Frequency-Response System Identification Under Variable Disturbances
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing system identification methods, such as those described in Patent Document 1, fail to accurately distinguish between model parameters that do not change and those that vary due to disturbances, leading to inaccurate controller design for systems with nonlinearity, as they estimate all parameters based on a single Bode diagram under a specific disturbance.
Innovation Solution
A system identification method that measures frequency responses under multiple disturbance magnitudes, calculates frequency responses for models with common and disturbance-variable parameters, and uses evaluation functions to search for parameter values that satisfy a predetermined condition, ensuring accurate identification of constant and varying parameters using nonlinear programming.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If all model parameters are estimated based on one Bode diagram under a specific disturbance, then the estimation process is simple, but the accuracy of distinguishing constant parameters from disturbance-variable parameters deteriorates
Solution Approach 1:
The patent segments the model parameters into two distinct groups: constant parameters that do not change with disturbance and disturbance-variable parameters that change with disturbance. This segmentation is achieved by performing parameter estimation under multiple disturbance conditions and identifying which parameters remain consistent across different disturbances versus those that vary. This resolves the contradiction by maintaining estimation simplicity through systematic classification while improving accuracy through multi-condition testing.
Solution Approach 2:
The patent applies dynamics by conducting parameter estimation under multiple varying disturbance conditions rather than a single static condition. By dynamically changing the disturbance magnitude and observing parameter behavior, the method can distinguish between constant and variable parameters. This dynamic approach improves identification accuracy while the systematic framework maintains process simplicity.
2Ease of operation
If model parameters are estimated without considering disturbance variations, then the estimation process is straightforward, but the controller responsiveness deteriorates
Solution Approach 1:
The patent performs preliminary parameter classification by estimating parameters under multiple disturbance conditions before controller design. This preliminary action identifies which parameters are constant and which are disturbance-variable, allowing the controller to be designed with appropriate compensation strategies. This resolves the contradiction by completing the classification work in advance, maintaining estimation simplicity while enabling responsive controller design that accounts for parameter variations.
Solution Approach 2:
The patent systematically changes disturbance parameters during the estimation phase to observe how model parameters respond. By deliberately varying disturbance conditions and tracking parameter changes, the method identifies disturbance-variable parameters that should be compensated in the controller. This approach maintains procedural simplicity through structured experimentation while improving controller responsiveness through informed parameter classification.
3Measurement precision
If multiple disturbance conditions are used for parameter estimation, then the accuracy of parameter identification improves, but the complexity of the estimation process increases
Solution Approach 1:
The patent reduces process complexity by segmenting parameters into constant and disturbance-variable groups based on their behavior under multiple disturbance conditions. This segmentation creates a systematic classification framework that organizes the estimation results, making the increased data from multiple disturbances manageable and interpretable. The structured approach maintains accuracy while reducing the perceived complexity through clear categorization.
Solution Approach 2:
The patent systematically varies disturbance parameters in a controlled manner to observe parameter responses. By deliberately changing disturbance conditions and tracking which model parameters change accordingly, the method creates a clear pattern recognition process. This systematic parameter changing approach improves identification accuracy while maintaining process manageability through structured experimentation and clear cause-effect relationships.
Data Source
AI summary
This system identification method includes: a step (S1) for measuring frequency responses (ω, HR1), (ω, HR2) . . . , and (ω, HRn) in a real system under n sets of disturbances of different magnitudes; a step (S3) for calculating frequency responses (ω, HM1), (ω, HM2) . . . , (ω, HMn) from input to output in n sets of mechanical models M1 to Mn including i sets (i is an integer of 1 or greater) of common parameters that do not change due to disturbance and j sets of disturbance variable parameters that do change due to disturbance; a step (S4) for calculating the values of a total of n sets of evaluation functions F (HRk, HMk) and the sum σF thereof, and steps (S3 to S6) for searching for the values of i sets of common parameters and j×n sets of disturbance variable parameters for which the sum σF would meet convergence conditions.