Multidimensional Signal Separation With Repetition-Time Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for signal processing in multichannel measurements, such as arithmetic mean calculation followed by factor analysis or blind signal separation, fail to accurately estimate the true signal value and reduce noise effectively, leading to distorted waveforms and loss of asynchronous components.
Innovation Solution
An information processing device and method that utilizes a calculation model with a constraint for repetitive signal parts to have the same value at identical relative times from a repetition start, enabling effective signal separation and analysis using multidimensionality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If arithmetic mean calculation is performed after factor analysis or blind signal separation, then noise reduction is achieved, but the estimated signal value does not approach the true value and waveform distortion occurs
Solution Approach 1:
The patent inverts the conventional processing sequence by performing factor analysis or blind signal separation first, then applying arithmetic mean calculation to the separated components. This reversal allows the signal separation to occur on the original multidimensional data before noise accumulation, followed by noise reduction through averaging of the separated signal estimates, thereby improving both signal estimation accuracy and waveform reliability
Solution Approach 2:
The patent segments the signal processing into distinct stages: first separating the target signal from noise and other components using factor analysis or blind signal separation, then independently calculating arithmetic means of the separated signal components. This segmentation allows each processing stage to optimize for its specific function, preventing waveform distortion while achieving noise reduction
2Object-affected harmful factors
If factor analysis or blind signal separation is performed after arithmetic mean calculation, then noise samples are reduced, but the number of asynchronous components is also reduced making target signal separation difficult
Solution Approach 1:
The patent inverts the processing order to perform factor analysis or blind signal separation before arithmetic mean calculation. This ensures that asynchronous components are preserved during the separation stage, and only the noise portions are reduced through subsequent averaging, preventing loss of target signal information
Solution Approach 2:
The patent extracts the target signal components from the mixed signal using factor analysis or blind signal separation before applying noise reduction. By taking out the signal components of interest first, the subsequent arithmetic mean calculation only affects noise levels without removing asynchronous signal components
3Device complexity
If conventional separate processing methods are used, then processing simplicity is maintained, but effective utilization of multidimensionality and repetitive structure is insufficient
Solution Approach 1:
The patent merges factor analysis or blind signal separation with arithmetic mean calculation into a unified processing framework where the separated signal components are directly subjected to averaging. This combination allows the method to simultaneously exploit multidimensional correlations and repetitive structures, improving signal analysis effectiveness while maintaining processing simplicity through the integrated approach
Data Source
Figure 1
Figure 2
Figure 3
AI summary
According to an embodiment, an information processing device includes an acquisition unit configured to acquire a multidimensional measurement signal obtained by measuring a target signal including a signal component having a repetitive structure and a signal processing unit configured to perform predetermined signal processing on the measurement signal acquired by the acquisition unit with a calculation model for performing signal separation using multidimensionality, the calculation model including a constraint that repetitive parts have the same value when a relative time from a repetition start time is identical.