شبکه فازی بیزی برای پیش بینی عملکرد کمپرسور تبرید و کاهش زمان آزمون

A typical characteristic of refrigeration compressor performance tests is their long duration. A reduction in the time periods related to this activity can be achieved using unsteady-state data analysis. This paper presents an original approach to predicting compressor performance using Bayesian networks and a hybrid Fuzzy–Bayesian network. All analysis was performed using real test data.

Performance tests are an experimental activity which aims to measure four fundamental characteristics of a compressor related to its performance: refrigerating capacity; power consumption; isentropic efficiency and coefficient of performance (COP). This kind of test is mainly used for three purposes: research and development (R&D), determination of catalog parameters, and quality control (ASHRAE, 2005 and ISO, 1989). A typical characteristic of these tests is the long time required to achieve stable conditions, under which the measurement can be performed. In general, performance tests last 2½ h. Most performance tests run by a typical compressor manufacturer are carried out for R&D purposes, to develop new technologies and improvement the current compressors (Borges, 2008). Additionally, improvements in the compressor performance must be statistically proven using a sufficient number of test results. To achieve shorter test times several studies have been carried out related to: optimization of test benches; use of measurement systems with lower uncertainties; and application of automation and control techniques to improve the performance tests. Nevertheless, the reduction achieved with these actions was insufficient, since unsteady periods of long duration are typical in refrigeration systems (Flesch and Normey-Rico, 2010, Poleto, 2006 and Scussel, 2006). An alternative and original approach to reducing the time required for refrigeration compressor performance tests is proposed in this paper. The technique proposed is based on unsteady-state data analysis using Fuzzy–Bayesian networks to predict compressor performance. This paper presents a Fuzzy–Bayesian network structure which can be used to achieve a significant time reduction of one third of the complete test time, on average, in the refrigerating capacity evaluation. The article is divided into the following sections: Section 2 describes the performance test; Section 3 briefly presents the Bayesian networks and Fuzzy–Bayesian networks; Section 4 presents the proposed networks; and Section 5 presents the conclusions.

This article presented an original approach to classifying and predicting the refrigerating capacity of compressors during their performance test. The prediction allows a significant reduction in time costs and can be used to improve the quality control and R&D performed by the compressor manufacturer. The application of Fuzzy–Bayesian networks could be extended to other performance variables like the power consumption, isentropic efficiency and coefficient of performance (COP).