Research On Key Techniques And Platform Of Emergency Supplies

Emergency Supplies (ES) mentioned in this paper refers to equipments and survival materials needed in the process of disaster rescue after earthquakes. The core problem of ES is to maximize protection of the people’s lives and properties with minum supplies in disaster. It is impossible to reserve appropriate relief supplies for all conditions because of the limitation of ES and randomness of disasters. Distributing and utilizing relief supplies reasonably to minimise losses furthest with limited supplies is the main goal of ES researches.Drawing up a different solution before each disaster is a necessary mean to ensure the correctness of rescue process. The basic infomation of supplies, such as quantity, quality, and variety, must be ascertained firstly for favoring relief supplies reserve. Treating intensity of historical disaster as signal sequence and modeling it with Gaussian Mixed Model, more exact prediction about disaster degree could be acquired. Emergency Material Requirement(EMR) could be acquired farther by computating maximum expectation of disaster degree.Because of information block, communication outage, terror and escape of victims after disaster, it’s difficult to grasp the disaster degree and its distribution exactly in a short time. Processing pictures of disaster area before and after disaster, which could be captured by aerial photo, via edge detection and Hough Transformation, disaster degree could be estimated objectively. As people are moving, the precise number of people in disaster area is not clear after disaster. We can estimate it from number of defectors. By simplifying escape route, via a method transfer information backward, we can deduce the distribution of EMR in disaster area.In the process of relief materials reserve, many factors, such as the source of materials, storage cost, transportation cost, and benefit to relief, should be considered. To support rapid decision, Principal Component Analysis (PCA) is introduced to simplify factors of site selection. Availability and validity to simplify reserve points selection are proved by simulation experiments. To reach relief points rapidly, ES should be reserved in one unified site, but a variety of materials should also be classified storage to quickly distribute, while saving storage costs. Clustering by Nearest Neighbor Graph, ES could be classfied effectively, saving storage cost, protecting rescue efficiency, and providing data for Emergency Supplies Reserves Selection(ESRS). Most traditional methods for site selection are based on Operational Research. It is difficult to modeling and computing for them. Ant Colony Optimization(ACO) is introduced to solve the ESRS problem, reducing computational complexity, while preventing local optima traps.For the life and property security in disaster area, relief pathes are choosed through external observation and history data only, before the actual survey traffic conditions. The vehicles run randomized and concurrent in the rescue transport process. Stochastic Petri Net is introduced for emergency rescue transport process modeling, which can easily solve the problem of the traffic situation estimating in disaster area. Without considering the resource cost and the shortage of resource conflicts, emergency logistics path selection method with time expectation is more adaptable than the path selection methods for shortest possible time or shortest possible distance.After the disaster, the discussion about donated material is one of the hot topic of Internet and other interactive media. The type, amount and distribution of donated material can be predicted through the attention of discusses about donations. Topic Detection and Tracking is introduced to collect and forecast information on donations. Introducing the Anisotropic Diffusion technique is introduced into WordNet to reflect the influence of the related words while protecting the boundary between theme. A limited memory and passive forgetting mechanism is proposed to index WordNet partly with limited storage. Donation Topic and its words are quickly detected by scanning and processing none of the inactive words.