Study On Statistical Sampling Methods Of Natural Gas Used In Industries In Chongqing

According to the features of natural gas and its administration in Chongqing, a new sampling plan about the natural gas used in industried in Chongqing is put forward in this dissertation, which is also based on the experience of Annual Sampling in Manufacutres (ASM) in America and the experience of planning of sampling in industries in China. This plan is applied in industries in Chognqing and the expected results have been obtained, which demonstrates that the plan is feasible for the industries in Chongqing and owns several advantages, for example, it can be apllied easily, it has high ratio of returning and it is of high accuracy.This dissertation chiefly discusses four aspects of this plan.(1) Due to a few large establishments consuming lots of natural gas every year, which are needed to do general investigations, the list sampling technology is employed to determine the important factories that should be investigated generally. The key problem is how to make sure the optimal cutt-off point. A new method named the optimal partitioning method is used, which is different from the traditional way based on the emperic formula.(2) In order to effectively reduce the errors and improve the efficency of sampling, several sampling method, including multi-objective stratified sampling, two-stage sampling and random sampling, are utilized in this dissertation. The consequences show that the complex sampling method mentioned above is feasible.(3) Because of the traditional way without the ability to copy with the multi-objective strata, the minimal dispersion cluster algorithm and the most neighbors clustering algorithm are applied in the process of industry stratification respectively. The results demonstrate that the two ways have the same effect.(4) A sample rotation algorithm combined with PPS and permanent random numbers is designed here to enhance the accuracy of sampling in the future. This algorithm could avoid the problem of"old"samples. What is more important, it could be operated more easily compared to the traditional sample rotation methods.