Ion currents in engines contain important information on combustion that sensors that are currently used are unable to detect. This study focuses on the analysis of ion current signals in a diesel engine. The various features of the ion current signal are described in the context of the heat release rate and in-cylinder pressure signal to better understand the components of the ion current. A variety of back-propagation artificial neural networks are used to predict expansion work, net indicated mean effective pressure and engine out torque from the ion current signal. The factors affecting the performance of the artificial neural networks that were investigated were the number of hidden layers, number of hidden nodes and the desired training error. It was found that more complex networks were better able to predict engine out torque and took longer to train. A reduction in the desired training error improved performance up to a point when the network was no longer able to converge. The effect of averaging on the signal was also tested. Averaging signals at the input to the network was found to reduce the accuracy of predictions of the network, as a result of the reduction in size of the training set.
AsanoMKumaTKajitaniM. Development of new ion current combustion control system. SAE paper 980162, 1998.
7.
GazisAPanousakisDChenR. Computationally inexpensive methods of ion current signal manipulation for predicting the characteristics of engine in-cylinder pressure. Int J Engine Res2006; 7: 271–282.
8.
MehreshPSouderJFlowersD. Combustion timing in HCCI engines determined by ion-sensor: experimental and kinetic modelling. Proc Combust Inst2005; 30: 2701–2709.
9.
MehreshPFlowersDDibbleRW. Experimental and numerical investigation of effect of fuel on ion sensor signal to determine combustion timing in HCCI engines. Int J Engine Res2005; 6: 465–474.
10.
GlavmoMSpadaforaPBoschR. Closed loop start of combustion control utilizing ionization sensing in a diesel engine. SAE paper 1999-01-0549, 1999.
11.
HeneinNABryzikWAbdel-RahimA. Characteristics of ion current signals in compression ignition and spark ignition engines. SAE Int J Engines2010; 3: 260–281.
12.
WickstromNTavenikuMLindeA. Estimating pressure peak position and air-fuel ratio using the ionization current and artificial neural networks. In: IEEE conference on intelligent transportation system, Boston, MA, 9–12 November 1997, pp.972–977. New York: IEEE.
13.
CurranHJGaffuriPPitzWJ. Autoignition chemistry of the hexane isomers: an experimental and kinetic modelling study. SAE paper 952406, 1995.
14.
PeronLCharletAHigelinP. Limitations of ionization current sensors and comparison with cylinder pressure sensors. SAE paper 2000-01-2830, 2000.
GhojelJHonneryD. Heat release model for the combustion of diesel oil emulsions in DI diesel engines. Appl Therm Eng2005; 25: 2072–2085.
17.
GhojelJHonneryDAl-KhaleefiK. Performance, emissions and heat release characteristics of a direct injection diesel engine operating on diesel oil emulsion. Appl Therm Eng2006; 26: 2132–2141.
18.
HeneinNABadawyTRaiN. Ion current, combustion and emission characteristics in an automotive common rail diesel engine. J Eng Gas Turb Power2012; 134: 0428011–0428017.
19.
LahdeTRonkkoTVirtanenA. Heavy duty diesel engine exhaust aerosol particle and ion measurements. Environ Sci Technol2009; 43: 163–168.
20.
ScappinFStefanssonSHHaglindF. Validation of a zero-dimensional model for prediction of NOx and engine performance for electronically controlled marine two-stroke diesel engines. Appl Therm Eng2012; 37: 344–352.
21.
WelsteadST. Neural network and fuzzy logic applications in C/C++. New York: Wiley, 1994, p.59.
22.
ShaylerPJGoodmanMSMaT. Transient air/fuel ratio control of an SI engine using neural networks. SAE paper 960326, 1996.
23.
AsikJRPetersJMMeyesGM. Transient A/F estimation and control using a neural network. SAE paper 980515, 1997.
24.
LenzUShroderD. Air-fuel ratio control for direct injecting combustion engine using neural networks. SAE paper 981060, 1998.
25.
LeonhardtSLudwigCSchwarzR. Real time supervision for diesel engine injection. Control Eng Pract1995; 3: 1003–1010.
26.
ThompsonGJAtkinsonCMClarkNN. Neural network modelling of the emissions and performance of a heavy duty diesel engine. Proc IMechE, Part D: J Automobile Engineering2000; 214: 111–126.
27.
De LucasADuranACarmonaM. Modelling diesel particulate emissions with neural networks. Fuel2001; 80: 539–548.
28.
HafnerMSchulerMNellesO. Fast neural networks for diesel engine control design. Control Eng Pract2000; 8: 1211–1221.
29.
TraverMLAtkinsonRJAtkinsonCM. Neural network-based diesel engine emissions prediction using in-cylinder combustion pressure. SAE paper 1999-01-1532, 1999.
30.
HeYRutlandCJ. Application of artificial neural networks in engine modelling. Int J Engine Res2004; 5: 281–296.
Bosch. Technical data sheet, hot-film air-mass meter, type HFM 2, 2012.
33.
RiedmillerMBraunH. A direct adaptive method for faster backpropagation learning: the RPROP algorithm. In: Proceedings of the IEEE international conference on neural networks, San Francisco, CA, 29 March–1 April 1993, vol. 1, pp.586–591. New York: IEEE.
34.
IgelCHuskenM. Empirical evaluation of the improved RPROP learning algorithms. Neurocomputing2003; 50: 105–123.
