CN118397796A

CN118397796A - Highway high slope landslide hazard early warning method based on evaluation function training

Info

Publication number: CN118397796A
Application number: CN202410561921.9A
Authority: CN
Inventors: 邵景干; 叶国永; 王钰轲; 刘旭玲; 冯忠居; 秦恒洁; 李森; 王鹏鹏; 王富春; 张志远
Original assignee: Henan Jiaoyuan Engineering Technology Group Co ltd; Zhengzhou Dongchen Science & Technology Co ltd; Zhengzhou University of Light Industry
Current assignee: Henan Jiaoyuan Engineering Technology Group Co ltd; Zhengzhou Dongchen Science & Technology Co ltd; Zhengzhou University of Light Industry
Priority date: 2023-08-07
Filing date: 2024-05-08
Publication date: 2024-07-26
Also published as: CN117058845A

Abstract

The invention discloses a highway high slope landslide hazard early warning method based on evaluation function training, which comprises the following steps: deploying a plurality of sensors at key points of the expressway high side slope, and acquiring real-time monitoring data by utilizing the sensors; satellite image data of a highway high slope area is acquired by utilizing a satellite remote sensing technology; integrating the acquired real-time monitoring data with the processed high-resolution image, and carrying out space-time scale fusion on the integrated data source; constructing a disaster early warning model based on evaluation function training, and calculating damage index scores of the slope states by using the disaster early warning model; making a corresponding slope state early warning threshold; if the damage index score exceeds the slope state early warning threshold, a corresponding warning signal instruction is sent out, and rescue measures are taken according to a preset early warning rule; and constructing a three-dimensional perspective view of the expressway high side slope area. The method and the device can quickly and accurately discover and early warn the landslide risk possibly occurring.

Description

Highway high slope landslide hazard early warning method based on evaluation function training

Technical Field

The invention relates to prediction of a road high slope landslide hazard, in particular to a highway high slope landslide hazard early warning method based on evaluation function training.

Background

The expressway high slope refers to a slope of an expressway arranged in a region with steeper terrain. In general, the construction of a highway requires the change of the original topography by means of excavation or water diversion, etc., while a high slope is a structure specially constructed for preventing landslide of a roadbed, protecting the road surface and safety of vehicles. Because the relief fluctuation of the high side slope is extremely large, the soil on the surface of the ground is loose and is obviously influenced by natural factors such as wind erosion, water erosion and the like, landslide disasters are extremely easy to occur on the high side slope.

The highway high slope landslide hazard is a natural disaster, and the occurrence of the disaster can lead to serious traffic accidents, casualties, property loss and other consequences. Therefore, the method has important significance in predicting the landslide hazard of the high slope and taking reasonable measures for the landslide hazard.

In the expressway high slope landslide hazard early warning method, in order to improve prediction accuracy, novel technologies such as a sensor monitoring technology, a remote sensing image processing technology and a machine learning algorithm are introduced, the conditions of dangerous areas are monitored in real time by combining the advantages of the novel technologies, environmental change data are collected in time, and therefore the risk of the high slope landslide hazard can be predicted better. However, the prior art also exposes some defects in the application process, such as inaccurate data acquisition, and the acquired remote sensing image data is influenced by various interference factors to reduce the image quality, and the recognition accuracy is seriously influenced, so that the early warning result is deviated. Therefore, the technical field of view is still required to be expanded, the early warning algorithm is optimized, a brand-new state prediction effectiveness guarantee mechanism is explored, and more accurate and reliable early warning of the expressway high slope landslide disaster is realized.

For the problems in the related art, no effective solution has been proposed at present.

Disclosure of Invention

Aiming at the problems in the related art, the invention provides an expressway high slope landslide hazard early warning method based on evaluation function training, so as to overcome the technical problems in the prior art.

For this purpose, the invention adopts the following specific technical scheme:

The highway high slope landslide hazard early warning method based on evaluation function training comprises the following steps:

s1, deploying a plurality of sensors at key points of a highway slope according to a terrain distribution rule, and acquiring real-time monitoring data by using the sensors;

s2, acquiring satellite image data of a highway high slope area by utilizing a satellite remote sensing technology, and performing geometric correction processing on the satellite image data;

S3, integrating the acquired real-time monitoring data with the processed high-resolution image, and carrying out space-time scale fusion on the integrated data source to extract relevant characteristic variables;

s4, constructing a disaster early warning model based on evaluation function training, and calculating damage index scores of the slope states by using the disaster early warning model;

s5, acquiring a history instance related to landslide disasters, and combining a high slope landslide generation mechanism to formulate a corresponding slope state early warning threshold;

S6, if the damage index score exceeds the slope state early warning threshold value, a corresponding warning signal instruction is sent out, and rescue measures are taken according to a preset early warning rule;

and S7, constructing a three-dimensional perspective view of the expressway high side slope area and providing visualization.

