Wavenet using artificial bee colony applied to modeling of truck engine powertrain components. (May 2015)

• Record Type: Journal Article
• Authors: Klein, Carlos Eduardo; Bittencourt, Mario; Coelho, Leandro dos Santos
• Is Part Of: Engineering applications of artificial intelligence. Volume 41(2015:May)
• Page Start: 41
• Page End: 55
• Publication Date: 2015-05
• Subjects: Nonlinear identification -- Artificial neural network -- Wavelet neural network -- Elman neural network -- Automotive applications -- Powertrain -- Artificial bee colony; Engineering -- Data processing -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Ingénierie -- Informatique -- Périodiques; Intelligence artificielle -- Périodiques; Systèmes experts (Informatique) -- Périodiques; Artificial intelligence; Engineering -- Data processing; Expert systems (Computer science); Periodicals; 620.00285
• DOI: 10.1016/j.engappai.2015.01.009
• Languages: English
• ISSNs: 0952-1976

Abstract

Abstract: The purpose of this paper is to validate an artificial wavelet neural network, or wavenet model, combined with artificial bee colony optimization, a swarm intelligence paradigm, to model powertrain components of a truck engine. The arrangement of artificial neural networks with wavelet based functions, called artificial wavelet neural network or wavenet (AWNN), creates a valuable tool to represent the nonlinear multivariable systems. AWNN can be considered a particular case of the feed-forward basis function neural network model. To illustrate the use of the proposed AWNN based on ABC optimization for the black-box modeling, we apply it to model a truck engine with a cubic displacement greater than seven liters. Identification results were carried out using AWNN implemented in Matlab computational environment and the model accuracy is evaluated based on performance indices. Final results were compared with Elman network, Jordan network and kernel adaptive filtering in order to check the AWNN performance. The comparison methods were tuned with the same optimization algorithm in order to find their best tuning parameters. The simulation results show that the artificial wavelet neural network approach can be useful and a promising technique in powertrain components modeling. This proposed AWNN combined with artificial bee colony approach allows modeling the dynamical behavior of powertrain components of a truck engine. Highlights: Simulation techniques are becoming more present in automotive development process. This paper proposed an AWNN approach combined with an artificial bee colony. A good model of the system dynamics can facilitate model-based design.