US 5,657,737 A · Air-fuel ratio control system

A vehicle control system which controls a controlled object in a vehicle, particularly in an engine, having an dead time between an input and an output thereof, particularly the dead time varying depending on operating conditions in a vehicle control, particularly in an engine control, said vehicle control system comprising:

a data sampling means which samples an input data sequence from various sensor output data, each of which contains the input or output values of said controlled object or at least one of a plurality of parameters, each of which becomes a factor for the dead time;
a teacher data generating means which generates a teacher data sequence corresponding to each element in said input data sequence sampled by said data sampling means, by manipulating the dead time varying depending on values of the parameters;
a connectivity-coefficients learning means which learns characteristics of said controlled object having the dead time to determine connectivity coefficients based on the input data sequence and the teacher data sequence;
a neuro-computing means which estimates at a timing point k an output result having the dead time corresponding to the input sampled at the timing point k using the connectivity coefficients;
and
a control-amount calculating means which calculates a control amount using the output result of said neuro-computing means as a feedback amount in place of the output of said controlled object sampled by said data sampling means.

3 more elements

An air-fuel ratio control system which controls an air-fuel ratio in an internal combustion engine, said air-fuel ratio control system comprising:

a state detecting sensor group which detects an operating state of said internal combustion engine;
an air-amount detecting sensor group which detects an intake air amount;
an air-fuel ratio sensor which detects the air-fuel ratio of an exhaust gas of said internal combustion engine;
a basic-fuel calculating means which calculates a basic injection amount of fuel based on output data of said sensors and a predetermined data group;
an air-fuel ratio storage means which stores past air-fuel ratio data detected by said air-fuel ratio sensor, renewing the stored data at each control cycle with the newest n data;
an injected-fuel amount storage means which stores past injection amount data of fuel injected to each cylinder in said internal combustion engine, renewing the stored data at each control cycle with the newest m data;
a conversion means which converts each of values of the data detected by said sensor groups and each of values of the air-fuel ratio data and the injected-fuel amount data stored by said storage means into input data for a neural network;
a forward neuro-computing means which reads the input data for the neural network given by said conversion means, and performs in real time a neuro-computation of the neural network, which has been trained beforehand, to output at a time k an estimate A/F.sub.NN of the air-fuel ratio A/F(k+n) detected at a time k+n, where n is a time delay between a fuel injection and a detection of its effect and varies with operating conditions;
a fuel injection correcting means which calculates a correction injection amount of fuel in accordance with output data of said forward neuro-computing means;
and
a fuel injection means which injects a fuel to said internal combustion engine according to a fuel amount obtained by adding the basic injection amount obtained by said basic-fuel calculating means and the correction injection amount obtained by said fuel injection correcting means.

8 more elements

An air-fuel ratio control system which controls an air-fuel ratio in an internal combustion engine, said air-fuel ratio control system comprising:

a state detecting sensor group which detects an operating state of said internal combustion engine;
an air-amount detecting sensor group which detects an intake air amount;
an air-fuel ratio sensor which detects the air-fuel ratio of an exhaust gas of said internal combustion engine;
a basic-fuel calculating means which calculates a basic injection amount of fuel based on output data of said sensors and a predetermined data group;
an air-fuel ratio storage means which stores, at each control cycle, past air-fuel ratio data detected by said air-fuel ratio sensor, renewing the stored data with the newest n data;
an injected-fuel amount storage means which stores, at each control cycle, past data of amount of fuel injected to each cylinder in said internal combustion engine, renewing the stored data with the newest m data;
a conversion means which converts each of values of data detected by said sensor groups, each of values of the air-fuel ratio data and the injected-fuel amount data stored by said storage means, and a target air-fuel ratio into input data for a neural network;
a backward neuro-computing means which reads the input data for the neural network given by said conversion means and performs in real time a neuro-computation of the neural network to output a correction injection amount of fuel so as to make the air-fuel ratio equal to the target air-fuel ratio;
and
a fuel injection means which injects a fuel to said internal combustion engine according to a fuel amount obtained by adding the basic injection amount obtained by said basic-fuel calculating means and the correction injection amount obtained by said backward neuro-computing means.

