Sensor Maintenance Timing Using Machine Learning Deterioration Models

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current sensor maintenance practices in industrial plants are inefficient, leading to increased costs due to unpredictable sensor deterioration and the need for regular adjustments or replacements, which can be triggered by factors like environmental conditions and usage situations, making it difficult to determine the optimal maintenance time and resulting in unnecessary maintenance work.

Innovation Solution

An information processing device that acquires sensor data and maintenance information, uses machine learning to generate a determination model by associating maintenance data with label information, and generates a maintenance plan to accurately predict sensor maintenance needs, reducing unnecessary maintenance and costs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If regular maintenance is performed on sensors at fixed intervals, then sensor reliability is maintained, but maintenance costs increase due to unnecessary maintenance work

Engineering Contradiction:

Improvesensor reliabilityVSAvoidmaintenance costs

Core Design Contradiction:

ReliabilityVSLoss of energy

Solution Approach 1:

The system changes the maintenance parameter from fixed time intervals to condition-based thresholds. By continuously monitoring sensor output values and comparing them against dynamically determined reference values, maintenance is triggered only when actual deterioration occurs, optimizing the balance between reliability and cost

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The sensor system performs self-diagnosis by comparing its own output values against reference values. The sensor effectively monitors its own health status and triggers maintenance alerts autonomously, eliminating the need for external定期检查 and reducing unnecessary maintenance interventions

Inventive Principle:

Principle #25Self-service

2Productivity

If machine learning is applied to sensor data to predict deterioration, then maintenance optimization is improved, but measurement precision deteriorates due to the influence of regular maintenance work on teacher data

Engineering Contradiction:

Improvemaintenance optimizationVSAvoiddeterioration detection accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system performs preliminary separation of maintenance-related data points before generating teacher data. By identifying and excluding data collected during or after maintenance activities, the training dataset cleanly represents only natural deterioration patterns, ensuring high measurement precision in deterioration detection

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system extracts and removes maintenance-influenced data points from the overall dataset. By isolating and excluding these problematic data points, the machine learning model trains only on pure deterioration signals, maintaining high accuracy in predicting actual sensor degradation

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS11126150B2Information processing device, information processing method, and storage medium

Publication Date: 2021.09.21 YOKOGAWA ELECTRIC CORP

AI summary

An information processing device according to one aspect of the present invention includes a first acquirer configured to acquire measured data of a sensor, a second acquirer configured to acquire maintenance information related to maintenance performed on the sensor, a learner configured to learn teacher data in which the acquired maintenance information as label information is associated with the acquired measured data to generate a determination model, and a storage storing the generated determination model.