加油站可燃性气体颗粒危险程度检测方式研究

摘要

Research accurate recognition of combustible gas concentration. Combustible gas kinds of gas station are complex. This paper put forward an explosion risk detection algorithm based on local characteristic value similarity polymerization. It obtains particle feature space distribution relationship by the mode of local discrete to test guarantee risks. The experiment results show that the algorithm can improve the accuracy of identification, and the satisfactory effect is gained.%研究加油站可燃性气体颗粒危险程度准确检测问题.加油站中大量不同种类的可燃性气体颗粒分布在空气中,达到标准浓度会发生火灾.多种可燃性气体颗粒混合,造成每种可燃性颗粒标准浓度发生改变.传统的检测方式通过单一可燃性气体颗粒浓度检测危险程度,一旦发生多种气体混合,颗粒浓度将发生畸变,导致无法确定检测浓度,造成检测准确率过低的问题.为了避免上述缺陷,提出了基于局域特征值相似性聚合算法的危险程度检测方式.通过局域离散方式获取颗粒特征空间分布关系,根据特征值相似性聚合方式实现颗粒危险程度检测.实验证明,这种检测方式提高了加油站混合可燃性气体颗粒危险程度检测的准确率,取得了满意的效果.