Research On Intelligent Habitat Pattern And Behavior Prediction Based On Data Mining Technology

Occupants' behavior is a key influence factor for building energy consumption,and it is also the main source of uncertainty in energy consumption simulation and prediction.The energy consumption behavior of occupants actually indicates their physical,physiological and psychological needs.At the present stage,how to balance the building energy conservation and human thermal comfort concerning about occupants is the emphases for building energy efficiecy research.With the rapid development of big data technology and the construction of cloud platforms,discovering the behavioral motives,patterns and regularities of residents becomes possible.To rely on the smart home management platform and use advanced data mining technologies,it can effectively overcome the limitations of behavioral research,solve the randomness and complexity of human behavior,guide people to energy conservation behavior from the hommization point of view,and lay the foundation for intelligent building automatioon system,which is of great significance both for the development of building energy efficiency and the improvement of human settlements.Based on the smart home management platform of residential buildings using a variety of data mining techniques and machine learning algorithms,this dissertation analyzes the motivations,patterns and regularities of each household's HVAC regulation behavior based on actual engineering cases.Firtly,a univeral framework for data preprocessing is presented,which applies KMeans,KNN,improved moving average and other data mining techniques to deal with different problems of missing data and abnormal data systematically in data platforms.The framework can realize data cleaning,data integration,data conversion and data specification effectively,so it is of great significance for improving data quality and ensuring the quality and efficiency of further data mining.Secondly,the influence factors of occupants' HVAC regulation behavior are put forward from multiple angles.Then the correlation between behavior and the factors is examined by binary logistic regression,nonparametric rank sum test and chi-square test.This dissertation further uses binary logistic regression and multiple logistic regression to excavate the key driving factors of occupant behavior,and the validity of the model was verified from the perspectives of model fitting goodness and fitting effect.In this way we can fully understand the motivation of occupant behavior.Thirdly,four behavior patterns,namely time condition pattern,outdoor environment pattern,indoor environment pattern and temperture setting pattern are proposed.This dissertation using K-Means cluster to identify the typical categories of occupant behavior patterns;and then using FP-growth association rule to analyze the cooling or heating setting refularities of each household under different outdoor environments and time conditions,as well as the indoor environment created by different outdoor environments and cooling or heating temperature setting levels,so that we can realize the regular activities,the thermal comfort preferences and the operation of building and system of each household.In this way,more practical strategies for energy consumption management and control will be put forward.Finally,three machine learning algorithms,namely random forest,BP neural network and support vector machine,are used to model the four types of occupants' HVAC regulation behaviors,and real-time behavior prediction is achieved.Then the models are verified and evaluated in order to select the best model that meets the need of real-time and accurate prediction of occupant behavior.The average calculation time of the best model is within 30 s.The average accuracy of the best model for heating switching,cooling switching and cooling temperature setting is above 95%,and the average absolute percentage error(MAPE)of the best model is only 1.00%.These occupants' behavior models lay the foundation for intelligent decision-making and personalized customization services of the smart home management and controling system.