Abstract
In this paper towboat shaft power was predicted using various artificial neural networks. This work is a step toward reducing errors in the prediction of towboat power as well as providing better understanding of powering characteristics by the crew of the towboat. An ensemble neural network (ENN) and the single neural network (ANN) with two hidden layers are proposed to predict towboat shaft power. These two models were compared on the basis of their calculated root mean squared errors, mean absolute errors and relative errors. The database used for training and testing of the proposed ANN and ENN has been collected from the full-scale speed-power trials. Trials are conducted on selected towboats and convoys of barges. The goal of the paper is to show that ENN can be applied on towboat shaft power prediction and can improve the accuracy of the results over the single ANN. Computational results from this numerical example show that ENN definitely outperforms single ANN with two hidden layers. The contribution of this paper is a proposal to use an AIC-based ENN method for predicting towboat shaft powers. The paper is the first one that addresses AIC-based ENN method to predict towboat shaft powers.
Similar content being viewed by others
Notes
A bias neuron is invented to adjust only weights during the training process and always emits +1 value.
In the case of Neural network model sample size corresponds to number of trained data (N tr).
Abbreviations
- AIC:
-
Akaike information criterion
- AICc :
-
Corrected AIC value
- B :
-
Pushed convoy beam in meters (m)
- \(\beta\) :
-
Constant
- ∆AICc :
-
Modified delta_AIC value
- DIV:
-
Diversity of the component network
- η 1 :
-
Relative error calculated for training data set
- η 2 :
-
Relative error calculated for testing data set
- f 1 :
-
Relative frequencies of the relative errors calculated for training data set
- f 2 :
-
Relative frequencies of the relative errors calculated for testing data set
- F n :
-
Froude number
- K :
-
Total number of estimated parameters of the ANN model
- L :
-
Pushed convoy length in meters (m)
- Λ :
-
Weight of the component network
- N :
-
Sample size or number of all measured data
- N:
-
The set of all natural numbers
- \(n_s^1\) :
-
Number of times the relative errors appear in subinterval s calculated for training data set
- \(n_s^2\) :
-
Number of times the relative errors appear in subinterval s calculated for testing data set
- N h :
-
Number of hidden neurons
- N i :
-
Number of input neurons
- N tr :
-
Number of training data
- O :
-
Set of all shaft powers obtained through measurements (real outputs)
- Ô :
-
Set of all predicted shaft powers
- Ô 1 :
-
Set of all towboat shaft powers predicted with training data set by ENN
- Ô 2 :
-
Set of all towboat shaft powers predicted with testing data set by ENN
- Ô 1(r) :
-
Set of all towboat shaft powers predicted with training data set by r-th component network
- \(\hat{O}_{j}^{1\left( r \right)}\) :
-
J-th towboat shaft power predicted with training data set by r-th component network
- R :
-
Number of component networks
- RMSE:
-
Root mean squared error
- S :
-
Number of subintervals
- \(\hat{\sigma }^{2}\) :
-
Mean squared error (MSE)
- SHP:
-
Towboat shaft power (kW)
- T :
-
Pushed convoy mean drought in meters (m)
- V :
-
Pushed convoy volume of displacement in cubic meters (m3) (volume of the underwater shape of the pushed convoy which is equal to the displacement of the pushed convoy on the rivers)
References
Abdel Nabya MM, Lehetaa HW, Banawana AA, Elhewya AH (2008) Investigation of various artificial neural networks techniques for the prediction of inland water units’ resistance. Ships Offshore Struct 3:247–254
Akaike H (1973) Information theory as an extension of the maximum likelihood principle. In: Second international symposium on information theory, Budapest, Hungary, pp 267–281
Authority for determination of seaworthiness, Ministry of Transport, Republic of Serbia (2012) Rules on the method of conducting trial trips and on navigation zones (in Serbian), Official Gazette of the Republic of Serbia, No 99/12, in Serbian language, internet adress: http://www.uprava-brodova.gov.rs/sr_lat/pdf/novosti/Pravilnik%20o%20probnoj%20voznji.pdf. Accessed 6 November 2013
Bertram V, Mesbahi E (2006) SWATH ship design formulae based on artificial neural nets. In: 5th international conference on high performance marine vehicles, Australia, pp 8–10, November 2006
Colic V (2006) Research of navigational, technical, energetic and propulsive characteristics of Danube towboats. Faculty of Transport and Traffic Engineering (in Serbian), University of Belgrade, Belgrade, p 350
Couser PR, Mason A (2004) Artificial neural networks for hull resistance prediction. Computer applications and information technology in the maritime industries (COMPIT’04), Siguenza, Spain, pp 9–12, May 2004
Economic Commission for Europe (2006) Recommendations on harmonized Europe-wide technical requirements for inland navigation vessels. Inland Transport Committee. ECE/TRANS/SC.3/172, New York and Geneva. Available from: http://www.unece.org/trans/doc/finaldocs/sc3/Resolution%20No61E.pdf. Accessed 5 November 2013
Giacinto G, Roli F (2001) Design of effective neural network ensembles for image classification purposes. Image Vis Comput 19(10):699–707
