Research On Emergency Material Demand Forecast Method Under Large-scale Earthquakes

In recent years, the 8.0 magnitude Earthquake in Wenchuan, the 7.1 magnitude earthquake in Yushu and 9.0 magnitude Earthquake in Japan have caused heavy casualties and severe property damage. A lot of people paid attention to them. To minimize the casualties resulted from earthquakes, the rescue work should begin as soon as possible, and the basic life of victims should be ensured after earthquakes. Usually, large-scale earthquakes resulted in great damage to infrastructure in disaster areas, which makes the information exchange blocked, so the information of emergency material demand cannot be received. This brings many problems to raise, distribute and supply emergency materials. Therefore, how to get information about emergency materials needed by people in disaster the first time after earthquakes has became a problem which needs be solved for rescue and relief work.In this paper, firstly, the characteristic and contents of emergency material demand are analyzed and studied, and the studying coverage is limited to the quantities of emergency materials in early time after earthquakes. Based the characteristics of sudden, uncertainty, supplier-controlled and the weak economy in emergency demand, the regular forecast models are not used step by step for forecasting emergency materials demand. Through research and analysis, a dynamic real-time estimation method has been made in this paper. The method can be divided into two stages. In the first stage, based on the information which can be collected quickly and easily, the principal component analysis and BP neural network model is constructed to predict the casualties in earthquakes. In the second stage, the estimate data of casualties and the inventory management knowledge are made used of to estimate the quantity of materials under the situation of unknown demand.As for the principal component analysis and BP neural network model for casualties estimation, firstly, the eight parameters including the magnitude, time, intensity, etc. are chose from many parameters which influenced the casualties, and this eight parameters can be obtained directly or by the rapid assessment methods the first time after earthquakes. Secondly, the information about several large-scale earthquakes in history is obtained form China Earthquake Information Net, books Database of History Earthquakes in China and Statistical Yearbooks as samples. To avoid correlation among parameters selected, the data dimension is reduced by principle component method. Then a three layer BP neural network model is constructed to train the data in samples. Through that the simulation data and actual data in test samples has been compared, the data simulated by the model can meet the error requirement. For the emergency material estimation method, according to its characteristics, the safety stock theory is employed to set up formulas for estimate the food, clothes and basic drugs used by people in disasters.Finally, the method structured in this paper is used to estimate the demand of convenient food, bottled water , tents and anti-inflammatory in the 7.1 magnitude earthquake which occurred in Yushu, Qinghai.