CN-121350849-B - Prediction method and system for casualties caused by earthquake landslide in consideration of population dynamic characteristics and environmental heterogeneity

CN121350849BCN 121350849 BCN121350849 BCN 121350849BCN-121350849-B

Abstract

The invention discloses a method and a system for predicting casualties caused by earthquake landslide with consideration of population dynamic characteristics and environmental heterogeneity, wherein the method comprises the following steps of collecting all environmental factor data and multisource population distribution data in a target earthquake region according to a preset space unit; the method comprises the steps of collecting environment factor data, inputting the collected environment factor data into a preset logistic regression model to obtain earthquake landslide probability, processing multi-source population distribution data by utilizing a data fusion method to obtain dynamic population distribution of an earthquake occurrence period, processing the environment factor data based on a random forest algorithm model and a SHAP method to obtain environment heterogeneity parameter and environment factor contribution degree, and obtaining probability distribution of casualties of earthquake landslide according to a probability prediction model of casualties caused by the earthquake landslide. The method improves the accuracy of the casualties assessment caused by the earthquake landslide, and provides a reliable basis for making an action scheme for earthquake rescue.

Inventors

GAO HUIRAN
XU CHONG

Assignees

应急管理部国家自然灾害防治研究院

Dates

Publication Date: 20260512
Application Date: 20251216

Claims (9)

1. A method for predicting casualties caused by earthquake landslide taking into account population dynamics and environmental heterogeneity, comprising: Collecting all environmental factor data and multi-source population distribution data in a target earthquake area according to a preset space unit, wherein the environmental factor data comprises a topography factor, a geological factor and a basic geographic environmental factor, and the multi-source population distribution data comprises population kilometer grid data and mobile positioning data; Inputting the collected environmental factor data into a preset logistic regression model to obtain the probability of earthquake landslide; processing the multisource population distribution data by using a data fusion method to obtain dynamic population distribution of earthquake occurrence time periods; processing the environmental factor data based on a random forest algorithm model and a SHAP method to obtain environmental heterogeneity parameters and environmental factor contribution; And fifthly, constructing a probability prediction model of casualties caused by the earthquake landslide according to the environment heterogeneity parameter and the contribution degree of the environment factors, and inputting the environment heterogeneity parameter, the earthquake landslide probability and the dynamic population distribution of the earthquake occurrence period into the probability prediction model of casualties caused by the earthquake landslide to obtain the probability distribution of casualties caused by the earthquake landslide.
2. The method for predicting casualties caused by earthquake landslide taking into consideration population dynamics and environmental heterogeneity according to claim 1, wherein the step two specifically comprises: obtaining the correlation between any two environmental factors in all environmental factor data by utilizing a spearman rank correlation formula, and screening the environmental factors according to the correlation to obtain screened environmental factors; And inputting the screened environmental factors into a preset logistic regression model to obtain the earthquake landslide probability.
3. The method for predicting casualties caused by earthquake landslide taking into consideration population dynamics and environmental heterogeneity as recited in claim 1, wherein said step three specifically includes: Respectively carrying out two-dimensional discrete Fourier transform on population kilometer grid data and mobile positioning data to obtain a frequency domain image, wherein the mobile positioning data comprises anonymous user equipment space position data sets in a preset time window before and after earthquake occurrence time acquired by an operator communication network; Obtaining amplitude information based on frequency domain images corresponding to the population kilometer grid data, and obtaining phase information based on frequency domain images corresponding to the mobile positioning data; And processing the amplitude information and the phase information according to the two-dimensional inverse discrete Fourier transform to obtain the dynamic population distribution of the earthquake occurrence period.
4. The method for predicting casualties caused by earthquake landslide taking into consideration population dynamics and environmental heterogeneity according to claim 1, wherein the step four specifically comprises: inputting the environmental factor data into a preset random forest algorithm model to obtain environmental heterogeneity parameters; And obtaining the contribution degree of the environmental factors based on the SHAP method and the base-Ni value index method on the basis of the random forest algorithm model.
5. The method for predicting casualties caused by earthquake landslide taking account of population dynamic characteristics and environmental heterogeneity as recited in claim 4, wherein the method for obtaining the contribution degree of the environmental factors based on the SHAP method and the radix value index method on the basis of a random forest algorithm model specifically comprises the following steps: calculating according to the environmental heterogeneity parameters to obtain characteristic contribution values, and constructing an n multiplied by m characteristic contribution value matrix according to the characteristic contribution values, wherein n is a sample size, and m is a characteristic number; Calculating the decrease amount of the base value of each feature when splitting nodes in the random forest algorithm model, and carrying out weighted fusion on the feature contribution value matrix and the base decrease amount to obtain the importance score of the environmental factor; And normalizing the importance scores of the environmental factors to obtain the contribution degrees of the environmental factors.
6. The method for predicting casualties caused by earthquake landslide taking account of population dynamics and environmental heterogeneity as set forth in claim 4, wherein the constructing a probability prediction model of casualties caused by earthquake landslide according to the environmental heterogeneity parameter and the environmental factor contribution degree specifically includes: the environmental factors are ordered according to the contribution degree of the environmental factors to obtain the environmental factors of the second ranking, and an environmental correction coefficient is obtained according to the environmental factors of the second ranking and the environmental heterogeneity parameters; And constructing a probability prediction model of casualties caused by earthquake landslide according to the environment correction coefficient.
7. The method for predicting casualties caused by seismic landslide with consideration of demographic dynamics and environmental heterogeneity as recited in claim 6 wherein said environmental correction factor is expressed by the following formula: Wherein, the As the environmental correction factor(s), As a parameter of the heterogeneity of the environment, And Respectively different normalized environmental factors, As the coefficient of the light-emitting diode, And y is the casualty distribution of the historical real earthquake case personnel.
8. The method for predicting casualties caused by earthquake landslide taking into account demographic dynamic characteristics and environmental heterogeneity as recited in claim 7, wherein the probability prediction model of casualties caused by earthquake landslide is expressed by the following formula: Wherein R pop is the probability distribution of casualties of earthquake landslide, H landslide is the occurrence probability of earthquake landslide, P dynamic is the dynamic population distribution of earthquake occurrence period, and E uncertainty is the environmental heterogeneity parameter.
9. A system for predicting casualties caused by seismic landslide taking into account demographic dynamics and environmental heterogeneity, comprising: The system comprises a data collection module, a data processing module and a data processing module, wherein the data collection module is used for collecting all environmental factor data and multi-source population distribution data in a target earthquake area according to a preset space unit, the environmental factor data comprise a topography factor, a geological factor and a basic geographic environmental factor, and the multi-source population distribution data comprise population kilometer grid data and mobile positioning data; the landslide prediction module is used for inputting the collected environmental factor data into a preset logistic regression model to obtain the probability of earthquake landslide; the dynamic population distribution module is used for processing the multisource population distribution data by using a data fusion method to obtain dynamic population distribution of earthquake occurrence time periods; The element extraction module is used for processing the environmental factor data based on the random forest algorithm model and the SHAP method to obtain the environmental heterogeneity parameter and the environmental factor contribution degree; The prediction module is used for constructing a probability prediction model of casualties caused by the earthquake landslide according to the environment heterogeneity parameter and the contribution degree of the environment factors, inputting the environment heterogeneity parameter, the earthquake landslide probability and the dynamic population distribution of the earthquake occurrence period into the probability prediction model of casualties caused by the earthquake landslide, and obtaining the probability distribution of casualties caused by the earthquake landslide.

