Calibration Envelope Modeling for High-Dimensional Control Variables

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Calibrating internal combustion engines and other technical systems with a large number of control variables is complex due to non-linear influences and dependencies, making it difficult to determine whether new control variables are within the drivability limit using existing discrete data points, especially in high-dimensional test spaces.

Innovation Solution

The method models a conical data envelope using different opening angles for radial basis functions and interpolation conditions, allowing for efficient checking of whether an optimized test point lies inside or outside the data envelope, even in high-dimensional spaces, and considers data points outside the data shell for more accurate modeling.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If discrete data points are used to define the drivability limit, then the calibration can be performed with limited measurements, but it becomes impossible to determine whether new control variable combinations are within the drivability limit

Engineering Contradiction:
Improvenumber of data pointsVSAvoiddeterminability of drivability limit compliance
Core Design Contradiction:
Quantity of substanceVSReliability

Solution Approach 1:

The patent introduces an implicit function f(u) as an intermediary between the discrete data points and the drivability limit determination. This function acts as a mediator that takes control variable vectors as input and returns a scalar value indicating whether the point lies within the drivability limit, thereby enabling continuous evaluation without requiring exhaustive discrete data coverage

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent creates a mathematical model (implicit function) that copies or represents the drivability limit surface defined by the discrete data points. Instead of working directly with the discrete points, the implicit function creates a continuous representation that can be evaluated at any point in the control variable space, allowing determination of drivability limit compliance for new control variable combinations

Inventive Principle:
Principle #26Copying

2Manufacturing precision

If the number of control variables is increased to improve system control, then the calibration accuracy improves, but the computational complexity and dimensionality of the test space increases

Engineering Contradiction:
Improvecalibration accuracyVSAvoiddimensionality of test space
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent extracts the essential characteristic of the drivability limit by representing it through a single implicit function f(u) = 0 rather than storing and processing all discrete data points. This extraction reduces the computational burden by focusing on the boundary definition rather than the complete data set, making high-dimensional calibration tractable

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent changes the parameter representation from discrete data points to a continuous implicit function with parameters that define the drivability limit surface. This parameter transformation allows the system to handle high-dimensional control variable spaces by working with a compact mathematical representation rather than exhaustive data sets

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3458699B1Method for calibrating a technical system
Publication Date: 2022.03.30 AVL LIST GMBH
  • EP3458699B1 patent drawingFigure 1~3
  • EP3458699B1 patent drawingFigure 2
  • EP3458699B1 patent drawing

AI summary

In order to be able to check the compliance with a data envelope in a simple and rapid manner when calibrating a technical system, provision is made for the data envelope (D) to be modelled by a function - formula (I) with formula (II) and a predefined centre (c) of the data points (Xn), or formula (III) with a radial base function (ϕ) and coefficients (cn), and for the test point (Zj) to be considered to be within the data envelope (D) if the condition in formula (IV) or formula (V) with a predefined opening angle (cp) or a predefined contour line (f) is satisfied.