The applicability of machine learning techniques for feedback control systems is limited by a lack of stability guarantees. Robust control theory offers a framework for analyzing the stability of feedback control loops, but for the integral quadratic constraint (IQC) framework used here, all components are required to be represented as linear, time-invariant systems plus uncertainties with, for IQCs used here, bounded gain. In this paper, the stability of a control loop including a recurrent neural network (NN) is analyzed by replacing the nonlinear and time-varying components of the NN with IQCs on their gain. As a result, a range of the NN's weights is found within which stability is guaranteed. An algorithm is demonstrated for training the recurrent NN using reinforcement learning and guaranteeing stability while learning.
While having the potential to significantly improve heating, ventilating and air conditioning (HVAC) system performance, advanced (e.g., optimal, robust and various forms of adaptive) controllers have yet to be incorporated into commercial systems.Controllers consisting of distributed proportional-integral (PI) control loops continue to dominate commercial HVAC systems. Investigation into advanced HVAC controllers has largely been limited to proposals and simulations, with few controllers being tested on physical systems. While simulation can be insightful, the only true means for verifying the performance provided by HVAC controllers is by Preprint submitted to Energy and Buildings 4 January 2006 actually using them to control an HVAC system. The construction and modeling of an experimental system for testing advanced HVAC controllers, is the focus of this article.A simple HVAC system, intended for controlling the temperature and flow rate of the discharge air, was built using standard components. While only a portion of an overall HVAC system, it is representative of a typical hot water to air heating system. In this article, a single integrated environment is created that is used for data acquisition, controller design, simulation, and closed loop controller implementation and testing. This environment provides the power and flexibility needed for rapid prototyping of various controllers and control design methodologies.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.