Time-Series Event Detection with 2D Shape Histograms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting events from time-series data sequences, particularly in vehicular contexts, face challenges in accurately identifying variable duration patterns and require subjective parameter settings, leading to inconsistent results.
Innovation Solution
A system and method utilizing a 2D shape histogram generation and concatenation technique, combined with a Support Vector Machine classifier, to analyze time-series data from multiple sensors, enabling accurate event detection by representing global distribution of sample points and employing a linear concatenation method to enhance feature discrimination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional pattern recognition techniques are used to detect events from time-series data, then the detection process is straightforward, but the accuracy is insufficient for variable duration patterns
Solution Approach 1:
The patent transforms 1D time-series data into 2D shape representations by plotting data points and connecting them to form visual patterns. This dimensional transformation enables the use of image processing techniques (like SIFT and HOG) that are more effective at recognizing variable duration patterns, thereby improving detection accuracy without proportionally increasing system complexity
Solution Approach 2:
The patent introduces shape histograms as an intermediary representation between raw time-series data and event classification. The shape histograms capture the global distribution of shape features, serving as a bridge that translates temporal patterns into a form suitable for machine learning classifiers, thus improving accuracy while managing complexity
2Reliability
If subjective parameter settings are used in existing detection methods, then the implementation is simple, but the results are inconsistent
Solution Approach 1:
The patent employs machine learning algorithms (SIFT, HOG, and support vector machines) that automatically learn optimal parameters from training data without requiring manual parameter tuning. The system self-adjusts to capture relevant features and distinguish events reliably, eliminating subjective parameter settings and ensuring consistent results across different scenarios
Solution Approach 2:
The patent transforms the detection problem from one requiring fixed subjective parameters to one where parameters are dynamically optimized through machine learning. The system learns optimal feature extraction parameters and classification thresholds from labeled training data, ensuring reliable and consistent detection performance across varying conditions
3Loss of information
If multiple sensors are used to capture time-series data, then the data coverage is improved, but the data processing complexity increases
Solution Approach 1:
The patent merges time-series data from multiple sensors by aligning them temporally and extracting features across all sensor streams simultaneously. The shape representation and histogram aggregation processes integrate information from multiple sources in a unified framework, maintaining complete data coverage while managing processing complexity through efficient feature aggregation
Data Source
Figure 1a~1b
Figure 2
Figure 3
AI summary
The present subject matter discloses a system and a method for detecting an event from time-series data sequences. The system receives time-series data sequences generated by sensors, wherein the time-series data sequences comprise sample points. The system pairs the sample points with one another for determining pairs of the sample points. The system computes Euclidean distances and angles between the sample points for determining distance matrix and angle matrix corresponding to the sample points. Further, the system determines global distribution of the plurality of pairs of sample points, wherein the global distribution of the plurality of pairs of sample points represent 2D shape histogram for the time-series data sequence. Further, the system concatenates the 2D shape histogram for each time-series data sequence to generate a concatenated shape histogram. Finally the system matches the concatenated shape histogram to pre-stored shape histograms for determining the event.