Adaptive Filter Mask for Heterogeneous Event Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing event detection systems struggle with accurately identifying heterogeneous events, such as varying step durations and amplitudes, and inconsistent intervals, due to their reliance on arbitrary threshold parameters and non-adaptive filtering methods.
Innovation Solution
The method and system employ localized adaptive filtering through a local time-frequency transform and an adaptive filter mask based on the time-frequency representation, allowing for real-time or batch processing of sensor data to detect events with varying characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a low-pass filter with a single cut-off frequency is applied to the entire data set, then the filtered data series shows discrete events more definitively, but events that vary in duration, amplitude, or separation time cannot be accurately detected
Solution Approach 1:
The patent segments the data set into multiple overlapping windows, applying a low-pass filter to each window independently with locally optimized cut-off frequencies. This segmentation allows the system to adapt to local variations in event characteristics while maintaining the benefits of filtering, thereby accurately detecting heterogeneous events with varying durations, amplitudes, and separation times.
Solution Approach 2:
The patent implements dynamic adaptation by adjusting the cut-off frequency of the low-pass filter based on local event characteristics detected in each data window. The system dynamically modifies filtering parameters to match the specific temporal and spectral properties of events in different regions of the data, enabling accurate detection of heterogeneous events that a static filter cannot handle.
2Ease of manufacture
If predetermined threshold parameters are used for event detection, then the system is simpler to implement, but it cannot accurately detect heterogeneous events with varying characteristics
Solution Approach 1:
The patent applies preliminary action by first segmenting the data into windows and analyzing local event characteristics before applying filtering and thresholding. This preliminary analysis allows the system to adapt thresholds and filter parameters to each local region, improving detection accuracy for heterogeneous events while maintaining a relatively simple overall system structure.
Solution Approach 2:
The patent implements parameter changes by adjusting the cut-off frequency of the low-pass filter and threshold values based on local event characteristics in each data window. This adaptive parameter modification enables the system to accurately detect events with varying durations, amplitudes, and separation times without requiring a completely complex system architecture.
3Productivity
If arbitrary threshold parameters are applied to sensor data, then the processing is faster and simpler, but events with varying amplitudes and durations are misidentified or missed
Solution Approach 1:
The patent segments the data into overlapping windows and applies localized filtering and thresholding to each segment. This segmentation enables parallel processing of multiple windows, maintaining high processing speed while allowing each window to use optimized parameters for its specific events, thereby improving event identification accuracy for heterogeneous data.
Solution Approach 2:
The patent dynamically changes threshold parameters and filter cut-off frequencies based on local event characteristics in each data window. This adaptive parameter adjustment maintains processing efficiency by avoiding global re-analysis while significantly improving event detection accuracy for events with varying amplitudes, durations, and separation times compared to fixed arbitrary thresholds.
Data Source
AI summary
A method and system for heterogeneous event detection. Sensor data is obtained and divided into discrete data windows. Each data window is defined by and corresponds to a time period of the sensor data. A time-frequency representation over the time period is calculated for each data window. A filter mask is calculated based on the data window corresponding to the time-frequency representation. The filter mask is applied for reverting the time-frequency representation to a time representation, resulting in filtered data. Features, such as extrema or other inflection points, are identified in the filtered data. The features define events, and transforming the time-frequency representation back into the time domain emphasizes differences between more and less prominent frequencies, facilitating identification of heterogeneous events. The method and system may be applied to body movements of people or animals, automaton movement, audio signals, light intensity, or any suitable time-dependent variable.