Real-Time Gas Turbine Engine Linearization for Control Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current model-based control systems for gas turbine engines rely on pre-calculated linearizations at steady state conditions, leading to inaccuracies and reduced control precision due to approximating engine states by a set of preselected steady states, resulting in diminished operating efficiency and increased risks of undesirable events like engine surge or blowout.
Innovation Solution
The system generates an operating point using a component-level model, analytically linearized by taking first partial derivatives of input parameters with respect to output parameters, forming a composite perturbational model that is inverted to solve for control commands as a function of target and measured parameters, providing a more accurate real-time control model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If pre-calculated linearizations at steady state conditions are used, then real-time implementation is achieved, but control precision and operating efficiency are reduced
Solution Approach 1:
The patent transforms the static pre-calculated linearization approach into a dynamic real-time linearization approach. The component-level model is analytically linearized at each time step based on current operating conditions, allowing the linearization to adapt dynamically to changing engine states rather than relying on fixed pre-selected steady states. This resolves the contradiction by enabling both real-time implementation and high control precision through dynamic adaptation.
Solution Approach 2:
The patent changes the parameter of linearization timing from fixed pre-calculated values to real-time dynamic values. By computing the linearization at each time step based on current operating parameters rather than using static pre-selected conditions, the system achieves both real-time capability and improved precision. The analytical linearization is performed using current operating point parameters, allowing the model to accurately represent the engine state at any given moment.
2Device complexity
If pre-selected steady states are used to approximate engine states, then computational simplicity is achieved, but operating efficiency is diminished
Solution Approach 1:
The patent segments the engine model into component-level models, where each component is analytically linearized independently. This segmentation allows for efficient computation while maintaining accuracy, as each component's linearization can be performed separately using current operating parameters. The segmented approach resolves the contradiction by providing computational simplicity through modular component linearization while achieving high operating efficiency through accurate real-time representation of each component's behavior.
3Measurement precision
If component-level models are analytically linearized in real-time, then control precision is improved, but computational complexity increases
Solution Approach 1:
The patent performs preliminary analytical linearization of the component-level model, deriving the linearization formulas in advance. This preliminary action allows the complex analytical linearization to be prepared beforehand, reducing the computational burden during real-time execution. The preliminary derivation of linearization relationships enables efficient real-time computation while maintaining high control precision, thus resolving the contradiction between precision and computational complexity.
Data Source
AI summary
A method for model-based control of a gas turbine engine is disclosed. An operating point of the gas turbine engine is generated from measured parameters using a component-level model. The component-level model is analytically linearized by taking the first partial derivative of output parameters of each component with respect to input parameters of each component, and evaluating the result at the operating point. Components of the linearized component-level model are combined to form a combined perturbational model of the gas turbine engine, which is inverted to solve for control commands as a function of target parameters and measured parameters.