Preferably, the satellite remote sensing technology is used for collecting satellite image data of the expressway high slope area, and the geometric correction processing of the satellite image data comprises the following steps:

s21, acquiring satellite images of a highway high-side slope area by using a high-resolution remote sensing satellite;

S22, selecting ground control points from satellite image data, and marking coordinate values of each ground control point under a pixel coordinate system;

S23, determining a geographic coordinate correction model according to the geometric distortion property and the selected characteristic control points;

S24, performing geographic coordinate correction calculation on the satellite image by using a geographic coordinate correction model to obtain a geographic coordinate set after pixel correction;

S25, performing image correction operation on the corrected satellite image under a geographic coordinate system to obtain a processed high-resolution image;

s26, resampling the high-resolution image by using a local search method, calculating a gray value of a target point, and adding the gray value into the processed high-resolution image.

Preferably, the resampling the high resolution image by using the local search method, calculating a gray value of the target point, and adding the gray value to the processed high resolution image includes the steps of:

s261, obtaining an image point of a gray value to be solved, and searching the image point geographical coordinates of the gray value to be solved in a geographical coordinate set after image point correction;

s262, taking an image point of a gray value to be solved as a center, taking a preset threshold value as a radius, and determining a search area;

S263, searching four pixel points closest to an image point of the gray value to be solved in a search area;

s264, reversely calculating by using a geographic coordinate correction model to obtain conjugate points, and determining a local area of the search area based on the conjugate points;

s265, acquiring gray values of four pixel points, performing distance weight calculation to obtain the gray value of the conjugate point, and adding the gray value into the processed high-resolution image.

Preferably, the integrating the acquired real-time monitoring data with the processed high-resolution image, and performing space-time scale fusion on the integrated data source, and extracting the relevant feature variables includes the following steps:

s31, respectively preprocessing the real-time monitoring data and the processed high-resolution image;

s32, extracting characteristic information in the processed high-resolution image, and extracting relevant physical quantity and index in real-time monitoring data;

s33, converting the space-time scale of the real-time monitoring data and the high-resolution image into a matched pixel size and a matched time interval;

S34, fusing related physical quantities and indexes to high-resolution images of corresponding positions by using a spatial interpolation algorithm, and establishing a space-time context relation;

and S35, extracting key characteristic variable information based on the matched data sources.

Preferably, the expression of the geographical coordinates corrected by the pixels is:

wherein X, Y each represent geographic coordinates in the real world;

a. r each represents a basic offset;

b. e each represents a parameter related to the scaling and the rotation angle.

Preferably, the constructing a disaster early warning model trained based on the evaluation function and calculating the damage index score of the slope state by using the disaster early warning model comprises the following steps:

s41, acquiring a history instance sample set related to landslide disasters, and constructing an error cost function according to the distribution of the history instance sample set;

S42, calculating respective misclassification costs of the fault sample and the normal sample by using an error cost function;

s43, extracting a sample set from the historical instance sample set by adopting an Adaboost algorithm, and forming a sample subset and a corresponding out-of-bag data set;

s44, generating a decision tree model for the sample subset;

s45, carrying out classification test on each decision tree by using the out-of-bag data set corresponding to the sample subset, and calculating the classification accuracy of each decision tree;

S46, giving weight to each decision tree by using the classification accuracy, and generating a final disaster early warning model based on the new data set.

Preferably, the generating a decision tree model for the subset of samples comprises the steps of:

s441, selecting representative features from a feature space of a historical instance sample set, and forming a feature subset;

S442, extracting characteristic attributes of the representative characteristics, and calculating misclassification cost reduction values of the characteristic attributes in the characteristic subsets;

S443, selecting the characteristic attribute with the largest misclassification cost reduction value as the splitting basis of the current node;

S444, distributing samples in the sample subset to different sub-nodes for processing according to the characteristic two-class criterion;

s445, repeating S442-S444 until the samples in the sample subset are classified or the maximum node layer number is reached, and finally generating a decision tree model.

Preferably, when the damage index score exceeds the slope state early warning threshold, a corresponding warning signal instruction is sent, and rescue measures are taken according to a preset early warning rule, including the following steps:

s61, comparing the damage index score with a slope state early warning threshold, and if the damage index score exceeds the slope state early warning threshold, sending a corresponding warning signal instruction to a management department;

s62, starting an emergency plan, and dispatching a professional team to rescue according to the scene disaster condition;

S63, constructing a risk assessment model based on landslide disasters, and assessing influence and diffusion degree caused by the disasters by using the risk assessment model;

s64, control and management of the highway high slope area are enhanced.

Preferably, the emergency plan includes emergency equipment material preparation, evacuation of personnel and property, emergency deployment planning and emergency instruction signaling.

Preferably, the constructing a risk assessment model based on landslide disasters and using the risk assessment model to assess the influence and the extent of spread caused by the disasters comprises the following steps:

S631, acquiring landslide hazard evaluation indexes, and classifying and grading the hazard evaluation indexes;

S632, determining the weight of each disaster evaluation index by utilizing an AHP algorithm, and acquiring a relative weight matrix among influence factors;

s633, building a relevant fuzzy set based on the weight level of the disaster evaluation index, and determining a membership function;

s634, determining a factor set and an evaluation set of an evaluation object, establishing single factor evaluation, and constructing a landslide hazard risk evaluation model;

and S635, evaluating the risk of the landslide disaster by using a risk evaluation model.

