Research On Personalized Smart Home System Based On Machine Learning

In recent years,with the development of science technology and the improvement of living standards,people pay more and more attention to intelligence.As a hot topic in intelligentization,smart home plays an important role in improving the home environment.However,the current mainstream smart home products on the market have problems of incompatible hardware protocols and low intelligence.In order to solve the above problems,a personalized intelligent home system based on machine learning is designed.The system uses a home gateway based on open HAB(open home automation bus)for device management to achieve compatibility with different protocol devices.At the same time,the intelligent and personalized control of household equipment has been realized by the proposed system through mining and analyzing household data,which is also verified by the experiment data.The project firstly studied Zig Bee,Wi Fi wireless networking and open HAB(open home automation bus)platform,completed the networking of home sensors and home devices;then built a home embedded gateway platform to achieve platform access for home devices;finally designed mobile phones Client and web,to achieve the viewing of home parameters and control of home devices.Secondly,in order to further realize the personalized home system on the basis of intelligence,we analyze the characteristics of the home data,and then draw the conclusion that the data has regularity,randomness and difference.Therefore,the LSTM model with the attention mechanism is intentionally designed to predict the regular equipment state parameters;The GRU-In FCN model is designed to classify the random events;Considering the different characteristics of the data from different families,a update module is designed and employed in the two models mentioned above,so that the model can update the parameters according to the data collected from the different families.The intelligent and personalized control strategy can be achieved by combining the updated prediction model and the recognition model results.Finally,the test of the network monitoring function of the smart home system,including the system performance,the personalized control function of the home parameterprediction model and the intelligent control function of the user behavior recognition model are completed.The test results show that the smart home system realizes the compatibility of various communication protocol devices,and the system realizes the personalized control of the home devices on the basis of intelligence.