Applications of Supervised-Learning Approaches for Air Conditioning Plants Condition-Based Maintenance

Abstract

Numerous data analytic techniques have been employed by the United States Navy for air conditioning (AC) plant condition-based maintenance. The developed diagnosis models include physics-based and data-driven approaches, particularly regression models have shown that AC plants operating conditions could be correctly diagnosed for poor conditions or faults. However, building effective condition-based maintenance models for AC plants faults prognosis based on classification has not been explored in-depth. This paper presents a comparative study of machine learning classification techniques that produces solutions that are as effective and better than solutions produced by the physics-based and regression approaches.