Development and Testing of an Adaptive Fuzzy Logic Controller for HVAC Applications

Grant # 01-04
Principal Investigator: Ron M. Nelson
Organization: Iowa State University
Technical Area: Energy Efficiency

Background and Significance:
The purpose of this project is to improve the energy efficiency of commercial heating, ventilating, and air conditioning (HVAC) systems by performing research on adaptive fuzzy logic controllers. These controllers offer improved performance for systems, but research is needed to make them easier to use. The results of this project are expected to reduce energy and electric demand costs while improving comfort for HVAC applications.

Fuzzy logic controllers (FLCs) are based on a set of fuzzy control rules that make use of people’s common sense and experience. The fuzzy logic used in these controllers is very well defined mathematically and is able to take into account the uncertainty in the knowledge used to develop the rules and the uncertainty in the operating conditions. The initial rules for FLCs come from the experience of people who work with the systems that are to be controlled. This experience is obtained in the form of linguistic rules such as “If the temperature is too low, open the hot water valve” and “If the temperature is too low and going down fast, open the hot water valve very quickly”, etc. These initial rules are then used to construct a fuzzy control matrix that implements the logic of the controller.

The initial fuzzy control matrix and other FLC parameters need to be refined using adjustment strategies that are usually based on manual trial and error methods to achieve improved performance. Several researchers, including the principal investigator for this project, have studied and implemented self-tuning or adaptive fuzzy logic controllers (AFLCs) that improve their performance as they adjust to the controlled process and the environment. The operation of an AFLC relies on past experience that looks at suitable combinations of control strategies (control rules, membership functions, and scale factors) and the effects they produce. A particular feature of AFLCs is they automatically improve their performance until they converge to a predetermined optimal condition.
All previous work on AFLCs, except the work by the principal investigator, has been on industrial and robotic systems. More work is needed to develop AFLCs for HVAC applications that typically have long (about a minute or so) system time constants and periodically change operating modes (i.e. heating to cooling, using setback, etc.). Having a good adaptive or self-tuning algorithm for fuzzy logic controllers would increase their use since considerable time is needed to manually tune FLCs. Properly tuned FLCs are superior to conventional control algorithms [such as proportional-integral-derivative (PID)], because they respond more quickly to a set point and environmental changes and have minimal overshoot of the controlled variables. The controlled systems reach a steady state faster with reduced oscillations about the set point. The quick response improves comfort and, since the systems spend less time in transient operation, energy consumption is reduced. The control logic can also implement electric demand and energy saving measures.

Project Objectives:
The main objective of this project is to develop, implement, and test an adaptive fuzzy logic controller on one of the air-handling units at the Iowa Energy Center’s Energy Resource Station (ERS). A second objective is to develop procedures for dealing with missing and erroneous data.

The tasks for this research are to (1) conduct a detailed literature and technology review (completed), (2) develop and implement a fuzzy control algorithm in the ERS, (3) compare its performance to current control strategies, (4) develop procedures for dealing with missing and erroneous data, (5) develop and implement an adaptive FLC in the ERS and (6) report the results.