Vehicle Performance Classification Through Clustered Usage Outliers

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current technologies for predicting vehicle performance issues are limited to diagnosing specific device failures and cannot accurately predict problems resulting from combinations of device interactions, lacking comprehensive analysis of vehicle usage patterns and wear patterns.

Innovation Solution

A method involving clustering reference data to identify vehicle performance by associating second vehicle data with cluster significant parameters, comparing it to clustered reference data, and defining outliers to classify and categorize vehicle performance, enabling predictive maintenance and improved reliability.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If current technology for predicting vehicle problems is used, then specific device failures can be diagnosed, but comprehensive analysis of vehicle usage patterns and wear patterns resulting from combinations of device interactions cannot be accurately predicted

Engineering Contradiction:
Improveprediction accuracyVSAvoidcomprehensive analysis capability
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent segments vehicle data into multiple dimensions including usage patterns, wear patterns, and device interaction patterns. Each dimension is analyzed separately through clustering to identify distinct patterns, which are then integrated to provide comprehensive predictions. This segmentation allows the system to handle complex combinations of causes by breaking them down into analyzable components.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces multiple dimensions of analysis beyond single-device diagnostics, including temporal dimensions (usage patterns over time), spatial dimensions (different operating conditions), and interaction dimensions (combinations of devices). This multi-dimensional approach enables accurate prediction of problems resulting from combinations of causes by analyzing vehicle data across these additional dimensions simultaneously.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Reliability

If clustered reference data is maintained for comprehensive vehicle performance identification, then vehicle usage patterns and wear patterns can be accurately identified, but data processing complexity increases

Engineering Contradiction:
Improvevehicle performance identification accuracyVSAvoiddata processing system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent performs preliminary clustering of reference data during an offline phase to create pre-organized clusters representing different usage patterns and wear patterns. This preliminary action prepares the data structure in advance, so that during online operation, the system only needs to compare new data against the pre-established clusters, significantly reducing real-time processing complexity while maintaining high identification accuracy.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces cluster significance parameters as intermediaries that bridge the gap between complex multi-dimensional vehicle data and simplified classification. These parameters serve as mediators that capture the essential characteristics of each cluster, allowing the system to perform accurate pattern recognition without directly processing the full complexity of the underlying multi-dimensional data at runtime.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Reliability

If outlier detection is implemented to identify irrational usage patterns, then premature failures can be detected early, but false positives may increase

Engineering Contradiction:
Improvefailure detection capabilityVSAvoiddetection accuracy
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent implements feedback mechanisms where the results of outlier detection are continuously evaluated and used to refine cluster definitions and significance parameters. When false positives are detected, the system adjusts the cluster boundaries and significance thresholds to better distinguish between genuine outliers and normal variations. This feedback loop continuously improves detection accuracy while maintaining the ability to identify premature failures.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent dynamically adjusts cluster significance parameters based on the detected data distribution and historical patterns. By changing parameters such as significance thresholds and cluster weights adaptively, the system optimizes the balance between detecting genuine outliers and avoiding false positives. This parameter adaptation allows the system to maintain high reliability in failure detection while minimizing false alarms.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4060576B1A method for identifying vehicle performance
Publication Date: 2024.12.25 VOLVO TRUCK CORP
  • EP4060576B1 patent drawingFigure 1
  • EP4060576B1 patent drawingFigure 2
  • EP4060576B1 patent drawingFigure 3

AI summary

The present invention relates to a method for identifying vehicle performance, the method comprising: maintaining a database with clustered reference data based on first vehicle data, in which each cluster in the clustered reference data is associated with a cluster significant parameter being a cluster threshold indicative of the association of the corresponding cluster; collecting second vehicle data; identifying the second vehicle data with regards to the clusters of the clustered reference data by means of an associated significant parameter and a cluster threshold, and in response of not being able to classify the second vehicle data into a cluster of the clustered reference data, define the second vehicle data as an outlier; identifying vehicle performance based on the cluster identification and determined outlier of the second vehicle data.