A Deep Learning Engine Power Model for Estimating the Fuel Consumption of Heavy-Duty Trucks

Author(s):  
Yuheng Kan ◽  
Hao Liu ◽  
Xiaoyun Lu ◽  
Qi Chen
Author(s):  
Ziming Wang ◽  
Shunhuai Chen ◽  
Liang Luo

Abstract In the downturn of the shipping industry, optimizing the speed of ships sailing on fixed routes has important practical significance for reducing operating costs. Based on the ship-engine-propeller matching relationship, this paper uses BP neural network to build main engine power model, and correction factors are introduced into the main engine power model to reflect the influence of wind and wave. The Kalman filter algorithm is used to filter the data collected by a river-sea direct ship during the voyage from Zhoushan to Zhangjiagang. The filtered data and the meteorological data obtained from the European Medium-Range Weather Forecast Center are used as the data set of the BP neural network to predict the main engine power. Based on the main engine power model, a multi-objective optimization model of ship speed under the influence of actual wind and waves was established to solve the conflicting goals of reducing sailing time and reducing main engine fuel consumption. This multi-objective model is solved by a non-dominated fast sorting multi-objective genetic algorithm to obtain the Pareto optimal solution set, thereby obtaining the optimal speed optimization scheme. Compared with the original navigation scheme, the navigation time is reduced by 8.83%, and the fuel consumption of the main engine is reduced by 12.95%. The results show that the optimization model can effectively reduce the fuel consumption and control the sailing time, which verifies the effectiveness of the algorithm.


2013 ◽  
Vol 60 (2) ◽  
pp. 185-197 ◽  
Author(s):  
Paweł Sulikowski ◽  
Ryszard Maronski

The problem of the optimal driving technique during the fuel economy competition is reconsidered. The vehicle is regarded as a particle moving on a trace with a variable slope angle. The fuel consumption is minimized as the vehicle covers the given distance in a given time. It is assumed that the run consists of two recurrent phases: acceleration with a full available engine power and coasting down with the engine turned off. The most fuel-efficient technique for shifting gears during acceleration is found. The decision variables are: the vehicle velocities at which the gears should be shifted, on the one hand, and the vehicle velocities when the engine should be turned on and off, on the other hand. For the data of students’ vehicle representing the Faculty of Power and Aeronautical Engineering it has been found that such driving strategy is more effective in comparison with a constant speed strategy with the engine partly throttled, as well as a strategy resulting from optimal control theory when the engine is still active.


2017 ◽  
Vol 110 ◽  
pp. 23-34 ◽  
Author(s):  
Bernardo Tormos ◽  
Leonardo Ramírez ◽  
Jens Johansson ◽  
Marcus Björling ◽  
Roland Larsson

Author(s):  
Mirko Baratta ◽  
Roberto Finesso ◽  
Daniela Misul ◽  
Ezio Spessa ◽  
Yifei Tong ◽  
...  

The environmental concerns officially aroused in 1970s made the control of the engine emissions a major issue for the automotive industry. The corresponding reduction in fuel consumption has become a challenge so as to meet the current and future emission legislations. Given the increasing interest retained by the optimal use of a Variable Valve Actuation (VVA) technology, the present paper investigates into the potentials of combining the VVA solution to CNG fuelling. Experiments and simulations were carried out on a heavy duty 6-cylinders CNG engine equipped with a turbocharger displaying a twin-entry waste-gate-controlled turbine. The analysis aimed at exploring the potentials of the Early Intake Valve Closure (EIVC) mode and to identify advanced solutions for the combustion management as well as for the turbo-matching. The engine model was developed within the GT-Power environment and was finely tuned to reproduce the experimental readings under steady state operations. The 0D-1D model was hence run to reproduce the engine operating conditions at different speeds and loads and to highlight the effect of the VVA on the engine performance as well as on the fuel consumption and engine emissions. Pumping losses proved to reduce to a great extent, thus decreasing the brake specific fuel consumption (BSFC) with respect to the throttled engine. The exhaust temperature at the turbine inlet was kept to an almost constant value and minor variations were allowed. This was meant to avoid an excessive worsening in the TWC working conditions, as well as deterioration in the turbocharger performance during load transients. The numerical results also proved that full load torque increases can be achieved by reducing the spark advance so that a higher enthalpy is delivered to the turbocharger. Similar torque levels were also obtained by means of Early Intake Valve Closing strategy. For the latter case, negligible penalties in the fuel consumption were detected. Moreover, for a given combustion phasing, the IVC angle directly controls the mass-flow rate and thus the torque. On the other hand, a slight dependence on the combustion phasing can be detected at part load. Finally, the simulations assessed for almost constant fuel consumption for a wide range of IVC and SA values. Specific attention was also paid to the turbocharger group functioning and to its correct matching to the engine working point. The simulations showed that the working point on the compressor map can be optimized by properly setting the spark advance (SA) as referred to the adopted intake-valve closing angle. It is anyhow worth observing that the engine high loads set a constraint deriving from the need to meet the limits on the peak firing pressure (PFP), thus limiting the possibility to optimize the working point once the turbo-matching is defined.


Author(s):  
G.K. Booto ◽  
R.A. Bohne ◽  
H. Vignisdottir ◽  
K. Pitera ◽  
G. Marinelli ◽  
...  

Sign in / Sign up

Export Citation Format

Share Document