Abnormal Sensor Estimation via Outlierness Degree
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for detecting abnormal sensors in complex systems, such as automobiles or manufacturing apparatuses, struggle with handling non-linear relationships among sensors, often misclassifying normal data as abnormal or vice versa due to their reliance on linear models.
Innovation Solution
An apparatus and method that acquire target data from multiple sensors, calculate the degree of outlierness relative to a reference data distribution, and estimate which sensors are sources of outlierness by generating reference data distributions compatible with the relational structure among sensors, allowing for the handling of both linear and non-linear relationships.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a linear model is used to analyze sensor relationships, then the processing method is simple, but it cannot accurately handle complex non-linear relationships among sensors
Solution Approach 1:
The patent transforms the sensor data by applying non-linear transformation functions to the original sensor readings. This parameter transformation allows the system to capture complex non-linear relationships among sensors while maintaining a computationally efficient processing framework. The transformed parameters are then used in the analysis to accurately identify abnormal sensors without requiring overly complex processing methods.
2Ease of operation
If conventional linear analysis is used, then processing is straightforward, but normal data may be misclassified as abnormal or abnormal data as normal
Solution Approach 1:
The patent introduces a new dimension to the analysis by computing both the degree of outlierness and the degree of association between sensor groups. This multi-dimensional approach allows for more accurate classification of sensor data by considering multiple aspects simultaneously - how much individual sensors deviate from normal patterns and how sensor groups are interconnected. This resolves the misclassification problem by providing a more nuanced view of sensor behavior.
Solution Approach 2:
The patent introduces sensor groups as intermediary entities between individual sensors and the final abnormality determination. By analyzing groups of sensors together and computing their collective degree of outlierness, the system can better distinguish between random fluctuations and genuine abnormalities. The degree of association metric acts as an intermediary that captures the relational structure among sensors, improving classification accuracy while maintaining operational simplicity.
Data Source
AI summary
Provided is an estimation apparatus including a memory device having program code stored thereon, and at least one processor operatively coupled to the memory device. The at least one processor is operable to execute program code stored on the memory device to calculate, for each of a plurality of sensor groups that each include two sensors among a plurality of sensors, a degree of outlierness of target data serving as an examination target relative to a reference data distribution of output from the sensor group, and to estimate a sensor among the plurality of sensors to be a source of outlierness, based on a comparison of the degrees of outlierness of the sensor groups that include the at least one sensor to the degrees of outlierness of the sensor groups that lack the at least one sensor.