Sensor Tag Autoencoder for Manufacturing Anomaly Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Manufacturers rely on scheduled and reactive maintenance, leading to lost revenue, excess costs, shortened asset life, poor product quality, and safety risks due to unplanned maintenance and increased field exposure.

Innovation Solution

A system using machine learning and autoencoders to predict and detect anomalies in processing pipelines by analyzing data from sensors, identifying correlations, and transmitting alerts for proactive maintenance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If scheduled and reactive maintenance is used to manage operating assets, then maintenance costs are controlled, but lost revenue from deferred production, excess costs from unplanned maintenance, shortened asset life, poor product quality, and personnel safety risks occur

Engineering Contradiction:
Improveasset reliabilityVSAvoidproduction continuity
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The system performs preliminary actions by continuously monitoring asset data and predicting failures before they occur. The anomaly detection system analyzes sensor data, operational parameters, and maintenance records to identify early signs of asset degradation, enabling maintenance to be scheduled proactively rather than reactively, thus preventing unplanned downtime while optimizing maintenance resource allocation

Inventive Principle:
Principle #10Preliminary action

2Ease of operation

If reactive maintenance is implemented, then immediate response to failures is achieved, but asset life is shortened and product quality deteriorates

Engineering Contradiction:
Improvemaintenance responsivenessVSAvoidasset life
Core Design Contradiction:
Ease of operationVSDuration of action of stationary object

Solution Approach 1:

The system implements continuous feedback loops by monitoring asset performance data, comparing it against predicted anomalies, and adjusting maintenance schedules accordingly. The system provides real-time feedback on asset health status, enabling operators to respond to actual conditions rather than following fixed schedules, thereby extending asset life through optimized maintenance timing while maintaining rapid responsiveness to actual failures

Inventive Principle:
Principle #23Feedback

3Measurement precision

If more sensors and monitoring systems are deployed to improve anomaly detection accuracy, then prediction accuracy improves, but system complexity and implementation costs increase

Engineering Contradiction:
Improveanomaly detection accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system achieves universal anomaly detection capability by implementing a unified machine learning framework that can analyze multiple data types from various sensors simultaneously. The platform is designed to handle diverse sensor inputs (vibration, temperature, pressure, flow) and operational data through a single integrated system, reducing overall complexity compared to implementing separate specialized systems for each sensor type while maintaining high detection accuracy across all asset types

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP4028963B1Methods, system and medium for predicting manufacturing process risks
Publication Date: 2025.09.24 C3 AI INC
  • EP4028963B1 patent drawingFigure 1
  • EP4028963B1 patent drawingFigure 2
  • EP4028963B1 patent drawingFigure 3

AI summary

The present disclosure provides system, methods, and computer program products for predicting and detecting anomalies in a subsystem of a system. An example method may comprise (a) determining a first plurality of tags that are indicative of an operational performance of the subsystem. The tags can be obtained from (i) a plurality of sensors in the subsystem and (ii) a plurality of sensors in the system that are not in the subsystem. The method may further comprise (b) processing measured values of the first plurality of tags using an autoencoder trained on historical values of the first plurality of tags to generate estimated values of the first plurality of tags; (c) determining whether a difference between the measured values and estimated values meets a threshold; and (d) transmitting an alert that indicates that the subsystem is predicted to experience an anomaly if the difference meets the threshold.