of oxygen (O2) and hydrogen (H2) or known as oxyhydrogen (HHO) gas via an electrolytic dry cell generator. The HHO was retrofitted experimentally to investigate the engine emissions and exhaust gas temperature from a 1.5 kW gasoline engine. The engine was tested with different power ratings (84–720 W) to investigate the performance and emissions of the engine using gasoline followed by the addition of HHO. The emissions of CO and NOx were measured with different amounts of HHO added. The exhaust temperature was calculated as one of the variables to be considered in relation to pollution. The air-fuel ratios are varied from 12 to 20% in the experiment. The most appropriate air-fuel ratio needed to start the generator with the most environmentally friendly gas emission was analysed. The results showed that the addition of HHO to the engine is successful in reducing fuel consumption up to 8.9%. A higher percentage of HHO added also has improved the emissions and reduced exhaust gas temperature. In this study, the highest quantity of HHO added at 0.15% of the volume fraction reduced CO gas emission by up to 9.41%, NOx gas up to 4.31%, and exhaust gas temperature by up to 2.02%. Generally, adding oxyhydrogen gas has significantly reduced the emissions, and exhaust temperature and provided an eco-friendly environment.

Real-time FPGA-based laser absorption spectroscopy using on-chip machine learning for 10 kHz intra-cycle emissions sensing towards adaptive reciprocating engines

Fast emissions sensing is needed to enable rapid optimization on-the-fly for increasingly adaptive engine architectures to improve performance over a wide range of loads and to offer fuel flexibility towards a low-carbon energy future. In this work, a real-time laser absorption spectroscopy technique is developed for 10 kHz on-line measurements of engine exhaust gas temperature and carbon monoxide. Latency due to data reduction is significantly shortened through a machine learning approach to spectral analysis and minimization of post-processing complexity for implementation on a high-bandwidth field-programmable gate array (FPGA). The data reduction method is tested on cycle-resolved laser-absorption thermochemistry measurements in reciprocating piston engine exhaust. The sensor employs a quantum cascade laser to spectrally-resolve two fundamental rovibrational absorption lines of carbon monoxide near 4.9 μm at a rate of 10 kHz. The recorded signals are passed to an FPGA to infer CO concentration and gas temperature through either a pre-trained ridge regression or artificial neural network model. Models are constructed to limit complexity with the aim of minimizing resource utilization and latency on the FPGA. Experimental data are used to evaluate the prediction accuracy of the models, with the neural network achieving RMS errors in CO mole fraction and gas temperature of 0.0390% and 15.0 K, respectively, compared to spectral fitting of the absorption lineshapes. The data reduction latencies are measured through hardware-in-the-loop demonstration, achieving a 10 kHz throughput and 25 μs latency. Computational time of the end-to-end data reduction process on the FPGA is measured at 300 ns and 600 ns for the ridge regression and neural network, respectively. The data reduction methods presented in this work expand the utility of laser absorption spectroscopy for low-latency sensors that match the timescales of combustion and increase the potential for real-time sensing and control to minimize engine exhaust emissions and maximize performance.

Solubility and stability enhancement of ethanol in diesel fuel by using tri-n-butyl phosphate as a new surfactant for CI engine

Nowadays, researchers are very interested in improving the stability and solubility of blending diesel fuel with a high percentage of ethanol. As a result, the goal of this paper was to find a way to use the surfactant of Tri-n-butyl phosphate (TBP) substance to blend ethanol with diesel fuel to a level of 40%. Diesel fuel is mixed with ethanol in volumetric proportions of 10%, 20%, 30%, and 40%, as well as a tiny amount of TBP from 1 to 4%. The prepared blends were the subject of an experiment evaluation by fueling a direct injection diesel engine. This engine is a water-cooled, commercial diesel engine, single cylinder, and four-stroke with 12 kW maximum power. The four blends were evaluated as clean fuel mixtures of 10% ethanol/90% diesel/1% TBP, 20% ethanol/80% diesel/2% TBP, 30% ethanol/70% diesel/3% TBP, and 40% ethanol/60% diesel/4% TBP. As the starting fuel, we used 100% diesel to compare the results. The engine’s output and emissions have been measured at various engine loads and constant speeds of 1500 rpm. According to the data gathered, even when the percentage of ethanol was increased to 40%, neither the base fuel nor the engine BTE changed significantly. The engine exhaust gas temperature was found to decrease slightly when the proportion of ethanol was increased. When bioethanol is increased to 40% of the base volume, it causes an increase in the combustion of unburned hydrocarbons and CO emissions. However, when the percentage of ethanol was increased from 100% diesel to the base fuel to 40%, CO2 emissions decreased, and O2 emissions slightly increased.

