Misfire detection device for internal combustion engine, misfire detection system for internal combustion engine, data analysis device, and controller for internal combustion engine

Patent metadata

Field	Value
Publication number	US-11099102-B2
Application number	US-202016785705-A
Country	US
Kind code	B2
Filing date	Feb 10, 2020
Priority date	Feb 15, 2019
Publication date	Aug 24, 2021
Grant date	Aug 24, 2021

Abstract

A misfire detection device for an internal combustion engine is provided. A mapping takes time series data of instantaneous speed parameters as inputs. Each instantaneous speed parameter corresponds to one of a plurality of successive second intervals in a first interval. The instantaneous speed parameters correspond to the rotational speed of the crankshaft. The first interval is a rotational angular interval of the crankshaft in which compression top dead center occurs. The second interval is smaller than an interval between compression top dead center positions. The mapping outputs a probability that a misfire has occurred in at least one cylinder that reaches compression top dead center in the first interval. The mapping data defining the mapping has been learned by machine learning.

First claim

What is claimed is:

1. A misfire detection device for an internal combustion engine, comprising:
memory circuitry; and
processor circuitry, wherein
the memory circuitry is configured to store mapping data that defines a mapping that takes time series data of rotation durations as inputs to output probabilities that a misfire has occurred in cylinders of the internal combustion engine, the probabilities being quantified into a continuous variable in a predetermined range that is greater than 0 and less than 1, each rotation duration corresponding to one of a plurality of successive second intervals in a first interval, the mapping including a neural network having one intermediate layer, and a softmax function which normalizes outputs of the neural network so that a sum of the probabilities of misfires is equal to 1,
the processor circuitry is configured to perform:
an obtainment process that obtains the rotation durations based on a detection value of a sensor that detects rotational behavior of a crankshaft of the internal combustion engine;
a determination process that determines presence or absence of a misfire based on an output of the mapping, which takes the time series data as inputs, by calculating original probabilities, which are parameters correlated positively with the probabilities that a misfire has occurred, inputting the original probabilities into the softmax function to obtain the probabilities that a misfire has occurred, and comparing a maximum value of the probabilities that a misfire has occurred to a threshold value; and
a response process that, when the determination process determines that a misfire has occurred, operates predetermined hardware to respond to a situation where the misfire has occurred,

the rotation durations correspond to rotational speed at which the crankshaft rotates by the corresponding one of the second intervals,
the first interval is a rotational angular interval of the crankshaft in which compression top dead center occurs,
the second interval is smaller than an interval between compression top dead center positions,
the mapping outputs a probability that a misfire has occurred in at least one cylinder that reaches compression top dead center in the first interval, and
the mapping data is data that has been learned by machine learning.

2. The misfire detection device for an internal combustion engine according to claim 1 , wherein
the inputs to the mapping include a parameter that defines an operating point of the internal combustion engine,
the obtainment process includes obtaining the parameter that defines the operating point, and
the determination process determines presence or absence of a misfire based on an output of the mapping that is obtained by further including, in the inputs to the mapping, the parameter that defines the operating point and is obtained in the obtainment process.

3. The misfire detection device for an internal combustion engine according to claim 1 , wherein
the inputs to the mapping include a moderator variable that is a parameter for controlling combustion speed of air-fuel mixture in a combustion chamber of the internal combustion engine by operating an operation portion of the internal combustion engine,
the obtainment process includes obtaining the moderator variable, and
the determination process determines presence or absence of a misfire based on an output of the mapping that is obtained by further including, in the inputs to the mapping, the moderator variable that is obtained in the obtainment process.

4. The misfire detection device for an internal combustion engine according to claim 1 , wherein
the inputs to the mapping include a state variable of a driveline system connected to the crankshaft,
the obtainment process includes obtaining the state variable of the driveline system, and
the determination process determines presence or absence of a misfire based on an output of the mapping that is obtained by further including, in the inputs to the mapping, the state variable of the driveline system that is obtained in the obtainment process.

5. The misfire detection device for an internal combustion engine according to claim 1 , wherein
the inputs to the mapping include a road surface state variable of a road surface on which a vehicle on which the internal combustion engine is mounted travels,
the obtainment process includes obtaining the road surface state variable, and
the determination process determines presence or absence of a misfire based on an output of the mapping that is obtained by further including, in the inputs to the mapping, the road surface state variable that is obtained in the obtainment process.

6. The misfire detection device for an internal combustion engine according to claim 1 , wherein
the mapping data includes a plurality of types of mapping data stored in the memory circuitry, and
the determination process includes a selection process that selects one of the plurality of types of mapping data to be used to determine presence or absence of a misfire.

7. The misfire detection device for an internal combustion engine according to claim 1 , wherein
the mapping data includes mapping data sets each associated with one of misfire types that are classified according to an interval between cylinders where misfires occur, and
the determination process uses the mapping data sets to determine presence or absence of misfires of the respective misfire types.

8. The misfire detection device for an internal combustion engine according to claim 1 , wherein
the mapping data includes input-side mapping data, which defines an input-side mapping, and output-side mapping data, which defines an output-side mapping,
the input-side mapping is a nonlinear mapping that outputs data having fewer dimensions than input data, and
the output-side mapping is a nonlinear mapping that takes an output of the input-side mapping as an input to output a probability of a misfire.

9. The misfire detection device for an internal combustion engine according to claim 8 , wherein
the output-side mapping includes:
a torque output mapping that takes the output of the input-side mapping as an input to output data on generated torque of the internal combustion engine; and
a probability mapping that outputs a probability that a misfire has occurred, based on an output of the torque output mapping, and the determination process includes:
a torque calculation process that calculates the data on generated torque by inputting the output of the input-side mapping, which takes the time series data of the rotation durations as inputs, to the torque output mapping; and
a probability calculation process that calculates the probability by inputting the data on generated torque calculated in the torque calculation process to the probability mapping.

10. The misfire detection device for an internal combustion engine according to claim 8 , wherein
the input-side mapping includes:
a plurality of input-side linear mappings that output linear combination data of the time series data of the rotation durations; and
a plurality of input-side nonlinear mappings that perform nonlinear transformation on an output of the respective input-side linear mappings, the output-side mapping includes:
output-side linear mappings that are equal in number to cylinders and output linear combination data of outputs of the input-side nonlinear mappings; and
output-side nonlinear mappings that perform nonlinear transformation on an output of the respective output-side linear mappings to output data on one of generated torque and a probability that a misfire has occurred in the respective cylinders,

in one of the input-side linear

Assignees

• Toyota Motor Co Ltd

Inventors

• Muto Harufumi
• Takamoto Satoshi
• Katayama Akihiro
• Murai Tatsunobu

Classifications

• G06N3/048

  Activation functions · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• F02D41/1405

  Neural network control · CPC title

• F02D41/008

  Controlling each cylinder individually · CPC title