Abstract: In order to improve the accuracy of the estimation model for earthquake disaster death, the paper puts forward5 indexes for estimating the death amount after earthquake disaster, including epicenter intensity, magnitude, earthquake disaster occurrence time, population density and damaged area of building. It regards the “two-phase” laws of earthquake death distribution as the theory, and put forward the earthquake death estimation model based on the partial normal distribution cure, in which the main variable is the epicenter intensity. To improve the accuracy and stability of the estimation model, it amends the partial normal distribution cure by using the method of the coefficient adjustment. In other words, it uses some coefficient, including the magnitude coefficient, earthquake disaster occurrence time coefficient, population density coefficient and building damaged area coefficient, to mend the partial normal distribution cure. The experiment results show that the partial normal distribution cure has some advantages, including low sensitivity and high accuracy, compared with ANN, multivariate linear equation, high order nonlinear equation and logarithmic function. In total, the estimation model of earthquake death can be applied to practice.

Key words: death estimation for earthquake disaster, earthquake disaster, partial normal distribution cure, “two-phase” laws

摘要： 为提高震灾人员受损估计模型的精度,本文提出震中烈度、震级、人口密度、房屋损毁面积和地震发生时间5个估计指标,并依据震灾人员受损的“两期”变动规律,将偏正态分布曲线引入震灾人员受损估计中,提出基于修正偏正态曲线的震灾人员受损估计模型。在模型改进上,以震中烈度为主变量,通过数据回归得到偏正态分布估计曲线;进一步,为提高模型的精度和稳定性,采用震级系数、人口密度系数、房屋损毁面积系数和地震发生时间系数对偏正态分布曲线的估计精度进行修正,通过系数的动态变动提高模型的估计精度。验证结果表明,与人工神经网络、多元线性估计曲线、高阶非线性和对数函数等估计模型相比,修正偏正态分布曲线具有敏感度低和估计精度高的优势,能为应急管理部门提供理论依据和方法借鉴。关键词:人员受损估计;地震灾害;偏正态分布曲线;“两期”规律

关键词: 人员受损估计, 地震灾害, 偏正态分布曲线, “两期”规律