Fuzzy Inference Procedure for Intelligent and Automated Control of Refrigerant Charging

摘要

Fuzzy logic controllers are readily customizable in natural language terms and can effectively deal with nonlinearities and uncertainties in control systems. This paper presents an intelligent and automated fuzzy control procedure for the refrigerant charging of refrigerators. The elements that affect the experimental charging and the optimization of the performance of refrigerators are fuzzified and used in an inference model. The objective is to represent the intelligent behavior of a human tester and ultimately make the developed model available for the use in an automated data acquisition, monitoring, and decisionmaking system. The proposed system is capable of determining the needed amount of refrigerant in the shortest possible time. The system automates the refrigerant charging and performance testing of parallel units. The system is built using data acquisition systems from National Instruments and programmed under LabVTEW. The developed fuzzy models, and their testing results, are evaluated according to their compatibility with the principles that govern the intelligent behavior of human experts when performing the refrigerant-charging process. In addition, comparisons of the fuzzy models with classical inference models are presented. The obtained results confirm that the proposed fuzzy controllers outperform traditional crisp controllers and provide major test time and energy savings. The paper includes thorough discussions, analysis, and evaluation.