Co-optimization of miller degree and geometric compression ratio of a large-bore natural gas generator engine with novel Knock models and machine learning

Large-bore natural gas reciprocating generator engines will play crucial roles in future low-carbon energy systems. Knock, as an abnormal combustion phenomenon, is a bottleneck for long time to restrict the engine performance improvement. Development and application of the reliable knock model can contribute the predictive engine control and optimization to further improve the engine efficiency and performance. The biggest challenge of developing the predictive knock model is the stochastic nature of the knock phenomenon. To bridge the technological gap, this study aims to develop a novel predictive knock model for a large-bore natural gas engine and showcase its application for co-optimization of the engine design and control parameters, together with the machine learning. The knock model, taking the effects of hot spots heat release rate and energy density into account, is calibrated by the statistical phenomenological knock factor. The 1-D engine simulation model embedded with the knock model is built and calibrated. The data-driven model based on machine learning is developed and used to predict the relation between the key parameters and the engine performance. The genetic algorithms are employed to achieve the multi-objectives global optimization of geometric compression ratio, Miller degree and other control parameters of the engine to reduce the exhaust gas temperature and improve the engine performance. The co-optimized results show that the engine exhaust gas temperature can be remarkably lowered and the indicated thermal efficiency is increased by 1–3% depending on the operational conditions, without degrading the engine power output and increasing the exhaust NOx emissions.

Exhaust heat recovery performance analysis of a Bi-fuel engine utilizing a thermoelectric generation kit and fuel economy evaluation

The internal combustion engine (ICE) exhaust gas temperature depends on fuel energy and combustion characteristics. The exhaust gas temperature of compressed natural gas fuel (CNG) is more than gasoline fuel and consequently, affect the thermoelectric generation (TEG) performance. This study evaluates how CNG and gasoline fuels affect exhaust temperatures, TEG power output and the brake specific fuel consumption (BSFC). In test setup, the engine exhaust channel is modified to support a thirty thermoelectric unit (TEM) in upper and lower of exhaust muffler; the TEM is clamped between exhaust mufflers and the cooling system to control the cold surface temperature of the TEM. The maximum power output and open circuit voltage of the TEG are described as functions of temperature difference. The electrical power output of the TEG is evaluated in terms of changing engine speed and load for both fuels. The result indicates that the CNG fuel has better temperature distribution of TEG Kit hot surface which can generate a maximum peak power output. The TEG kit can reach 178.3 W, while the gasoline fuel was 164.4 W. The average saving BSFC for a bi-fuel engine was (2.71%:3.20%) with TEG kit for both fuels.

Knock probability determination in a turbocharged gasoline engine through exhaust gas temperature and artificial neural network

In the present study, a new combination of different available engine sensors along with an exhaust gas temperature sensor is used as the input of an artificial neural network to recognize knock incidents probability. In this regard, comprehensive experimental test trips on a turbocharged gasoline-powered vehicle are defined, and about 5 million data sets are extracted. These data are filtered, and then the optimization process is conducted on the network to minimize errors in the presented network. The output of the presented network would be the probability of knock occurrence in different operating conditions which a probability of more than 50% would be recognized as a knock occurrence. The results reveal that the neural network can detect knocking by traditional engine sensors plus exhaust gas temperature sensors with 92% precision. The exhaust gas temperature increases in most knock occurrences due to spark time retarding. The network reveals the little effect of the air temperature sensor on knock incidence, while the strong correlation of intake manifold pressure on the knock occurrence is shown. The spark timing, engine speed, intake temperature, and pressure sensor have the most considerable effect on knock detection, while the camshaft angle sensor has an almost negligible effect on knock detection. The current network promotes knock detection methods by introducing the knock probability, and additional improvement on the presented method can lead to a reliable procedure for knock detection.