7 more elements

An air-fuel ratio control system which controls an air-fuel ratio in an internal combustion engine, said air-fuel ratio control system comprising:

a state detecting sensor group which detects an operating state of said internal combustion engine;
an air-amount detecting sensor group which detects an intake air amount;
an air-fuel ratio sensor which detects the air-fuel ratio of an exhaust gas of said internal combustion engine;
a basic-fuel calculating means which calculates a basic injection amount of fuel based on output data of said sensors and a predetermined data group;
an air-fuel ratio storage means which stores, at each control cycle, past air-fuel ratio data detected by said air-fuel ratio sensor, renewing the stored data with the newest n data;
an injected-fuel amount storage means which stores, at each control cycle, past data of amount of fuel injected to each cylinder in said internal combustion engine, renewing the stored data with the newest m data;
a conversion means which converts each of values detected by said sensor groups and each of values of the air-fuel ratio data and the injected-fuel amount data stored by said storage means into input data for a neural network;
a forward neuro-computing means which reads the input data for the neural network given by said conversion means and performs in real time a neuro-computation of the neural network, which has been trained during the time when a canister purge does not occur, to output an estimate A/F.sub.NN of the air-fuel ratio varying with a delay against a fuel injection;
a purge-amount estimating means which estimates, at each control cycle, a purge amount from past p pieces of the air-fuel ratio estimates A/F.sub.NN and past q pieces of the detected air-fuel ratio data respectively output from said forward neuro-computing means and said air-fuel ratio sensor;
a correction-amount calculating means which calculates a correction injection amount of fuel based on output data of said forward neuro-computing means and output data of said purge-amount estimating means so as to make the air-fuel ratio equal to a target air-fuel ratio;
and
a fuel injection means which injects a fuel to said internal combustion engine according to a fuel amount obtained by adding the basic injection amount obtained by said basic-fuel calculating means and the correction injection amount obtained by said correction-amount calculating means.

9 more elements

An air-fuel ratio control system which controls an air-fuel ratio in an internal combustion engine, said air-fuel ratio control system comprising:

a state detecting sensor group which detects an operating state of said internal combustion engine;
an air-amount detecting sensor group which detects an intake air amount;
an air-fuel ratio sensor which detects the air-fuel ratio of an exhaust gas of said internal combustion engine;
a basic-fuel calculating means which calculates a basic injection amount of fuel based on output data of said sensors and a predetermined data group;
a valve-state detecting means which detects a state of a purge valve that opens and closes a purge line in said internal combustion engine;
an air-fuel ratio storage means which stores, at each control cycle, past data detected by said air-fuel ratio sensor, renewing the stored data with the newest n data;
an injected-fuel amount storage means which stores, at each control cycle, past data of amount of fuel injected to each cylinder in said internal combustion engine, renewing the stored data with the newest m data;
a conversion means which converts each of values detected by said sensor groups, the state of the purge valve detected by said valve-state detecting means, and each of values of the air-fuel ratio data and the injected-fuel amount data stored by said storage means into input data for a neural network;
a purge-amount calculating means which reads output values output from said conversion means and performs in real time a computation of the neural network to output a purge amount of a canister;
a correction-amount calculating means which calculates a correction injection amount of fuel using the output data of said air-fuel ratio sensor and output data of said purge-amount calculating means so as to make the air-fuel ratio equal to a target air-fuel ratio;
and
a fuel injection means which injects a fuel to said internal combustion engine according to a fuel amount obtained by adding the basic injection amount obtained by said basic-fuel calculating means and the correction injection amount obtained by said correction-amount calculating means.

9 more elements

An air-fuel ratio control system which controls an air-fuel ratio in an internal combustion engine, said air-fuel ratio control system comprising:

an engine speed detecting sensor which detects an engine speed of said internal combustion engine;
an intake manifold pressure sensor which detects a pressure inside an intake manifold;
a throttle position sensor which detects a position of a throttle valve;
an intake air temperature sensor which detects a temperature of an intake air;
a coolant temperature sensor which detects a temperature of a coolant;
a basic-fuel calculating means which calculates a basic injection amount of fuel based on output data of said sensors and a predetermined data group;
an EGR amount detecting means which detects an EGR amount;
an air-fuel ratio sensor which detects an air-fuel ratio of an exhaust gas of said internal combustion engine at a timing corresponding to a rotation cycle of said engine, particularly at every crankshaft angle of 180.degree.;
an air-fuel ratio storage means which stores past data detected by said air-fuel ratio sensor, renewing the stored data at each engine step, that is 180.degree. crankshaft rotation, with the newest N samples, where N is a number of cylinders in said engine;
an air-fuel-ratio input selecting means which selects input data for a neural network from the data stored by said air-fuel ratio storage means;
an injected-fuel amount storage means which stores, at each engine step, past data of amount of fuel injected to each cylinder, renewing the stored data with the newest N+1 data, where N is the number of the cylinders;
an injected-amount input selecting means which selects input data for the neural network from the data stored by said injected-fuel amount storage means;
a conversion means which converts each of values detected by said sensors, the EGR amount, the injected-fuel amount data selected by said injected-amount input selecting means, and the air-fuel ratio data selected by said air-fuel ratio input selecting means into input data for the neural network;
a neuro-computing means which reads the input data for the neural network given by said conversion means and performs in real time a neuro-computation of the neural network to output an air-fuel ratio, which is a value of an internal state variable of said internal combustion engine;
a differential coefficients identifying means which expresses the air-fuel ratio as a non-linear function f.sub.NN Of internal state variables, whose values are the input data for said neural network obtained by said conversion means, and estimates differential coefficients of a function F.sub.NN which is obtained by partially differentiating said function f.sub.NN with said internal state variables, based on connectivity coefficients and intermediate-layer output data of said neuro-computing means;
a correction-amount calculating means which calculates a correction injection amount of fuel using parameters obtained by said differential coefficients identifying means so as to make the air-fuel ratio equal to a target air-fuel ratio;
and
a fuel injection means which injects a fuel to said internal combustion engine according to a fuel amount obtained by adding the basic injection amount obtained by said basic-fuel calculating means and the correction injection amount obtained by said correction-amount calculating means.

