论文部分内容阅读
图象识别是近二十年来发展起来的一门学科,它已广泛应用于许多领域中。盖尔芬德(I.M.Gelfand)、普雷斯(F.Press)等人将它用于地震危险区的划分。本文将图象识别方法用于地震预测中,以识别强震发生的时间。 按一定标准将所研究的全部时间划分为危险时间段D和不危险时间段N。以问题表的形式提出大地震前中等地震活动的特性,然后分两步进行图象识别: 1.“学习”。对P个时间段m个问题的回答是m×p的矩阵,回答以二进制(是或非)表示。通过“学习”,识别出一个、两个或三个问题组合的新“特征”,称之为D和N的“性质”。 2.“投票”。D和N“性质”数目的差是△,当△大于或等于某阈值时,则识别为危险段D,否则为N。 结果表明,大地震发生前的一定时期内,中等地震活动增至一定水平、相差半级的中等地震活动水平的比值较正常情况增高以及大震前中等地震活动随时间增强等“性质”的综合,表明未来时间段內可能发生大地震。 此外还作了控制试验,说明图象识别结果是稳定的。
Image recognition is a discipline developed in recent twenty years. It has been widely used in many fields. I.M.Gelfand, F. Press and others use it for the division of seismic hazards. In this paper, image recognition method is used in earthquake prediction to identify the occurrence of strong earthquakes. According to a certain standard all the time studied is divided into dangerous time D and non-hazardous time N. In the form of a question table, the characteristics of moderate earthquakes before large earthquakes are proposed, and then image recognition is carried out in two steps: 1. “Learning”. The answer to m questions in P periods is m × p matrix, and the answer is binary (yes or no). Through Learning, new “features” of one, two, or three problem combinations are identified as the “nature” of D and N. Vote The difference between the numbers of D and N “properties” is Δ, and when Δ is greater than or equal to a certain threshold, it is identified as dangerous segment D; The results show that within a certain period prior to the earthquake, the moderate seismicity increases to a certain level, and the ratio of moderate seismicity with a difference of half level is higher than that of the normal and the nature of the moderate seismic activity increases with time , Indicating that a major earthquake may occur in the future time period. In addition, a control experiment was made, which shows that the result of image recognition is stable.