Algorithm Design And Implementation Of Air Conditioning Refrigerant Charge Amount Anomaly Detection

Air conditioning in our daily life has been widely used,with the air conditioning running,it may break down. Refrigeration system is the core part of the air-conditioning, in the refrigeration system, the fault of refrigerant charge is one of the most important and typical fault, and is of high research value. It is of great importance for improving the air conditioning operational efficiency and service life, saving resources to detect the air conditioner refrigerant fault scientifically and accurately. With the advances in computer technology, data mining technology used to improve the air conditioning unit operating safety and efficiency is an effective solution ideas.This paper describes the research status and the related technologies of fault diagnosis, and studies the related machine learning algorithm and its improvement.At the same time we do research on how to use machine learning algorithms to complete the detection of air conditioning refrigerant charge failure.This paper studies the content of the decision tree algorithm, introduces several common decision tree algorithms include ID3 algorithm, C4.5 algorithm, CART algorithm. Aiming at the shortcomings of the CART algorithm when dealing with continuous attributes segmentation, the paper studies some related improvements. Then we study the SVM algorithm, and on the basis of studying svm,we learn to use genetic algorithm to optimize penalty factor C and kernel function parameter sigma. To complete the detection of air conditioning refrigerant charge failure, espectively,we use the improved decision tree algorithm and GA-SVM algorithm to establish classification models,and draw relevant conclusions by comparing the two classification models.