15 more elements

An air-fuel ratio control system which controls an air-fuel ratio in an internal combustion engine, said air-fuel ratio control system comprising:

an engine speed detecting sensor which detects an engine speed of said internal combustion engine;
an intake manifold pressure sensor which detects a pressure inside an intake manifold;
a throttle position sensor which detects a position of a throttle valve;
an intake air temperature sensor which detects a temperature of an intake air;
a coolant temperature sensor which detects a temperature of a coolant;
a basic-fuel calculating means which calculates a basic injection amount of fuel based on output data of said sensors and a predetermined data group;
an EGR amount detecting means which detects an EGR amount;
an injected-fuel amount detecting means which detects a fuel amount actually injected to each of cylinders in said internal combustion engine;
an air-fuel ratio sensor which detects the air-fuel ratio of an exhaust gas of said internal combustion engine;
a data storage means which stores values of data detected by said sensors, the EGR amount, and the injected fuel amount at each engine step, that is 180.degree. crankshaft rotation, renewing the stored data with the newest data at timing points k.sub.1 to k.sub.n ;
a conversion means which converts the values of the data stored by said data storage means into input data for a neural network;
a neuro-computing means which reads the input data for the neural network given by said conversion means and performs in real time a neuro-computation of the neural network to output an estimate of the air-fuel ratio, which is a value of an internal state variable of said internal combustion engine;
a correction-amount calculating means which calculates a correction injection amount of fuel based on the air-fuel ratio estimate so as to make the air-fuel ratio equal to a target air-fuel ratio;
and
a fuel injection means which injects a fuel to said internal combustion engine according to a fuel amount obtained by adding the basic injection amount obtained by said basic-fuel calculating means and the correction injection amount obtained by said correction-amount calculating means.

12 more elements

An air-fuel ratio control system which controls an air-fuel ratio in an internal combustion engine, said air-fuel ratio control system comprising:

an engine speed detecting sensor which detects an engine speed of said internal combustion engine;
an intake manifold pressure sensor which detects a pressure inside an intake manifold;
a throttle position sensor which detects a position of a throttle valve;
an intake air temperature sensor which detects a temperature of an intake air;
a coolant temperature sensor which detects a temperature of a coolant;
a basic-fuel calculating means which calculates a basic injection amount of fuel based on output data of said sensors and a predetermined data group;
an EGR amount detecting means which detects an EGR amount;
an O.sub.2 sensor which detects an oxygen concentration in an exhaust gas;
an oxygen-concentration integrating means which integrates output data of said O.sub.2 sensor to calculate an integral value of the oxygen concentration;
an oxygen-concentration differentiating means which calculates a differential value of output data of said O.sub.2 sensor;
an injected-amount detecting means which detects a fuel amount actually injected to each of cylinders in said internal combustion engine;
a conversion means which converts values of the data detected by said sensors, the EGR amount, the detected amount of fuel injection, the integral value of oxygen concentration, and the differential value of oxygen concentration into input data for a neural network;
a neuro-computing means which reads the input data for the neural network given by said conversion means and performs in real time a neuro-computation of the neural network to output an estimate of actual air-fuel ratios, which are values of an internal state variable of said internal combustion engine;
a correction-amount calculating means which calculates a correction injection amount of fuel based on the air-fuel ratio estimate so as to make the air-fuel ratio equal to a target air-fuel ratio;
and
a fuel injection means which injects a fuel to said internal combustion engine according to a fuel amount obtained by adding the basic injection amount obtained by said basic-fuel calculating means and the correction amount obtained by said correction-amount calculating means.

13 more elements

Air-fuel ratio control system

Life of this patent

Air-fuel ratio control system

Life of this patent

Abstract

Claims22