Controller Setting Selection Using Spectral Entropy for Oscillation Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Complex technical systems like gas turbines, wind turbines, and power grids often experience oscillations due to complex dynamical interactions, which impair efficiency and increase wear, and current machine learning-based controllers struggle to effectively detect and reduce these oscillations.
Innovation Solution
A method that processes operational data signals into the frequency domain to determine entropy values, allowing for the selection of optimal controller settings from a plurality of options, which can reduce oscillations by identifying and amplifying periodic components and using machine learning methods like neural networks to configure the system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If machine learning methods are used to model complex dynamical behavior and provide control policies, then the control efficiency is improved, but complex oscillation patterns are induced that are difficult to detect and reduce
Solution Approach 1:
The patent introduces an intermediary evaluation process that uses signal processing and entropy calculation to assess control policies without directly modifying the machine learning model. This intermediary layer detects oscillations in operational signals and provides feedback for selecting control policies with lower oscillation complexity, thus resolving the contradiction between control efficiency and oscillation generation
Solution Approach 2:
The patent implements a feedback mechanism where operational signals are analyzed to determine oscillation characteristics, and this information feeds back into the selection of controller settings. The entropy-based evaluation provides quantitative feedback that guides the selection of control policies that minimize oscillation complexity while maintaining control efficiency
2Reliability
If a plurality of different controller settings are evaluated to reduce oscillations, then the ability to detect and reduce oscillations is improved, but the assessment time and computational complexity increase
Solution Approach 1:
The patent replaces complex time-domain analysis with frequency-domain analysis using entropy calculation. This substitution transforms the oscillation assessment from a computationally intensive time-series evaluation to a more efficient spectral entropy computation, enabling rapid evaluation of multiple controller settings while maintaining high detection reliability
Solution Approach 2:
The patent changes the assessment parameter from time-domain signal analysis to frequency-domain entropy measurement. By transforming operational signals into the frequency domain and calculating spectral entropy, the system achieves a compact representation of oscillation characteristics that can be computed rapidly, allowing evaluation of hundreds or thousands of controller settings in acceptable time
Data Source
AI summary
For reducing oscillations in a technical system plurality of different controller settings for the technical system is received. For a respective controller setting signal representing a time series of operational data of the technical system controlled by the respective controller setting is received, the signal is processed, whereby the processing includes a transformation into a frequency domain, and an entropy value of the processed signal is determined. Depending on the determined entropy values a controller setting from the plurality of controller settings is selected, and the selected controller setting is output for configuring the technical system.