Trend in tropopause warming and its influence aircraft performance

Purpose Temperature anomalies in the upper troposphere have become a reality as a result of global warming, which has a noticeable impact on aircraft performance. The purpose of this study is to investigate the total air temperature (TAT) anomaly observed during the cruise level and its impact on engine parameter variations. Design/methodology/approach Empirical methodology is used in this study, and it is based on measurements and observations of anomalous phenomena on the tropopause. The primary data were taken from the Boeing 747-8F's enhanced flight data recorder, which refers to the quantitative method, while the qualitative method is based on a literature review and interviews. The GEnx Integrated Vehicle Health Management system was used for the study's evaluation of engine performance to support the complete range of operational priorities throughout the entire engine lifecycle. Findings The study's findings indicate that TAT and SAT anomalies, which occur between 270- and 320-feet flight level, have a substantial impact on aircraft performance at cruise altitude and, as a result, on engine parameters, specifically an increase in fuel consumption and engine exhaust gas temperature values. The TAT and Ram Rise anomalies were the focus of the atmospheric deviations, which were assessed as major departures from the International Civil Aviation Organizations–defined International Standard Atmosphere, which is obvious on a positive tendency and so goes against the norms. Research limitations/implications Necessary fixed flight parameters gathered from the aircraft's enhanced airborne flight recorder (EAFR) via Aeronautical Radio Incorporated (ARINC) 664 Part 7 at a certain velocity and altitude interfacing with the diagnostic program direct parameter display (DPD), allow for analysis of aircraft performance in a real-time frame. Thus, processed data transmits to the ground maintenance infrastructure for future evaluation and for proper maintenance solutions. Originality/value A real-time analysis of aircraft performance is possible using the diagnostic program DPD in conjunction with necessary fixed flight parameters obtained from the aircraft's EAFR via ARINC 664 Part 7 at a specific speed and altitude. Thus, processed data is transmitted to the ground infrastructure for maintenance to be evaluated in the future and to find the best maintenance fixes.

Internal combustion engine test plan execution order optimization using Travelling Salesman Problem heuristics approach

For a given internal combustion (IC) engine stationary test plan, significant time savings during its realization could be achieved if the sequence of test points execution is adequately determined. The criterion for stabilizing the engine operating point is determined by the magnitude of the change in the most inert parameter, and in the general case, it is the temperature of the engine exhaust gases. A certain level of prior knowledge about the examined object is necessary to conduct such an analysis. If there are no results from previous tests, simulation models, or experiences, the Slow Dynamic Slope (SDS) tests are a great way to quickly gather the necessary information. The task of finding the optimal sequence for a stationary engine testing plan can be set as Travelling Salesman Problem (TSP). This paper will present the application of one of the heuristic methods for solving the TSP on the example of testing the IC engine, which is a very complex dynamic system. Following this model, it is possible to optimize the stationary test plan for any other dynamic system. The basic idea is to find such a sequence of stationary operating points, during the realization of which a minor deviation of the engine exhaust gas temperature is obtained, resulting in the operating point's shortest stabilization time.

Application of the Gaussian Process Regression Method Based on a Combined Kernel Function in Engine Performance Prediction.

At present, regression modeling methods fail to achieve higher simulation accuracy, which limits the application of simulation technology in more fields such as virtual calibration and hardware-in-the-loop real-time simulation in automotive industry. After fully considering the abruptness and complexity of engine predictions, a Gaussian process regression modeling method based on a combined kernel function is proposed and verified in this study for engine torque, emission, and temperature predictions. The comparison results with linear regression, decision tree, support vector machine (abbreviated as SVM), neural network, and other Gaussian regression methods show that the Gaussian regression method based on the combined kernel function proposed in this study can achieve higher prediction accuracy. Fitting results show that the R 2 value of engine torque and exhaust gas temperature after the engine turbo (abbreviated as T4) prediction model reaches 1.00, and the R 2 value of the nitrogen oxide (abbreviated as NOx) prediction model reaches 0.9999. The model generalization ability verification test results show that for a totally new world harmonized transient cycle data, the R 2 value of engine torque prediction is 0.9993, the R 2 value of exhaust gas temperature is 0.995, and the R 2 value of NOx emission prediction result is 0.9962. The results of model generalization ability verification show that the model can achieve high prediction accuracy for performance prediction, temperature prediction, and emission prediction under steady-state and transient operating conditions.

