Anomaly Detection Using Circumstantial Features for Multi-State Acoustic Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly detection technologies fail to effectively detect anomalies in acoustic signals generated by a generation mechanism with multiple states, as they do not account for changes in the state of the mechanism, leading to missed detections when signal patterns differ between states.
Innovation Solution
An anomaly detection apparatus and method that incorporates a pattern storage part, a circumstantial feature extraction part, an anomaly detection feature calculation part, and a score calculation part, using both acoustic signal patterns and circumstantial feature values from other modal signals to calculate an anomaly score, thereby accounting for state changes in the generation mechanism.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single generation mechanism model is used to detect anomalies in acoustic signals, then the device complexity is low, but the detection accuracy deteriorates when the mechanism has multiple states
Solution Approach 1:
The patent segments the single generation mechanism model into multiple state-specific models (first generation mechanism model and second generation mechanism model), each corresponding to a different operational state. This segmentation allows the system to accurately detect anomalies in each state independently while managing complexity through structured organization of the multiple models.
Solution Approach 2:
The patent implements a dynamic state identification mechanism that automatically determines which generation mechanism model to use based on the current operational state. This dynamic approach enables the system to adapt to changing conditions and select the appropriate model, thereby maintaining high detection accuracy across multiple states without requiring a static complex structure.
2Measurement precision
If multiple generation mechanism models are used to account for state changes, then the anomaly detection accuracy improves, but the device complexity increases
Solution Approach 1:
The patent applies partial action by using only the necessary subset of models based on the identified operational state. Instead of continuously managing all possible models, the system selectively activates and uses only the relevant generation mechanism model for the current state, reducing the effective complexity while maintaining detection accuracy.
Solution Approach 2:
The patent introduces a state identification mechanism as an intermediary that mediates between the multiple generation mechanism models and the anomaly detection process. This intermediary automatically determines the current state and selects the appropriate model, simplifying the management of multiple models and reducing the complexity burden on the overall system.
3Measurement precision
If state-specific modeling is implemented, then the detection of anomalies during state transitions improves, but the loss of time for state identification occurs
Solution Approach 1:
The patent performs preliminary action by pre-training and storing multiple generation mechanism models corresponding to different operational states before actual anomaly detection begins. This preliminary preparation allows the system to quickly identify and switch between states during operation without requiring time-consuming analysis, thereby maintaining high detection precision while minimizing time loss.
Solution Approach 2:
The patent replaces complex mechanical or manual state identification processes with a computational model-based approach. The state identification mechanism uses learned patterns from training data to automatically and rapidly determine the current operational state, substituting time-consuming traditional methods with efficient computational analysis that maintains precision without significant time penalty.
Data Source
AI summary
An anomaly detection apparatus extracts a circumstantial feature value for anomaly detection corresponding to a circumstantial feature value for learning from other modal signal for anomaly detection different in modal from acoustic, calculates a signal pattern feature related to an acoustic signal of anomaly detection target based on the acoustic signal of anomaly detection target, the circumstantial feature value for anomaly detection and a signal pattern model learned based on an acoustic signal for learning and the circumstantial feature value for learning calculated from other modal signal for learning, and calculates an anomaly score for performing an anomaly detection of the acoustic signal of anomaly detection target based on the signal pattern feature.