Description

Prediction method and system for casualties caused by earthquake landslide in consideration of population dynamic characteristics and environmental heterogeneity Technical Field The invention relates to the technical field, in particular to a method and a system for predicting casualties caused by earthquake landslide in consideration of population dynamic characteristics and environmental heterogeneity. Background The earthquake triggered geological secondary disasters are represented by various slope material movement forms, and the mechanical mechanism of the earthquake triggered geological secondary disasters mainly comprises various typical types, namely rock collapse, and is mainly characterized in that a steep rock mass is instantaneously disintegrated under the action of earthquake vibration load, rolling stone disasters in a discrete body movement form are represented by bouncing rolling processes of single or multiple rock masses along a slope surface, continuously deformed soil landslide, the mechanical essence of which is shear damage on a potential sliding surface, mud-rock flow movement with the characteristics of solid-liquid two-phase flow and the like. These disaster types cause significant casualties in multiple strong jolts due to their sudden and interlocking effects. The existing earthquake landslide population loss evaluation system mainly adopts a statistical method based on field investigation, and in the implementation process, the method needs to organize multi-disciplinary joint investigation groups, and disaster data is summarized in a point-by-point check mode. The mode exposes obvious timeliness defects in earthquake emergency response, and the time spent for checking all landslide points from the occurrence of an earthquake is long, and the time is seriously delayed from a golden rescue period. Meanwhile, due to the interference of factors such as the landfill depth of the landslide body, the rescue progress and the like, obvious uncontrollable delayed reporting phenomenon exists in field statistics. The earth observation technology provides a new technical means for landslide identification, but is limited by the revisiting period, meteorological influence and data precision of the optical remote sensing satellite, so that the real-time requirement of emergency response is difficult to meet in remote sensing monitoring. In order to meet the requirements of timeliness and accuracy of earthquake disaster emergency, quantitative assessment means for the number of casualties caused by earthquake landslide are proposed in recent years. Currently, the prior art provides a rapid population loss assessment method based on the occurrence probability of earthquake landslide. According to the method, firstly, an earthquake landslide probability field is constructed based on a Newmark displacement model or a machine learning algorithm, and population loss caused by earthquake landslide is estimated by combining population density distribution. The technology compresses the evaluation timeliness of the casualties of the earthquake landslide from a plurality of days to a plurality of hours in the traditional method, and provides key decision support for emergency rescue and resource allocation. However, a more perfect technical scheme is not formed in the aspect of earthquake landslide casualties evaluation at present. In the existing earthquake landslide casualties evaluation research, on one hand, static population assumption is generally adopted, population dynamic characteristics are ignored, systematic deviation exists between casualties risk evaluation and spatial distribution prediction results and actual disaster conditions, and on the other hand, the spatial characteristics of environmental elements and complex association mechanisms thereof are often simplified into single factors or deterministic mapping relations, and space-time non-stationarity of the action relation between the spatial heterogeneity of the environmental elements and human activities in actual disaster scenes is not considered. Static assumptions about demographics and heterogeneity that do not take into account environmental impact factors are key issues that affect prediction accuracy. The gap between the casualties assessment result and the actual demands of emergency treatment and rescue is still larger. Disclosure of Invention In order to solve the problem that the prediction accuracy of casualties caused by the heterogeneity of population dynamic characteristics and environmental influence factors is low in the prior art, the invention provides a prediction method and a system for casualties caused by earthquake landslide taking the population dynamic characteristics and the environmental heterogeneity into consideration. In order to achieve the above object, the technical scheme of the present invention is as follows: the first aspect of the invention provides a method for predicting casualties caused b