Sensor Monitoring Threshold Control During Signal Changes

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In large-scale infrastructure systems, such as power plants and manufacturing apparatuses, erroneous anomaly detections occur due to the conjunction of prediction values with measurement values and relative shortages in training data, leading to unnecessary alarms and economic losses.

Innovation Solution

A monitoring apparatus that acquires measurement values from multiple sensors, uses a machine learning model to predict sensor values, and adjusts thresholds based on control signal changes to prevent erroneous detections by setting higher thresholds during periods of control signal changes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a machine learning model is used to predict sensor values for anomaly detection, then anomaly detection capability is improved, but erroneous detections increase due to conjunction of prediction values with measurement values and shortages in training data

Engineering Contradiction:
Improveanomaly detection capabilityVSAvoiddetection accuracy
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The system performs preliminary action by acquiring control signals and determining periods of interest before conducting anomaly detection. During these predetermined periods when control signals change, the system suppresses anomaly detection to prevent erroneous detections caused by the conjunction of prediction values with measurement values. This preliminary identification and handling of critical periods resolves the contradiction by proactively eliminating the conditions that cause detection errors while preserving the anomaly detection capability during normal operation.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system applies local quality by differentiating the detection threshold based on the operational period. During periods of interest when control signals change, the system sets a higher threshold for anomaly detection compared to normal periods. This localized adjustment of detection sensitivity allows the system to maintain high anomaly detection capability during stable operation while reducing erroneous detections during control signal transitions, thus resolving the reliability-precision contradiction.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If the threshold for anomaly detection is lowered to reduce false negatives, then detection sensitivity is improved, but erroneous detections increase during control signal changes

Engineering Contradiction:
Improvedetection sensitivityVSAvoidfalse alarm rate
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The system implements dynamics by making the anomaly detection threshold variable rather than fixed. The threshold is dynamically adjusted based on whether the current time falls within a predetermined period of interest associated with control signal changes. During these periods, the threshold is raised to reduce false alarms, while during normal periods, the threshold remains lower to maintain high detection sensitivity. This dynamic adaptation resolves the contradiction between sensitivity and false alarm rate.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes the detection threshold parameter based on operational conditions. By monitoring control signal changes and identifying periods of interest, the system modifies the threshold parameter to be higher during control signal transitions and lower during normal operation. This parameter change strategy allows the system to achieve both high detection sensitivity and low false alarm rate under different operational conditions, resolving the technical contradiction.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4152115B1Monitoring apparatus, method, and program
Publication Date: 2024.12.18 KK TOSHIBA
  • EP4152115B1 patent drawingFigure 1
  • EP4152115B1 patent drawingFigure 2~3
  • EP4152115B1 patent drawingFigure 4(a)~4(d)

AI summary

According to one embodiment, a monitoring apparatus (100) includes a processing circuit. The processing circuit is configured to generate second data including a prediction value of a second sensor correlated with a first sensor from first data including a measurement value of the first sensor of which a measurement value changes suddenly in a case where the control signal changes, detect an anomaly of the system or an anomaly of at least one sensor, and make it difficult to detect the anomaly in a case where the determination signal indicates that there is a change in the control signal.