Research On Emergency Warning System For Natural Disasters Based On Multi-sensor Information Fusion

Natural disasters are anomalies which occur in the nature world that humansdepend, and have a wide distribution and strong uncertainty, they are very destructiveand inevitable. If they occur, the harm on human society is often startling, resulting ina large number of casualties, property damage, social instability, and resource damage.The natural disasters cause serious disaster losses and the information is uncertain andnon-linear, Thesis builds a natural disaster emergency warning decision model basedon multi-sensor technology, information fusion technology and other advancedtechnologies, thus the accuracy of the prediction of natural disasters is effectivelyimproved. Model data layer, feature layer and the decision-making level fusionmodule functions and system structure are also analyzed and researched, providing astrong technical and system support for natural disaster emergency warning decisionmanagement.I study and analyze the structure of the general framework of the multi-sensorinformation fusion, information fusion system for natural disaster warning decisionstructure, as well as hierarchical fusion model which filters complex multipleheterogeneous information, normalizes and fuses in layer level in order to get accurateinformation to support decision-making. By analyzing the characteristics of themulti-sensor information in various stages of fusion, the research selects the thresholdmethod as a feature level fusion algorithm and Dempster-Shafter (D-S) evidencetheory as a method of decision-level fusion. It also uses analog data to verifythreshold value method and DS evidence theory method, it proved the modelincreases the feasibility and reliability of the system of decision-making model.In this paper, I expounds the importance and urgency of establishing the warningdecision model of natural disasters frequent natural disasters because of the serioussituation of disaster losses; I also study the characteristics of multi-sensor, thestructure of the basic principles of information fusion, information fusion model andalgorithm. In mechanism research, I study the principle of multi-sensor detection ofnatural disasters, the level of multi-sensor information fusion as well as the level ofthe algorithm. On this basis, the paper design a natural disaster warning decisionmodel based on information fusion and verify it using data. It proves that the systemimproves the accuracy of detection of natural disasters and reduces the rate of falsepositives. The paper takes forest fires for example, the model analyzes firstly multi-sensorfusion forecast system data sources, and in accordance with the goals it selects theappropriate monitoring sensor, then it filters the multi-sensor multi-dimensionalheterogeneous information and makes fusion processing. The information processedthrough information fusion model eliminates the ambiguity and redundancy of theinitial information and finally gets accurate decisions.