| 摘要 |
PROBLEM TO BE SOLVED: To remarkably increase chances for learning dispersion of the injection quantity of an injector and correct the flow characteristic early and effectively in a cylinder injection type engine which is set in a stratified combustion mode in a low load territory after warming up, openly controls a fuel injection quantity based on output of an air flow sensor, in a high load territory, feedback- controls based on output of an O2 sensor, meanwhile at non-warming up, feedback-controls even in a low load territory. SOLUTION: In a stoichiometric mode at non-warming up of an engine and after warming up, the average value cfb-ave of a feedback correcting value is respectively learned as injection quantity deviation (SC6, 14). The learning value ClrnB at small injection pulse width Ti in the idle running condition, the learning value ClrnC at middle injection pulse width Ti in the low/middle load condition, and the learning value ClrnC at large injection pulse width Ti in the middle/high load condition, are respectively learned (SC8, 10, 16), and while giving weight to the respective learning values according to reliability, an injector flow characteristic map M2 is rewritten (SC18).
|
