Machine Learning Valve Fault Detection in Variable Displacement Engines

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Internal combustion engines, particularly variable displacement engines, face challenges in detecting faults with intake and exhaust valves that do not properly open or close as commanded, leading to issues like oil seepage into the cylinder due to low in-cylinder pressure during skipped cycles, for which existing technologies lack effective solutions.

Innovation Solution

A system and method utilizing machine learning to detect valve faults by training a neural network model with operational parameters, generating a valve fault flag when the probability of valve behavior does not match the commanded action, thereby identifying faulty valve operations and preventing oil seepage.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Use of energy by moving object

If variable displacement operation is implemented by skipping cylinders, then fuel efficiency is improved, but in-cylinder pressure becomes too low causing oil to be sucked into the working chamber

Engineering Contradiction:

Improvefuel efficiencyVSAvoidoil seepage into cylinder

Core Design Contradiction:

Use of energy by moving objectVSObject-affected harmful factors

Solution Approach 1:

The re-breathing strategy applies preliminary anti-action by opening the exhaust valve at the end of one working cycle and the intake valve during the next skipped working cycle to increase in-cylinder pressure before oil seepage can occur, preventing the harmful effect of oil entering the cylinder during variable displacement operation

Inventive Principle:

Principle #9Preliminary anti-action

Solution Approach 2:

The machine learning model performs preliminary action by continuously monitoring valve operation and predicting potential valve faults before they cause significant performance degradation or oil seepage issues, allowing preventive maintenance actions to be taken

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If machine learning model is implemented for valve fault detection, then valve fault detection accuracy is improved, but device complexity increases

Engineering Contradiction:

Improvevalve fault detection accuracyVSAvoidsystem complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The machine learning model applies self-service by automatically monitoring valve operation, comparing actual valve behavior against expected patterns, and generating fault flags without requiring external intervention or complex additional hardware, making the system self-diagnostic

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces complex mechanical valve monitoring systems with a machine learning-based software model that processes existing sensor data to detect valve faults, substituting mechanical complexity with computational intelligence

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS11434839B2Use of machine learning for detecting cylinder intake and/or exhaust valve faults during operation of an internal combustion engine

Publication Date: 2022.09.06 TULA TECHNOLOGY INC

AI summary

A system and method for the use of machine learning for detecting faults for cylinder intake and/or exhaust valves that do not properly open or close as commanded and for generating a flag for such faults.