Design And Research Of Intelligent Home Environment Monitoring System Based On LSTM

In order to adapt to the development of emerging science and technology trends,meet the needs of future smart cities,improve the intelligent level of human living environment,the development of smart home in the society is playing a very important role.This article through to the present situation of smart home to do some analysis and research,the artificial intelligence emerging technologies combined with the traditional smart home control system,intelligent household environment data acquisition,data upload,user interaction,model training,such as home appliances control set in one,designed an improved monitoring and control system of smart home environment.Through the analysis of requirements,the overall structure of the system is planned.The user interaction of B/S architecture is adopted.The intelligent monitoring of the smart home environment is completed by collecting the overall structure of the lower computer--server--control node--household appliances,and the main software and hardware are studied and designed.In full to the artificial intelligence technology were studied after the machine learning theory,this paper chooses the length of the memory(LSTM)neural network model as the basis of smart home,after a brief analysis of GOOGLE’s open source framework,this paper chooses TensorFlow as LSTM neural network deployment platform,and studied the prediction model is designed.The environmental monitoring system based on LSTM neural network is mainly divided into three modules for research and design.The data management module collects environmental information through the raspberry pie of the lower computer,and displays the information to the server web page and mobile terminal interface of B/S architecture in real time through communication.At the same time,the data is stored in SQLite3 database,and the prediction model is provided for prediction.The network communication module mainly consists of raspberry pie and server uploading and storing data through MQTT protocol.The server communicates with the control center node through WIFI;Finally,the control center node sends control commands to different household appliances through IR,RF,EnOcean and other technologies.The prediction module mainly includes the LSTM neural network model based on TensorFlow.After data input,the model predicts and outputs control instructions.Finally,the intelligent home environment monitoring system designed in this paper was tested.The integrity of environmental information collection reached 100%,and the control rate of household appliances reached 95%.The wireless network test includes that the network can recover itself within 2 minutes after power failure.As the height changes,the communication quality is maintained below-95 db.The test of the prediction model mainly predicts the environmental information of one day,and the accuracy of the output householdappliance control instruction reaches 90%.Through testing,the intelligent home environment monitoring system designed in this paper meets the requirements as a whole.