Hansen LK, Salamon P (1990) Neural network ensembles. IEEE Trans Pattern Anal Mach Intell 12(10):993–1001
Hartman EJ, Keeler JD, Kowalski JM (1990) Layered neural networks with Gaussian hidden units are universal approximations. Neural Comput 2(2):210–215
Hashem S (1995) Optimal linear combinations of neural networks. Neural Netw 10(4):599–614
Holtrop J, Mennen GGJ (1982) An approximate power prediction method. Int Shipbuild Prog 29(335):166–170
Hornik K, Stinchcombe M, White H (1989) Multilayer feedforward networks are universal approximators. Neural Netw 2(5):359–366
Huang G-B (2003) Learning capability and storage capacity of two-hidden-layer feedforward networks. IEEE Trans Neural Netw 14:274–281
Isnel M (2008) Uncertainty in the analysis of speed and power trials. Ocean Eng 35:1183–1193
International Towing Tank Conference (ITTC) (1999) The Specialist Committee on Trials and Monitoring––final report and recommendations to the 22nd ITTC. In: Proceedings of 22nd ITTC, Seoul/Shanghai
International Towing Tank Conference (ITTC) (1957) Proceedings of the 8th international towing tank conference, Madrid, Spain
Karlaftis MG, Vlahogianni EI (2011) Statistical methods versus neural networks in transportation research: differences, similarities and some insights. Transp Res Part C 19:387–399
King KK, Yasukawa H, Hirata N, Kose K (2008) Maneuvering simulations of pusher-barge systems. J Mar Sci Technol 13(2):117–126
Mason A, Couser PR, Mason G, Smith CR, Konsky BR (2005) Optimisation of vessel resistance using genetic algorithms. Computer applications and information technology in the maritime industries, COMPIT’05, Hamburg, Germany, pp 440–454
Nash S (2000) A survey of truncated-Newton methods. J Comput Appl Math 124(1):45–59
Nocedal J, Wright SJ (1999) Numerical optimization. In: Mikosch TV, Resnick SI, Robinson SM (eds) Springer series in operations research and financial engineering, 2nd edn. New York, USA, p 634
Pedersen B, Larsen J (2009) Prediction of full-scale propulsion power using artificial neural networks. In: Proceedings of the 8th international conference on computer and IT applications in the maritime industries (COMPIT’09), Budapest, Hungary, pp 537–550, May 10–12. http://www2.imm.dtu.dk/pubdb/p.php?5840
Perrone MP, Cooper LN (1993) When networks disagree: Ensemble method for neural networks. In: Mammone RJ (ed) Artificial neural networks for speech and vision. Chapman & Hall, New York, pp 126–142
Petersen JP, Jacobsen DJ (2012) Statistical modelling for ship propulsion efficiency. J Mar Sci Technol 17:30–39
Radonjic A (2011) Strategy to reduce pollution from Serbian towboats. Int J Traffic Transp Eng 1(2):59–72
Radonjic A, Hrle Z, Colic V (2012) Application of different learning algorithms for the prediction of power of inland pushboats. In: Proceedings of the first international conference on traffic and transport engineering ICTTE 2012, Scientific Research Center Ltd, Belgrade, pp 29–39, 29–30.11.2012
Ren LQ, Zhao ZY (2002) An optimal neural network and concrete strength modeling. Adv Eng Softw 33(3):117–130
Ricketts Udr JB, Hundley LL (1997) Naval ship self-assessment of hull powering performance using propulsion shaft Torsionmeters and GPS. Naval Eng J 109(3):181–192
Riedmiller M, Braun H (1993) A direct adaptive method for faster backpropagation learning: The RPROP algorithm. In: Proceedings of the IEEE international conference on neural networks, IEEE Press, pp 586–591
Sugiura N (1978) Further analysis of data by Akaike’s information criterion and the finite corrections. Commun Stat Theory Methods 7(1):12–26
Sollich P, Krogh A (1996) Learning with ensembles: How over-fitting can be useful. In: Touretzky DS, Mozer MC, Hasselmo ME (eds) Advances in neural information processing systems 8. Denver, CO, MIT Press, Cambridge, MA, pp 190–196
Teodorovic D, Vukadinovic K (1998) Traffic control and transport planning: a fuzzy sets and neural networks approach, 1st edn. Springer Science + Business Media, New York, p 387 Originally Published by Kluwer Academic Publishers, ISBN: 978-94-010-5892-6
Vukadinovic K, Teodorovic D, Pavkovic G (1997) A neural network approach to the vessel dispatching problem. Eur J Oper Res 102(3):473–487
West D, Dellana S, Qian J (2005) Neural network ensemble strategies for financial decision applications. Comput Oper Res 32(10):2543–2559
Wu G-D, Lo S-L (2010) Effects of data normalization and inherent-factor on decision of optimal coagulant dosage in water treatment by artificial neural network. Expert Syst Appl 37:4974–4983
Zhao Z, Zhang Y, Liao H (2008) Design of ensemble neural network using the Akaike information criterion. Eng Appl Artif Intell 21(8):1182–1188
Zheng J (2009) Predicting software reliability with neural network ensembles. Expert Syst Appl 36(2):2116–2122
Zhu C, Byrd RH, Nocedal J (1997) Algorithm 778: L-BFGS-B, FORTRAN routines for large scale bound constrained optimization. ACM Trans Math Softw 23(4):550–560
Acknowledgments
This work has been supported by Serbian Ministry of Science and Technological Development, Grant No. TR36002.
Author information
Authors and Affiliations
Corresponding author
About this article
Cite this article
Radonjic, A., Vukadinovic, K. Application of ensemble neural networks to prediction of towboat shaft power. J Mar Sci Technol 20, 64–80 (2015). https://doi.org/10.1007/s00773-014-0273-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00773-014-0273-2