Research on Key Parameters Operation Range of Central Air Conditioning Based on Binary K-Means and Apriori Algorithm

Liangwen Yan,Wei Li,Fengfeng Qian

doi:10.3390/en12010102

Abstract

As the energy-saving control of central air conditioning has been widely applied in modern architecture, research of real-time optimal control based on historical data and identification of its optimal control strategies are of great importance for reducing energy wasting of buildings. However, due to the property of easily falling into local optimum, conventional k-means approach cannot achieve the goal of real-time optimal control, we therefore propose an innovative binary k-means clustering algorithm which is used to adjust the target value of temperature difference (TD) in the control system of chilled water and cooling water of central air conditioning system (CACS). Thanks to the clustering control, among the 304 test data, the coefficient of performance (COP) of 211 sets of data, which accounted for 69.41%, are higher than those of the traditional control method. In the simulation system, the COP of 191 sets of data, which accounted for 62.83%, are higher than those of traditional control methods, achieving better energy efficiency. To achieve the goal of identify potential energy-saving control strategies, the Apriori algorithm is proposed to correlate the key parameters and energy consumption efficiency of the CACS. The results show when the chilled water temperature difference (CWTD) > 2.0 °C and the cooling water temperature difference (COWTD) > 2.4 °C, some rules are discovered as follows: 1. The probability of a larger system COP will increase if the CWTD is set lower than the third quartile value or the COWTD is set lower than the first quartile value. 2. The probability of a larger system COP will also increase if the CTWD is set lower than the first quartile and the COWTD is set between the first and the third quartile. These underlying regularity is useful for technicians to adjust the control parameters of the equipment, to improve energy efficiency and to reduce energy consumption.

Highlights

In recent years, with the popularization of central air conditioning in modern buildings, energy-saving control of central air conditioning has been the focus of research
The temperature difference (TD) of the chilled water and the TD of the cooling water were the control parameter that needed to be obtained through cluster analysis, they were considered as target indicators and illustrated in Figures 6 and 7
Data mining research on data of central air conditioning system (CACS) is demonstrated by clustering and association in this paper

Summary

Introduction

With the popularization of central air conditioning in modern buildings, energy-saving control of central air conditioning has been the focus of research. Studies have pointed out that using advanced control technology can save more than 20% of energy output [1]. A large amount of central air conditioning data has been stored in the CACS. The large amount of data stored in the CACS cannot be utilized with traditional data analysis methods because of the variety of data structures in their subsystems [2]. DM is believed to be a powerful tool to learn the underlying regularity in large dataset [3]. Energy-saving methods can be found [4] through the advanced DM methods on the operational data of the CACS [5].

Methods

Discussion

Conclusion