Design Of Smart Home Assistance System Based On Machine Learning

In recent years,social progress has stimulated people’s demand for higher quality home life,and smart home with family users as its goal has emerged.Today,most of the smart home products on the market have been built into high-end products of the family.It is more thinking about how to make the intelligent home control more intelligent,so it is relatively neglecting the cost of family life,but the family users are enjoying the convenience brought by the intelligence,and they also want to reduce the family life as much as possible.Therefore,the effective balance between intelligent control and reducing household energy consumption in smart home system is a hot topic at present.How can the smart home provide the convenient and comfortable service to the user,analyze the basic life scene in the home life,the data produced by the daily life of the user,after a certain time accumulation,can form a huge basic data set,which contains a lot of valuable information,but usually the letter of these data sets The interest value cannot be obtained intuitively,and the distribution is sparse.How to use these data effectively and discover the effective information to provide intelligent life service for the users is always the goal that the researchers strive to pursue.At present,the machine learning technology is applied to the smart home.Through the information mining of the user data,it can improve the initiative of the home equipment control to a certain extent and make the intelligent home system service more automatic and intelligent.In this paper,a neural network model,Res-BPNN,which is based on the improvement of residual learning,is applied to the intelligent home system.This model can solve the deep effective signal attenuation by the original network reconfiguration and design the outburst and burst problem of the neural network ladder by the identity mapping.The network has more hidden layers under the condition of hidden layer nodes.Compared with the typical BPNN,it has stronger abstraction ability and generalization ability.This paper designs an auxiliary system for smart home,which is divided into auxiliary control subsystem and auxiliary scheduling subsystem from function.The auxiliary control subsystem records the user’s operation information data and records the home environment information through the sensor.The system machine learning unit will learn the two aspects of the data,then predict the user’s behavior,do some behavior feedback to the home system,and then help users to do some operating parties to the intelligent home system.The auxiliary control of the surface.In order to reduce the cost of household energy consumption,the auxiliary scheduling subsystem proposes a device scheduling algorithm based on the relaxation prediction.This algorithm is applied to the time-sharing charging mechanism,introduces the concept of relaxation degree,measures the latest scheduling time of home equipment,and predicts the relaxation degree of home equipment as a device adjustment.The basis is to reduce the power consumption at peak hours and smooth the energy consumption curve.The experiment proves that the auxiliary system designed in this paper has a high accuracy in the intelligent control of the equipment and has a certain role in reducing the family expenditure.