Developing a neural network and real genetic algorithm combined tool for an engine test bed

Abstract

In the automotive industry, engine test engineers are required to deal with a huge quantity of experimental data obtained from engine test beds each day. Those data must be analysed to evaluate engine performance and to guide further engine test operations. In order to improve efficiency and reduce expenditure of time in engine testing, it is very important for engine test bed controllers to develop a mathematical model from existing engine test data. This paper presents an investigation of a neural network-genetic algorithm (GA) combined tool for engine modelling. In the modelling tool, a real-coded GA has been employed to train three different groups of neural networks (a multilayer perceptron group, a radial basis function group, and a bar function networks group) and then finally to find the most suitable neural network model for engine modelling. The experimental results given in this paper show that the proposed tool has been successfully used for Rover engine testing.result, based on a two-cylinder, four-stroke direct JAUTO229

Soft computing techniques

“…They have been used for many optimization tasks, in particular by combining them with other techniques of artifi cial intelligence, as shown in [31][32][33][34][35][36]. One of the most fundamental principles in our world is the search for an optimal state [32].…”

Neuro-genetic optimization of disc brake performance at elevated temperatures

“…As one of the popular artificial intelligence methods, genetic algorithms represent search algorithms based on the mechanics of natural selection and natural genetics [21]. Until now, they have been used for many optimization tasks, particularly combined with other techniques of artificial intelligence, as it is shown in [22,23,24]. The basic GA optimization cycle is shown in Fig.…”

Investigation of a multi-objective optimization tool for engine calibration

“…In this section, engine multi-objective optimization by the proposed approach is performed on the Jaguar engine AJV8. The experiments are based on neural network modelling [33] for the Jaguar engine AJV8. The input variables are the engine speed, load efficiency (GN), ignition timing (IG), variable valve timing (VVT), and exhaust gas recirculation (EGR) rate.…”

“…It is also the size of the non-dominated solutions of each trial, the size of the final non-dominated solutions of each MOGA, and the size of the final non-dominated solutions of all MOGAs. In each single run of MOGA, a random combination of selection, crossover, and mutation operators [33] is applied. The reason for this is that different selection, crossover, and mutation methods may influence the efficiency of the search process since GAs are random search techniques [1][2][3].…”