Extremum seeking controller and method for controlling a vapor compression system
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional extremum-seeking controllers for vapor compression systems suffer from slow convergence rates, making real-time optimization of performance metrics challenging due to the need for averaging perturbations and reliance on model-based assumptions that do not account for installation-specific variations and dynamic changes in system conditions.
Innovation Solution
The method employs a time-varying parameter estimation approach that updates the gradient of the performance metric recursively, eliminating the need for averaging and allowing faster convergence by tracking the true gradient, even in noisy environments, and can be applied to optimize multiple actuators simultaneously.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional extremum-seeking controllers use perturbation-based methods with averaging, then they can achieve optimal performance metrics, but the convergence rate is slow
Solution Approach 1:
The patent changes the fundamental parameter estimation approach from conventional averaging-based gradient estimation to a time-varying parameter estimation that directly tracks the true gradient. This is achieved by formulating the gradient as a time-varying parameter and using a recursive estimation algorithm that adapts to changing system conditions, thereby improving convergence rate while maintaining optimization accuracy.
Solution Approach 2:
The patent implements a feedback mechanism where the estimated gradient is continuously updated based on the difference between predicted and actual performance metric values. This feedback loop allows the controller to correct estimation errors in real-time, enabling faster convergence without sacrificing the precision needed to achieve optimal performance.
2Loss of energy
If model-based methods are used to predict optimal input combinations, then energy efficiency can be improved, but the models deviate from actual system behavior due to simplifying assumptions
Solution Approach 1:
The patent enables the system to self-adapt by continuously estimating time-varying parameters directly from operational data without relying on pre-built mathematical models. The extremum-seeking controller learns the system's actual behavior through recursive parameter estimation, allowing it to optimize energy consumption while accurately reflecting the true system dynamics, including installation-specific characteristics and variations.
Solution Approach 2:
The patent transitions from static model-based predictions to dynamic parameter estimation that adapts to changing system conditions. By treating the gradient as a time-varying parameter and using recursive estimation, the system can track changes in system behavior over time, maintaining reliability even as operating conditions, component characteristics, and environmental factors change.
3Ease of operation
If a single mathematical model is used for vapor compression systems, then model-based control can be implemented, but it cannot accurately describe variations among different manufactured copies
Solution Approach 1:
The patent enables each specific system instance to self-characterize through recursive parameter estimation. Instead of requiring a universal model that attempts to cover all manufacturing variations, the extremum-seeking controller learns the specific characteristics of each system through continuous adaptation, naturally accommodating manufacturing variations and installation-specific differences without requiring separate models.
4Loss of information
If conventional ESC controls the plant in a quasi-steady manner without exciting dynamic response, then phase information can be distinguished, but the convergence rate becomes slow
Solution Approach 1:
The patent embraces dynamic behavior rather than suppressing it. By formulating the gradient as a time-varying parameter and using recursive estimation, the method can handle transient responses and dynamic effects. This allows the system to converge faster by utilizing dynamic information while the feedback mechanism ensures that phase information remains distinguishable through proper signal processing and estimation techniques.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for controlling a vapor compression system determines a value of a metric of performance of the vapor compression system using a previous value of an estimated parameter and a previous value of a control signal determined for a previous time step of the control. The values of the estimated parameter represent a relationship between values of the control signal and values of the metric of performance. A current value of the estimated parameter is determined based on the previous value of the estimated parameter and an error between the determined value of the metric of performance and a measured value of the metric of performance. Next, a current value of the control signal is determined based on the current value of the estimated parameter.