Multifidelity Simulation Research on the Low Reynolds Number Effect on the Engine Performance at Different Altitudes

Abstract With the rapid development of unmanned aerial vehicles, the effect of the low Reynolds number on gas turbine performance and high-altitude endurance has received extensive attention. However, the existing three-dimensional component modeling cannot meet the design requirements of the whole engine level, and the accuracy and physical principles of the existing engine empirical correction cannot be guaranteed. Through the study of a single-shaft turbojet engine, this paper adopts a versatile and accurate coupling method, which combines the volume method and the full coupling method and conducts multifidelity simulation research on the zero-dimensional engine model and the three-dimensional component model. Then, based on the high-altitude test data, under typical operating conditions, compared with the existing empirical correction method in gasturb, the accuracy of the engine inlet flow, fuel flow, thrust, and exhaust gas temperature predicted by the volume-based fully coupled method is improved by 6.2%, 7.9%, 4.7%, and 11.4%, respectively. Next, as the flight altitude rises from 0 km to 21 km, the Reynolds number reduces, the working lines approach the surge lines, and the maximum mass flow rate and the efficiency of the engine components gradually decrease. In addition, in the flow field of the components, the thickness of the boundary layer increases, the shock wave intensity decreases, and the position moves forward. The core innovation of this article is that it provides a creative multifidelity evaluation method for gas turbines to effectively solve the problems of insufficient accuracy of the existing empirical correction methods and the inability of the component design to meet the gas turbine requirements in the study of the low Reynolds number effect at different altitudes, which significantly strengthens the connection among the component internal flow field information extraction, the component characteristics analysis, and the gas turbine performance matching. Moreover, it is conducive to the scientific design of the advanced unmanned aerial vehicles' power.

Study of Absorbing CO2 from Emissions Using a Spray Tower

In order to reduce the environmental impact caused by CO2 emissions from ships and achieve the goal of green shipping, a spray tower using NaOH solution for the absorption of CO2 has been established in this paper. Using the characteristics of a 6135G128ZCa marine diesel engine, the CO2 absorption system was designed and mathematical models of CO2 absorption efficiency were developed. The effects of the variation in engine exhaust gas temperature, the concentration of NaOH solution, the exhaust gas velocity, different load conditions, and different nozzle types on the absorption efficiency of CO2 were thoroughly investigated experimentally. Moreover, the mechanism of CO2 absorption was analyzed. The developed model was verified by comparing the test results with the simulation results. The results of the study proved that using NaOH solution to absorb CO2 from ship exhausts could reduce the level of CO2 emissions from ships by more than 20%, which indicates that this technology could be used in the future to reduce the level of CO2 emissions from ships.

Research on fault prediction of marine diesel engine based on attention-LSTM

The marine diesel engine is an important guarantee for the safe operation of the ship. It is of great significance to predict the running trend of the diesel engine and infer whether the diesel engine is in an abnormal working state in advance to ensure the safe operation of the ship. In this paper, combining the feature extraction ability of the attention mechanism (Attention) and the time-series memory ability of the long short-term neural network (LSTM), a prediction model of the exhaust gas temperature of the marine diesel engine is constructed. Meanwhile, in order to improve the prediction accuracy of the model, the improved particle swarm algorithm is used to optimize the structural parameters of the model, and a model with higher prediction performance is obtained. Then, according to the residual value distribution between the predicted value and the actual value of the model, the process control method is used to set the fault threshold, and the experimental data of normal and abnormal conditions are used for verification. The results show that the proposed model can accurately predict developing failures and provide a new method for ship condition maintenance.

