Caterpillar Inc.

Greg Armstrong

Kranti Kumar Nellutla

Anand Krishnamurthygopalan

James Ulstad

Sylvain Joseph Charbonnel

Steven Y Tian

Brian P Brennan

Sasidhar Rayasam

A fuel injector controller is disclosed. The fuel injector controller may determine respective values of a set of parameters of the engine; process, using a neural network, the respective values to determine a target fuel output of a fuel injector, wherein the neural network is configured to determine the target fuel output based on the set of parameters being an input layer of the neural network; determine, based on an output of the neural network, the target fuel output; and provide the determined target fuel output to the fuel injector to permit the fuel injector to inject fuel according to the target fuel output.

Fuel injection control using a neural network

An engine has an injector communicating with a combustion chamber to introduce a plurality of dosing shots to mix with exhaust gases and regenerate an aftertreatment device downstream of the combustion chamber. An apportionment strategy can apportion a per cylinder quantity, representing the quantity of dosing fuel to introduce per cylinder per combustion cycle, among a predefined number of dosing shots each having a first per shot quantity. The strategy compares the predefined number of dosing shots with a temporal dosing window to determine if predefined number of dosing shots can be conducted within the temporal dosing window. If so, the strategy proceeds to introduce the predefined number of dosing shots and if not, the strategy may recalculate a reduced number of dosing shots and reapportions a second per shot quantity.

Fuel apportionment strategy for in-cylinder dosing

Growing community of inventors

Edwards, IL, United States of America

Greg Armstrong