Signal Processor Zone Segmentation for Measurement Data Filtration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing signal processing methods for filtering measurement data struggle to accurately remove noise while preserving the features of the data, particularly in cases with low noise levels or abrupt changes, leading to distortions and loss of accurate surface representation.
Innovation Solution
A signal processor that divides digital signal values into zones and calculates a weighting factor based on the median of the sum of squares of residual errors for each zone, allowing for accurate filtration by updating the weighting factor independently for each zone, thereby avoiding the issues of abnormal data treatment and distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single median weighting factor is applied to all measurement data, then the filtering process is simple, but local fluctuations of the data cannot be grasped and distortions are generated at both ends of the measurement area
Solution Approach 1:
The measurement data is divided into multiple zones along the measurement route, and a separate median weighting factor is calculated for each zone. This segmentation allows each zone to have its own optimized weighting factor, capturing local fluctuations while maintaining overall filtering effectiveness.
Solution Approach 2:
Instead of applying a uniform weighting factor across all data, the invention calculates and applies different weighting factors for different zones. Each zone's weighting factor is derived from the median of residual errors specific to that zone, allowing local characteristics to be preserved in the filtered output.
2Reliability
If the robust Gaussian regression filter updates weighting factors based on median residual error, then abnormal data is handled robustly, but data with very small noise levels are treated as abnormal and weight becomes zero
Solution Approach 1:
The data is segmented into zones, and the median residual error is calculated separately for each zone. This prevents the global median from being skewed by local characteristics, ensuring that zones with small noise levels maintain appropriate weights rather than being incorrectly identified as abnormal.
Solution Approach 2:
Each zone receives a weighting factor based on its local residual error characteristics. This local adaptation ensures that zones with genuinely small noise levels retain non-zero weights, preserving their data representation accuracy while still robustly handling abnormal data within each zone.
3Object-affected harmful factors
If initial processing for filtration is applied to remove noise, then high frequency factors are removed, but distortions are generated at both ends of the measurement area
Solution Approach 1:
The measurement data is divided into multiple zones along the measurement route, with the first zone starting from the beginning and the second zone starting from a predetermined position. This segmentation allows different filtering strategies for different regions, reducing edge distortions while maintaining noise removal effectiveness.
Solution Approach 2:
Different weighting factors are applied to different zones based on their local characteristics. The first zone uses a weighting factor based on its local residual error, while the second zone uses a different weighting factor, allowing each region to be filtered optimally for its specific characteristics and reducing overall distortion.
Data Source
AI summary
A digital signal value is divided to a plurality of zones along a route. A median is computed based on a difference between a digital signal value and a sum of squares of each component of a filter output value for the digital signal values for each zone. A weighting factor for digital signals in each zone is computed and updated based the median for the zone. A filter output value for the digital signal value is obtained by executing filtration using the computed weighting factor.