Research Of Fire Alarming Technology Based On Multi-sensor Information Fusion

With the rapid development of world’s economy and the increasing frequent of people’sexchanges, ships play an important part in the transportation sector. The developed countries needmore petroleum than other developing countries, and most petroleum is shipped from Middle EastCountries. Ships are floating in the sea, which is different from other transports. Once ships arefired, it is hard to put out the fire depending on sailors and fire extinguishers on the ship. And otherships in the sea can hardly see the firing ship and people on the island can’t help rescue the ship ontime. So it is important to alarm the fire before the fire causes damage. Our research is derived fromthe project of“fire alarm system of power engine room on ships based on multi-sensor informationfusion”. Our main aim is to design a fire alarm system based on multi-sensor information fusion.When fire is flaming in the power engine room, our system can detect the fire timely and give anexact alarm, at the same time the monitor center shows the information of the power engine room.For people, it is easy to detect fire from pictures. Usually, picture holds more information thansensors. Nowadays it is a growing trend to detect fire from pictures. In our designed system, wedevise a method using computer vision to spot flames from a picture. We use HSI and RGB modelsto find the fire color. We also make use of codebook model to find foreground pixels. Combiningthe two features we can detect a flame. Results show that our algorithm is reliable and robust.In our paper, we take advantage of the multi-sensor information fusion technique to judge ifthe power engine room is on fire or not. Single sensor sometimes sends out wrong message andgives a false alarm. In the monitor center, we calculate a fire probability in the characteristic layerusing fuzzy reasoning. If the probability of burning on fire is larger than a threshold, we give analarm.At the beginning we introduce the background and importance of the fire alarm system onships, and analyze the status and developing tendency of fire alarm systems on ships. Gettingthrough the discussion before, we decide to design a fire alarm system based on multi-sensorinformation fusion. The whole system consists of three parts. They are lower computer, uppercomputer and the monitor center. The lower computer is responsible for collecting the sensor dataand transmitting the data to the upper computer through serial communication. The upper computeris in charge of capturing the room picture and processing the picture using computer vision. If theupper computer discovers flames, it transmits the picture to the monitor center. The monitor centertakes charge of computing the fusion algorithm and giving a reliable alarm. In the lower computer,we choose the sensor AM2303 to collect the humidity and temperature and MQ-2 to detect smoke. We choose P89C52 as the main processor in the lower computer and S3C2440A in the uppercomputer. We use personal computer as the monitor center. We use serial communication totransmit data from the lower to upper computer. We design Ethernet network and CAN Busredundancy to send data from upper computer to the monitor center. In the paper, we give thedetails of how to realize the Ethernet and CAN bus in soft and hardware development, and discussin details how to transmit data timely and reliably.