Model-based cylinder charge detection for an internal combustion engine

Patent metadata

Field	Value
Publication number	US-10533510-B2
Application number	US-201515111345-A
Country	US
Kind code	B2
Filing date	Jan 19, 2015
Priority date	Jan 17, 2014
Publication date	Jan 14, 2020
Grant date	Jan 14, 2020

Abstract

A method for a model-based determination of a cylinder charge of a combustion chamber of an internal combustion engine as well as an internal combustion engine in a computer program product. The method utilizes a neuronal network having at least three input values. A pressure quotient is used as one of the input values. The pressure quotient is determined as the ratio of the pressure of the air set by the engine over the operating pressure of the engine. The pressure of the air set by the internal combustion engine may be determined by utilizing a measured value, a computed value, and/or a value determined from a characteristic map. It is also possible to include a combination of these in the pressure quotient.

First claim

The invention claimed is:

1. A method for model-based determination of a cylinder charge of a combustion chamber of an internal combustion engine, wherein the internal combustion engine has a variable valve lift which is considered in the model-based determination, comprising the steps of:
modeling, in a computer, the cylinder charge of the combustion chamber using at least one neuronal network and inputting at least three input values into the neuronal network;
determining a pressure quotient in a module of the computer for use as one of the at least three input values, the pressure quotient being at least indirectly determined from a pressure of the air sucked by the internal combustion engine and from an operating pressure, at least one of the other inputs selected from the group of a rotary speed of the internal combustion engine, a valve lift of the internal combustion engine, an inlet valve camshaft phase, and an outlet valve camshaft phase; and
using the determined cylinder charge for control of the internal combustion engine.

2. The method according to claim 1 , wherein in the pressure quotient, is a pressure of the air sucked by the internal combustion engine, a suction tube pressure or a charging pressure.

3. The method according to claim 1 , wherein the pressure quotient includes, as a pressure of the air sucked by the internal combustion engine, a measured value, a computed value and/or a value determined from a characteristic map.

4. The method according to claim 1 , comprising multiplying an output value of the neuronal network by an operating pressure and then dividing by a value characterizing a standard pressure-whereby a correction of the determined cylinder charge is performed in dependence on the geographic altitude where the internal combustion engine is located.

5. The method according to claim 1 , wherein one of the at least three input values includes at least one phase of a camshaft phase, wherein use is made of an inlet valve camshaft phase and/or outlet valve camshaft phase.

6. The method according to claim 1 , wherein, for correction of deviations, there is performed an adjustment of values used in the model-based determination, utilizing of values of an output value of the neuronal network, of a lambda control, of a suction tube pressure controller and/or of an air-mass measurement device.

7. The method according to claim 6 , wherein an adjustment of the output value of the neuronal network is performed with the aid of values from an air-mass measurement device as soon as, both by the air-mass measurement device and the lambda control, a respective deviation is determined from the value of the cylinder charge delivered by the neuronal network.

8. The method according to claim 7 , wherein an adjustment of a fuel amount is performed by the lambda control as soon as a respective deviation is determined from the value of the cylinder charge delivered by the neuronal network both by the air mass measurement device and the lambda control.

9. The method according to claim 6 , wherein a respective deviation is determined from the value of the cylinder charge delivered by the neuronal network by means of the output value of the neuronal network an adjustment of a value of an air-mass measurement device is performed as soon as, by the air-mass measurement device as well as by the neuronal network as well as by the suction tube pressure controller.

10. The method according to claim 6 , wherein an adjustment of a modeled throttle flap mass flow is performed by the suction tube pressure controller as soon as, both by the air-mass measurement device and the suction tube pressure controller, a respective deviation is determined from the value of the cylinder charge delivered by the neuronal network.

11. The method according to claim 1 , wherein a value of an output value is computed by performing a run through the neuronal network for a first time, with a pressure quotient using a measured pressure and with further input values, and by performing a run through the neuronal network for a second time, with a pressure quotient using a computed pressure and with further input values, and, subsequently, performing a local linear regression between the output value of said first run and the output value of said second run.

12. A computer program product for an engine computer program product being loaded in an engine control, the engine control having a storage medium that comprises one or more programming instructions stored thereon for causing the engine control to control a cylinder charge for a combustion chamber, an engine having a sensor measuring a measured pressure of air sucked by the engine, the computer program product comprising:
a neuronal network adapted to receive at least three inputs, one of the inputs being a pressure quotient determined by the ratio of the measured pressure received from the sensor by the engine control and an operating pressure, at least one of the other inputs selected from the group of a rotary speed of the internal combustion engine, a valve lift of the internal combustion engine, an inlet valve camshaft phase, and an outlet valve camshaft phase, the neuronal network further having an output a module adapted to determine the cylinder charge using the output of the neuronal network for control of the engine.

13. An internal combustion engine having a combustion chamber and a variable lift valve comprising;
a sensor for measuring a pressure of air sucked by the engine;
an engine control connected to the variable lift valve;
the engine control further having a neuronal network and a module of a computer program product, the neuronal network adapted to receive at least three inputs, one of the at least three inputs being a pressure quotient determined by the ratio the measured pressure from the sensor and an operating pressure, the other inputs selected from the group of a rotary speed of the internal combustion engine, a valve lift of the internal combustion engine, an inlet valve camshaft phase, and an outlet valve camshaft, the module of the computer program product using the neuronal network to determine the cylinder charge for the combustion chamber, the determined cylinder charge is used for control of the internal combustion engine.

14. A method of controlling the cylinder charge for a combustion chamber of an internal combustion engine having a lift valve, the method comprising;
measuring the air pressure sucked by the combustion chamber;
determining a pressure quotient as a ratio of the measured air pressure to the operating pressure;
inputting at least three inputs into a neuronal network, one of the inputs being the pressure quotient and at least one of the other inputs selected from the group of a rotary speed of the internal combustion engine, a valve lift of the internal combustion engine, an inlet valve camshaft phase, and an outlet valve camshaft phase to determine an output;
using the output of the neuronal network to determine the charge for the combustion chamber; and
using the determined charge of the combustion chamber for control of the internal combustion engine.

Assignees

• Fev Gmbh
• Hyundai Kefico Corp
• Hyundai Autron Co Ltd

Inventors

• Gotter Andreas
• Merts Menno
• Karaaslan Nuri

Classifications

• F02D2041/001

  for engines with variable valve actuation · CPC title

• F02D2200/0411

  Volumetric efficiency · CPC title

• F02D41/1405 Primary

  Neural network control · CPC title

• F02D13/0207

  changing valve lift or valve lift and timing · CPC title

• F02D2200/0406

  Intake manifold pressure · CPC title