Abstract
This paper presents a concept-to-implementation study of a control technique based on neurofuzzy technology. The aim of the programme was to explore the issues surrounding the practical utilisation of the neurofuzzy technology in engine control. The control application was engine idle-speed regulation, chosen due to its well defined but demanding non-linear nature.
Fuzzy logic and neurofuzzy techniques provide a strong non-linear framework for the construction of real-world controllers. The solutions generated are simple but robust with the opportunity to incorporate expert knowledge. Large engine non-linearities are dealt with via a controller gain-scheduling mechanism. Tuning of the controllers is achieved by assigning responses to engine operating conditions, and is directly equivalent to modifying large areas of the calibration maps. This will reduce the calibration effort. In this study, controller performance was verified with experimental data from a vehicle and the results discussed with reference to practical controller requirements.