Correction and Fitting Civil Aviation Flight Data EGT Based on RPM: Polynomial Least Squares Analysis

There are different missing flight data due to various reasons in the process of acquisition and storage, especially in general aviation, which cause inconvenience for flight data analysis. Effectively explaining the relationship between flight data parameters and selecting a simple and effective method for fitting and correcting flight data suitable for engineering applications are the main points of the paper. Herein, a convenient and applicable approach of missing data correction and fitting based on the least squares polynomial method is introduced in this work. Firstly, the polynomial fitting model based on the least squares method is used to establish multi-order polynomial by existing flight data since the order of the least squares polynomial has a direct impact on the fitting effect. The order is too high or too small, over-fitting or deviation will occur, resulting in improper data. Therefore, the optimization and selection of the model order are significant for flight data correction and fitting. Because the flight data of the aircraft engine exhaust gas temperature (EGT) are often lost because of the immature detection technology, a series of the multi-order polynomial are established by the relationship of aircraft engine exhaust gas temperature and Revolutions Per Minute (RPM). Case study results confirm the optimal model order is four for the fitting and correction of aircraft engine exhaust temperature, and the least squares polynomial method is applicable and effective for EGT flight data correction and fitting based on RPM data.

Diagnostic Information Analysis of Quickly Changing Temperature of Exhaust Gas from Marine Diesel Engine Part I Single Factor Analysis

Abstract In this paper, attention was paid to the problem of low controllability of marine medium- and high-speed engines during operation, which significantly limits the parametric diagnosis. The measurement of quickly changing temperature of engine exhaust gas was proposed, the courses of which can be a source of diagnostic information. The F statistic of the Fisher-Snedecor distribution was chosen as a statistical tool. Laboratory tests were carried out on the bench of a Farymann Diesel engine. The tests consisted of introducing the real changes in the constructional structure of the considered functional systems of the engine. Three changed parameters for the structure were reviewed: the active cross-sectional area of the inlet air channel, injector opening pressure and compression ratio. Based on the recorded plots of the quick-changing temperatures of the exhaust gases, three diagnostic measures were defined and subjected to statistical tests. The following data were averaged over one cycle for a 4-stroke piston engine operation, (1) the peak-to-peak value of the exhaust gas temperature, (2) the specific enthalpy of the exhaust gas, and (3) the rate of increase and decrease in the values for the quick-changing exhaust gas temperature. In this paper will present results of the first stage of the elimination study: the one-factor statistical analysis (randomised complete plan). The next part will present the results of the second stage of studies: two-factor analysis (block randomised plan), where the significance of the effect of changing the values of the structure parameters on the diagnostic measures was analysed in the background of a variable engine load.

EXPERIMENTAL INVESTIGATION ON THE EFFECT OF ADDING CYCLOHEXANONE TO GASOLINE IN SI ENGINE EMISSIONS AND PERFORMANCE

The aim of the present work is to investigate the influence of adding some ketone compounds on the performance, emissions, heat balance and exhaust gas temperature of spark ignition engine. The ketone used in this study is cyclohexanone (C6H10O). This ketone has been added to the base fuel (gasoline) with three concentration ranges (3, 6 and 9%) respectively. All experimental tests were carried out on gasoline engine type (Nissan QG18DE), four cylinders, 4-stroke, direct injection, with compression ratio (9.5:1). The acquired results showed that adding of ketones affect the physical properties of gasoline. Where the density changed from (710 kg/m3) for net gasoline to (740.8 kg/m3) for cyclohexanone at adding ratio of (9%). The octane number also increased from (86) for pure gasoline to (97.7) for fuel with 9% cyclohexanone. The calorific value will be decrease from (43000 kJ/kg) for gasoline to (42077.5) for cyclohexanone at adding ratio of (9%). The addition of ketones improves the emissions characteristic of engine. The best reduction of (UHC, CO_2, CO and NOx) was (49.04, 22.43, 35.02 and 42.14%) recorded by cyclohexanone addition at ratio of (9%). In the case of performance, all parameters of performance improved by adding ketones. The brake specific fuel consumption reduced by (8.9%) by adding (9%) of cyclohexanone which recorded as the best reduction through all types. The best increment of brake power, brake thermal efficiency, brake mean effective pressure and volumetric efficiency was (17.3, 8.98, 17.25 and 12.7%) is achieved by adding (9%) of cyclohexanone. Also, the exhaust gas temperature will be increase by adding ketones. The percentage increasing of exhaust gas temperature was (28.31%) recorded by cyclohexanone addition at ratio of (9%). In the case of heat balance, the best increment of total heat internal energy was (6.59) at (9%) of cyclohexanone.

