Engine Controller Dual Learning Mechanism Friction Adaptation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing engine controllers face challenges in accurately learning and adapting to variations in loss torque characteristics, particularly due to instantaneous friction changes, which can lead to prolonged convergence times for reliable learning values.

Innovation Solution

An engine controller with dual learning mechanisms: a first learning means that updates values based on stored learning values for improved reliability and a second means that updates values directly based on actual loss torque characteristics, allowing for prompt convergence to accurate learning values.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the learning value is smoothed to avoid instantaneous variation in loss torque, then the reliability of the learning value is improved, but the convergence speed to the actual loss torque characteristic deteriorates

Engineering Contradiction:
Improvereliability of learning valueVSAvoidconvergence time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent implements a dynamic switching mechanism between two learning modes: a first learning means that provides smoothed, reliable learning values under normal conditions, and a second learning means that provides rapid convergence when friction changes are detected. The system dynamically selects the appropriate learning mode based on real-time detection of friction change conditions, thereby resolving the contradiction between reliability and convergence speed.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes the learning parameter update strategy based on operating conditions. When friction change is detected, the system switches from gradual smoothing updates to direct updates using the second learning means, allowing the learning value to rapidly adapt to new friction characteristics while maintaining reliability during stable operation.

Inventive Principle:
Principle #35Parameter changes

2Stability of the object's composition

If the learning value is updated gradually based on stored learning values, then the reliability and stability are improved, but the adaptability to actual loss torque characteristic deteriorates

Engineering Contradiction:
Improvestability of learning valueVSAvoidadaptability to loss torque characteristic
Core Design Contradiction:
Stability of the object's compositionVSAdaptability or versatility

Solution Approach 1:

The system dynamically adjusts the learning update strategy by switching between two learning means. The first learning means maintains stability through gradual updates, while the second learning means enables rapid adaptation when friction changes are detected. This dynamic switching resolves the contradiction between stability and adaptability.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent incorporates feedback through the friction change detection mechanism that monitors whether actual friction conditions have changed. Based on this feedback, the system selects the appropriate learning mode: gradual updates for stability during normal operation, or rapid updates for adaptability when friction changes are detected.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP2241739B1Controller of an internal combustion engine for synchronizing the engine during the engine shutdown
Publication Date: 2019.12.25 DENSO CORP
  • EP2241739B1 patent drawingFigure 1
  • EP2241739B1 patent drawingFigure 2A~2B
  • EP2241739B1 patent drawingFigure 3A~3B

AI summary

An ECU includes a backup RAM which stores a learning value of a loss torque characteristic of an engine. The learning value is updated based on the loss torque characteristic which is computed based on an actual engine rotation behavior. The ECU includes a first learning portion (S109) for updating the learning value based on the learning value stored in the memory (Tloss) and a presently computed loss torque characteristic (Tg), and a second learning portion (S112) for updating the learning value based on the presently computed loss torque characteristic (Tlossg) without using the learning value stored in the memory. A switching portion (S107) switches between the first learning portion and the second learning portion in order to update the learning value (Tloss).