Adaptive Frequency Response Sampling for Speed-Accuracy Tradeoffs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional sinusoidal excitation-based frequency response measurement methods face challenges with measurement speed, accuracy, and data inheritance, particularly in complex systems, due to issues with step size selection and inefficient use of existing sampling information.
Innovation Solution
A two-point sampling optimized method that iteratively adds new sampling points to sub-frequency bands with the largest interpolation error, using the trapezoidal rule to estimate errors and strategically place new points for optimal sampling, ensuring each point is globally optimal based on existing information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the step size of frequency sweep measurement is chosen too large, then the measurement speed is improved, but the measurement accuracy deteriorates
Solution Approach 1:
The patent applies local quality by making the sampling density adaptive rather than uniform. The algorithm calculates interpolation errors for different sub-frequency bands and concentrates sampling points in regions with large interpolation errors while using fewer points in regions with small errors. This resolves the contradiction by locally increasing sampling density where accuracy is needed and reducing it where accuracy is already sufficient, thereby improving overall measurement accuracy without sacrificing measurement speed.
Solution Approach 2:
The patent implements dynamics by making the sampling strategy adaptive and iterative. Instead of a fixed step size, the algorithm dynamically adjusts the distribution of sampling points based on real-time calculation of interpolation errors. The process repeatedly identifies sub-frequency bands with largest interpolation errors and adds sampling points there, creating a dynamic sampling strategy that optimizes both speed and accuracy throughout the measurement process.
2Measurement precision
If the step size of frequency sweep measurement is chosen too small, then the measurement accuracy is improved, but the measurement time is greatly prolonged
Solution Approach 1:
The patent resolves this contradiction by applying local quality through adaptive sampling density. Rather than uniformly small step sizes across the entire frequency range, the algorithm calculates interpolation errors and concentrates sampling efforts only in sub-frequency bands where errors are large. This ensures measurement accuracy in critical regions while avoiding unnecessary measurements in regions where accuracy is already sufficient, thereby significantly reducing overall measurement time.
Solution Approach 2:
The patent applies partial action by performing sampling only where necessary. Instead of uniformly sampling across the entire frequency range with small step sizes, the algorithm identifies and focuses sampling efforts on specific sub-frequency bands with largest interpolation errors. This partial sampling strategy achieves the required measurement accuracy in critical regions without the time cost of exhaustive uniform sampling across the entire range.
3Ease of operation
If conventional frequency sweep method is used, then the measurement process is simple, but the data inheritance is not good
Solution Approach 1:
The patent applies feedback by using the results of interpolation error calculations to guide subsequent sampling decisions. The algorithm calculates interpolation errors based on existing sampling points, uses this feedback information to identify sub-frequency bands with largest errors, and then adds sampling points there. This feedback loop ensures that each sampling decision is informed by previous measurements, maximizing data inheritance and efficiency while maintaining a relatively simple automated process.
Data Source
AI summary
The present disclosure discloses a two-point sampling optimized method and system for sinusoidal excitation-based frequency response measurement. A plurality of points are sampled at equal intervals within the starting frequency and ending frequency range as initial information, the interpolation error of each sub-frequency band is estimated according to existing sampling information, the sub-frequency band with the largest interpolation error is selected, two new sampling points are added within the sub-frequency band, and the above steps are repeated until the quantity of sampling points reaches the total quantity set by a user; and the user is asked whether new sampling points need to be added, if so, after the user specifies a new quantity of sampling points, the interpolation error of each sub-frequency band is estimated again and sampling continues, otherwise, the sampling ends. The present improves the practicality of a sinusoidal excitation-based frequency response measurement method.