PURPOSE: To sufficiently actuate a learning function even if the deviation of an air-fuel ratio which exceeds a learning range of a conventional learning value occurs by introducing a learning value whose learning speed is much slower than the conventional learning value in addition to the conventional learning value.
CONSTITUTION: A learning value A in a learning area in response to an operational condition at the time is read out and a learning value B of a memory 34 is read out from a learning map 33 respectively by an air-fuel ratio learning value calculation means 35, and the sum learning values A, B is calculated as the air-fuel ratio learning value. A basic injection quantity is corrected on the basis of the calculated air-fuel ratio learning value by an injection quantity calculation means 36. On the other hand, the learning value A is renewed based on an air-fuel ratio feedback correction quantity a by a learning value renewing means 38 in a narrow learning range but a comparatively learning speed but the learning value B is renewed by another learning value renewing means 39 based on the air-fuel ratio learning value at a learning speed much slower than the speed of the learning value A.
MATSUNO OSAMU
KISHIMOTO YOICHI
KAWAMURA KATSUHIKO