Foreground Information Fusion And Evacuation Path Selection Based On Dynamic Triangulation Network

Fire is a catastrophic problem faced by the human being cause its frequently occurrence, it not only brings us a huge economic losses, but also a serious threat to people’s life safety. During the fire rescue, the damage of advanced inherent network equipment leads to communication breakdown and firefighters’ casualties. So the cable network features that low efficiency and fragile makes it difficult to be applied under the bad environment. Dynamic triangulation network is a special kind of mobile ad-hoc network, it doesn’t depend on any fixed facilities to form a strong anti-destroying ability of network, and it is more suitable for fireground rescue. In the network, Robots act as the network reference node, the robots can not only cover fire information, but also can they help firefighters search and rescue work. It provides guarantee for the firefighters to have a comprehensive control of fire condition and help them search, rescue and evacuation path selection. This paper is mainly aimed at the fire information fusion and firefighters evacuate problem based on dynamic triangulation network in fire and simulation for it.Firstly, the paper builds dynamic triangle network based on two-dimensional large fire environment. Then it launches research about the fire data fusion problem. D-S evidence theory is the fusion method that commonly used now to solve uncertain information fusion problem and does not require a priori probability. Considering that the fire information sensor can bring data conflict due to poor environmental problems, it is bound to affect the result based on D-S data fusion algorithm results. This paper improves the D-S data fusion algorithm with putting forward an improvement of D-S fusion algorithm based on JS-C(Joint Similarity and Credibility). It improves the concepts of the similarity function. Joint similarity function is proposed that it includes similarity function, cosine similarity function and distance similarity function. It can consider the direction similarity and make up for cosine similarity numerical insensitivity. A modified data fusion approach based on joint similarity function and credibility is presented to combining conflicting evidence. Reliability grade is added during the process of calculating certainty. This algorithm can solve the invalidation problem of evidence theory effectively. Then, the paper studies on the path planning in the process of evacuation. At the point of firefighter, the shortest path isn’t the optimal path when evacuation. With the spread of the fire, the nearby of shortest path may not be safe, which would cause damage to evacuees. So the evacuation should consider both the safety and path length. The dynamic triangulation network mentioned above solves the tough problem of communicating in the fire scene and assessing environment. Evidence theory based on similarity function is used to determine the node security levels after fusing the fire-data such as smoke concentration, temperature, oxygen concentration. Prediction mechanism is introduced in order to predict the changes of the security level of node position in the forward path in a short period of time, then risk nodes in network are processed dynamically A*algorithm based on distance conversion is used periodically to calculation the shortest and safest path in real time. By computing and data analyzing, it can adapt to the characteristics of the fire environment changes flexibly.At last, we point out problems to be solved.