Intelligent Monitoring System For Home Based On Multisensor Information Fusion

To create a safe and reliable home residential environment has been one of the most important hotspot. Traditional home monitor system is usually unit with single-family. When an emergency occurs, the system would send message to the home owner or dial the phone that was set by the owner. However, because the distance between house and relative staff is uncertain, measures may not be taken in time. At the same time, the low sensitivity and reliability of systems often failed to alarm which wasted a lot of manpower and resource. To develop a new type of high sensitivity and reliability intelligent monitoring system based on residential units has important significance.A new fuzzy neural network(FRBF) is put forward as the classifier to collect data of the field. In order to prove the theory the fire detection is analyzed, Multi-sensor technology is used to improve the alarm algorithm of the system, the time window is used to extract the necessary information. Principal component analysis(PCA) is used to analyze the data and extract the feature as the input of neural network. The system is also can be used in illegal envasion and the leak of gas.STM32 is used as main controller of home and area management center. CAN bus module and CC1100 wireless module is adopted between the main controller and main controller family residential district, the main controller uploads data to the management center through the serial port. The computer software is designed to communicate with the management center and to save the information by sensors. The software of client is designed for user to control the main controller to turn the device on or off.The data from NIST is used in the experiment. Performance of system on error rate, alarm time and stability is compared with BP and RBF neural networks. Experiments show that the system can provide accurate detection and the degree of error rate is greatly reduced.