Logfile Trigger Token Detection for Predictive Maintenance

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for predicting technical defects in computer-based systems are inefficient and fail to account for diverse and non-uniform event descriptions in logfiles, leading to manual effort and limited analysis capabilities, especially with Natural Language Processing (NLP) methods.

Innovation Solution

A computer-implemented method extracts tokens from event descriptions in logfiles based on token categories, correlating their time stamps with the system's technical defect, to identify a trigger token that predicts maintenance needs, using regular expressions and trained functions for efficient analysis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual monitoring of event descriptions is performed to predict technical defects, then relevant events can be identified, but the manual effort is very great and the most suitable event descriptions are frequently not defined

Engineering Contradiction:

Improveprediction accuracyVSAvoidmanual effort

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces manual monitoring and analysis of event descriptions with an automated computer program that processes logfiles. The system automatically extracts tokens, determines trigger tokens based on time stamp correlations, and identifies technical defects without human intervention, thereby eliminating the excessive manual effort while maintaining prediction accuracy.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables automated self-diagnosis and self-monitoring by having the computer program independently analyze logfiles, extract meaningful tokens, correlate time stamps with technical defects, and identify patterns without requiring manual definition of event descriptions. The system serves itself by automatically performing tasks that would otherwise require human analysts.

Inventive Principle:

Principle #25Self-service

2Loss of information

If NLP methods are applied to analyze logfile event descriptions, then relevant events can be recognized, but the processing effort is very great and not all relevant event descriptions are taken into account

Engineering Contradiction:

Improveevent description analysis completenessVSAvoidprocessing effort

Core Design Contradiction:

Loss of informationVSPower

Solution Approach 1:

The patent extracts only the necessary tokens from event descriptions based on predefined token categories rather than analyzing entire sentences or sequences. This selective extraction approach removes unnecessary processing overhead while retaining the essential information needed for defect prediction, thereby reducing processing effort without compromising analysis completeness.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent segments event descriptions into discrete tokens categorized by type (e.g., error codes, status indicators, component identifiers). This segmentation allows the system to process individual tokens independently and efficiently, reducing the overall processing burden compared to analyzing complete event descriptions while maintaining the ability to identify all relevant patterns.

Inventive Principle:

Principle #1Segmentation

3Reliability

If complete analysis of all event descriptions is performed to ensure rare predictive events are not missed, then prediction accuracy improves, but the processing effort becomes unmanageable

Engineering Contradiction:

Improvedefect prediction reliabilityVSAvoidanalysis throughput

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent replaces exhaustive manual or comprehensive automated analysis with a targeted automated system that uses time stamp correlation to efficiently identify trigger tokens. The computer program automatically processes logfiles at high speed, correlating token occurrences with subsequent technical defects, thereby maintaining reliable defect prediction while achieving high analysis throughput that would be impossible through complete manual analysis.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS12380273B2Method for provision of a trigger token

Publication Date: 2025.08.05 SIEMENS HEALTHINEERS AG

AI summary

A trigger token correlates with a need for maintenance of a system. A computer-implemented method comprises: receiving a logfile of the system, wherein the logfile includes a plurality of event descriptions, and a subset of the plurality of event descriptions is assigned a time stamp; receiving a point in time of a technical defect of the system; extracting tokens from the subset of the event descriptions based on a token category, wherein each of the extracted tokens is contained as a character string in the subset of the plurality of event descriptions; determining the trigger token from the extracted tokens based on a correlation of time stamps assigned to the extracted tokens with the point in time of the technical defect of the system; and provisioning the trigger token.