Large Two-Stroke Marine Diesel Engine Operation with a High-Pressure SCR System in Heavy Weather Conditions

Abstract The transient performance of a direct-drive large two-stroke marine diesel engine, installed in a vessel operating in a seaway with heavy weather, is investigated via simulation. The main engine of the ship is equipped with a selective catalytic reduction (SCR) aftertreatment system for compliance with the latest International Maritime Organization (IMO) rules for NOx reduction, IMO Tier III. Because of limitations of exhaust gas temperature at the inlet of SCR systems and the low temperature exhaust gases produced by marine diesel engines, in marine applications, the SCR system is installed on the high-pressure side of the turbine. When a ship sails in heavy weather, it experiences a resistance increase, wave-induced motions, and a time-varying flow field in the propeller, induced by ship motions. This results in a fluctuation of the propeller torque demand and, thus, a fluctuation in engine power and exhaust gas temperature, which can affect engine and SCR performance. To investigate this phenomenon and take into account the engine–propeller interaction, the entire propulsion plant was modeled, namely, the slow-speed diesel propulsion engine, the high-pressure SCR system, the directly driven propeller, and the ship’s hull. To simulate the transient propeller torque demand, a propeller model was used, and torque variations due to ship motions were taken into account. Ship motions in waves and wave-added resistance were calculated for regular and irregular waves using a 3D panel code. The coupled model was validated against available measured data from a shipboard propulsion system in good weather conditions. The model was then used to simulate the behavior of a Tier III marine propulsion plant during acceleration from low to medium load, in the presence of regular and irregular waves. The effect of the time-varying propeller demand on the engine and the SCR system was investigated. Introduction The effect of waves on a marine propulsion system is a complex phenomenon involving interactions between different subsystems of the propulsion plant, i.e., the prime mover, the propeller, and the ship’s hull. Ships sailing in heavy weather conditions experience a resistance increase, wave-induced motions, and a time-varying flow field in the propeller. This leads to a fluctuation of the propeller torque demand which results in a fluctuation in engine-produced power and exhaust gas temperature.

Machine learning assisted prediction of exhaust gas temperature of a heavy-duty natural gas spark ignition engine

Exhaust gas temperature is a key parameter for optimizing engine performance and emissions. Of particular interest is forecasting the exhaust gas temperature in a diesel engine converted to spark ignition natural gas operation, as the combustion process in this engine is significantly different from the one in a conventional engine. The goal was to assess four different machine learning algorithms namely the artificial neural network, random forest, support vector regression, and gradient boosting regression trees, with respect to a physical 1D CFD model and relative to one another, when predicting the exhaust temperature. The spark timing, equivalence ratio, and engine speed were model inputs. First, the artificial neural network predicted the exhaust temperature more accurately than the physical model, because of the complex premixed combustion phenomena inside a conventional diesel chamber. When compared relative to one another, all machine learning models predicted the exhaust gas temperature with acceptable error while also capturing its relationship with the three model inputs. The gradient boosting regression trees predicted the best, but it usually requires high quality noise-free data. The random forest had the least accuracy, but it required the least amount of calibration. The support vector regression had the smallest error, but it required the highest computational resources. The artificial neural network algorithm was the most appropriate, but it required effort in tuning its hyperparameters. Overall, the results showed that well-trained machine learning models can complement more complex physical model while also helping with optimizing the engine performance, emissions, and life.

Physical, data-driven and hybrid approaches to model engine exhaust gas temperatures in operational conditions

ABSTRACT Fast diesel engine models for real-time prediction in dynamic conditions are required to predict engine performance parameters, to identify emerging failures early on and to establish trends in performance reduction. In order to address these issues, two main alternatives exist: one is to exploit the physical knowledge of the problem, the other one is to exploit the historical data produced by the modern automation system. Unfortunately, the first approach often results in hard-to-tune and very computationally demanding models that are not suited for real-time prediction, while the second approach is often not trusted because of its questionable physical grounds. In this paper, the authors propose a novel hybrid model, which combines physical and data-driven models, to model diesel engine exhaust gas temperatures in operational conditions. Thanks to the combination of these two techniques, the authors were able to build a fast, accurate and physically grounded model that bridges the gap between the physical and data driven approaches. In order to support the proposal, the authors will show the performance of the different methods on real-world data collected from the Holland Class Oceangoing Patrol Vessel.