Test Bench Dynamic Parameter Identification via Frequency Response
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying system parameters of a test setup on a test bench are limited by the complexity of the setup, require assumptions about model structure, and are affected by measurement noise, leading to inaccurate results and difficulty in adapting to changes in mechanical structure or aging components.
Innovation Solution
A method that uses a non-parametric identification to determine the frequency response of the test setup, allowing for a systematic identification of system parameters, which are then used to derive a suitable model structure for parametric identification, accounting for measurement noise and enabling improved accuracy and adaptability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If parametric identification methods are used to determine system parameters, then the identification process requires assumptions about model structure, but this leads to inaccurate results when the assumed model structure does not match the actual system
Solution Approach 1:
The patent inverts the traditional identification approach by first performing non-parametric identification to determine the actual frequency response of the system, then using this empirical data to inform the parametric model structure. This reverses the conventional workflow where parametric methods assume a model structure first, thereby avoiding the inaccuracy caused by mismatched assumptions.
Solution Approach 2:
The patent performs non-parametric identification as a preliminary step before parametric identification. This preliminary action captures the actual system behavior through frequency response measurement, which then serves as the foundation for selecting an appropriate parametric model structure, ensuring subsequent parametric identification is based on accurate empirical data rather than potentially incorrect assumptions.
2Measurement precision
If traditional identification methods are used, then the process is simpler, but measurement noise significantly degrades the identification quality
Solution Approach 1:
The patent introduces non-parametric frequency response analysis as an intermediary step between raw noisy measurements and final parametric model identification. This intermediary process filters and processes the measurement data to extract the frequency response characteristics, which are then used to guide parametric identification, thereby mediating the impact of measurement noise on final identification quality.
3Adaptability or versatility
If the mechanical structure of the test setup is adapted or components are replaced, then the test setup can be improved or updated, but the dynamic system parameters change, requiring re-identification
Solution Approach 1:
The patent employs a two-stage identification methodology that is dynamically adaptable to structural changes. The non-parametric frequency response measurement can be quickly performed to detect changes in system behavior, and only when significant changes are detected does the more time-consuming parametric identification need to be repeated, allowing the system to adapt efficiently to structural modifications.
4Measurement precision
If non-parametric identification is used to determine frequency response, then measurement noise is accounted for, but additional processing steps are required
Solution Approach 1:
The patent segments the identification process into two distinct stages: non-parametric frequency response determination and parametric model identification. This segmentation allows each stage to focus on its specific strength - the non-parametric stage handles noise robustly through frequency response analysis, while the parametric stage leverages this cleaned data for accurate model parameter extraction, reducing the complexity burden on any single stage.
Data Source
Figure 1~6
Figure 2a~2b
Figure 3~5
AI summary
In order to improve the identification of system parameters of a test setup of a test bench, in particular in terms of the quality of the identification, there is provision for the test setup (PA) to be dynamically stimulated on the test bench (1) by virtue of a dynamic input signal (u(t)) being connected to the test setup (PA) and, in the process, measured values (MW) of the input signal (u(t)) of the test setup (PA) and of a resultant output signal (y(t)) of the test setup (PA) being recorded, a frequency response (G(Ωk)) of the dynamic response of the test setup (PA) between the output signal (y(t)) and the input signal (u(t)) being ascertained from the recorded input signal (u(t)) and output signal (y(t)) using a nonparametric identification method, a model structure of a parametric model that maps the input signal (u(t)) onto the output signal (y(t)) being derived from the frequency response (G(Ωk)), the model structure and a parametric identification method being used to ascertain at least one system parameter (SP) of the test setup (PA), and the at least one identified system parameter (SP) being used to perform the test run.