Engine universal characteristic modeling based on improved ant colony optimization

There have been some mathematics methods to model farm vehicle engine universal characteristic mapping (EUCM). Nevertheless, any of different mathematics methods used would possess its own strengths and weaknesses. As a result, these modeling methods about EUCM are not the same among the most vehicle manufacturers. In order to obtain a better robustness EUCM, an improved ant colony optimization was introduced into a traditional cubic surface regression method for modeling EUCM. Based on this method, the test data were regressed into a three-dimensional cubic surface, after that it was cut by some equal specific fuel consumption (ESFC) planes, more than twenty two-dimensional ESFC equations were obtained. Furthermore, the engine speed in every ESFC equation was discretized to obtain a set of ESFC points, and this set of ESFC points was linked into a closed curve by a given sequence via the improved ant colony algorithm. In order to improve the modeling speed, dimensionality reduction and discretization methods were adopted. In addition, a corresponding simulation platform was also developed to obtain an optimal system configuration. There were 48 000 simulation search tests carried out on the platform, and the major parameters of the algorithm were determined. In this way the EUCM was established successfully. In contrast with other methods, as a result of the application of the novel bionic intelligent algorithm, it has better robustness, less distortion and higher calculating speed, and it is available for both gasoline engines and diesel engines.