Application And Research Of Smart Home Decision Based On Activity Recognition

With the emergence and application of the technology of Internet of things and machine learning,the phenomenon of smart homes has witnessed a rapid development.In order to provide personalized services to different users,smart homes employed the technology of Internet of things to connect various household devices in the home,thus allowing remote lighting and air conditioner temperature control.Knowledge base is used to describe the environment information perceived by wireless sensors,followed by uncertain reasoning and information decision-making.Finally,concrete operation commands are produced and fed back to the controller to perform the corresponding operations.This paper mainly covers the following three parts:A smart home application framework is proposed.By using a hierarchical knowledge model,different modules can reason and communicate by means of the same concepts and relationships.A hybrid reasoning method is proposed(ER-OT),which combines the improved evidential theory with the ontology reasoning using Jena reasoning,and is applied to activity recognition to solve the uncertainty of activities and conflicts between reasoning results.Experiments are conducted to show that the uncertainty reasoning method designed in this paper achieves higher degree of reasoning accuracy.The GA-SVM algorithm is improved and applied to decision making of different users in different situations after intelligent space activity recognition,which provides personalized service for different users.By using the data of home environment as the input to the training classification model and using KNN classification algorithm to match user groups,making the home environment truly different from person to person,and at the same time enhanced the degree of intelligence of smart home.