The beneficial effects of the invention are as follows:

1. According to the invention, the acquired real-time monitoring data and the processed high-resolution image are integrated, and the integrated data sources are subjected to space-time scale fusion, so that valuable data can be widely collected on the time and space levels, further, the change rule and unknown information of the expressway high slope area can be more comprehensively and rapidly analyzed and researched, hidden danger points in the expressway high slope area are identified and classified, and the acquired remote sensing image is subjected to geometric correction, so that the image quality is further improved, the identification precision is also improved, and the deviation of the early warning result is avoided.

2. According to the invention, through constructing the disaster early warning model based on evaluation function training, various sensor data, remote sensing image information, geological data and the like can be input into the model in the model training stage, a proper disaster risk evaluation index is formed according to data characteristics and related rules, the set index coefficient rule is learned, and the optimal detection and early warning discrimination parameters are obtained, so that not only can cause analysis be more accurately carried out, but also quantitative treatment can be carried out on the prediction mechanism flow, and through the landslide disaster early warning model trained by the evaluation function, the possible landslide risk can be found and early warned rapidly and accurately, the actual guarantee is provided for the manager to formulate a more accurate coping scheme, and the traffic safety can be improved.

3. According to the invention, the risk assessment model based on landslide disasters is constructed, and the influence and the diffusion degree caused by the disasters are assessed by using the risk assessment model, so that the influence range and the diffusion degree after the disasters occur can be comprehensively assessed, and the influence and the diffusion degree thereof which are possibly caused by the landslide disasters of the high slope can be analyzed and explored, so that early warning personnel can refine design and implement precautionary measures according to the prediction result of the risk assessment model, and meanwhile, the method plays a certain guiding and promoting role in high-quality, high-efficiency and safe construction of highways.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a flowchart of an expressway high slope landslide hazard early warning method based on evaluation function training according to an embodiment of the invention.

Detailed Description

For the purpose of further illustrating the various embodiments, the present invention provides the accompanying drawings, which are a part of the disclosure of the present invention, and which are mainly used for illustrating the embodiments and for explaining the principles of the operation of the embodiments in conjunction with the description thereof, and with reference to these matters, it will be apparent to those skilled in the art to which the present invention pertains that other possible embodiments and advantages of the present invention may be practiced.

According to the embodiment of the invention, the expressway high slope landslide hazard early warning method based on evaluation function training is provided.

The invention will now be further described with reference to the accompanying drawings and detailed description, as shown in fig. 1, a configuration method according to an embodiment of the invention, the method comprising the steps of:

S1, deploying a plurality of sensors at key points of a highway slope according to a terrain distribution rule, and acquiring real-time monitoring data by using the sensors.

In particular, the deployment of multiple sensors at the key points can improve the effectiveness and accuracy of the monitoring data. And the sensors are deployed according to the terrain distribution rule, so that the whole high slope area can be better covered, and the global performance of the monitoring data is improved. Based on the real-time monitoring data, the early warning system can establish a prediction model by using a machine learning algorithm to identify and predict potential landslide hazards.

Specifically, the detection content of the sensor includes: compression, stretching, expansion and other physical deformations of the rock-soil mass; the change conditions of soil moisture, pore water level and water pressure; temperature and heat release conditions of soil and rock mass, vibration of the slope, vibration and frequency thereof, etc.

S2, satellite image data of the expressway high slope area are collected by utilizing a satellite remote sensing technology, and geometric correction processing is carried out on the satellite image data.

Specifically, the content collected by the satellite remote sensing technology comprises: morphological properties such as slope morphology, elevation, gradient and the like; earth surface evolution processes such as collapse, landslide, earth surface subsidence and the like; the vegetation changes, which vary depending on the type of vegetation.

The method for acquiring satellite image data of the expressway high slope area by utilizing the satellite remote sensing technology and performing geometric correction processing on the satellite image data comprises the following steps of:

S23, determining a geographic coordinate correction model according to the geometric distortion property and the selected characteristic control points.

Specifically, image distortion factors including deviation, scaling, rotation, nonlinear distortion and the like are calculated by using a feature control point through a least square method, and a geometric inversion model (a polynomial model, a perspective transformation model and the like) is established according to the calculated distortion parameters and pixel coordinates and real world coordinates of the feature control point to realize geographic coordinate correction.

S24, performing geographic coordinate correction calculation on the satellite image by using the geographic coordinate correction model to obtain a geographic coordinate set after pixel correction.

Specifically, the expression of the geographic coordinate correction calculation is:

the matrix form is as follows:

The expression of the geographical coordinates after pixel correction is:

wherein X, Y each represent geographic coordinates in the real world;

a. r each represents a basic offset;

b. e each represents a parameter related to the scaling and the rotation angle.

The method for resampling the high-resolution image by using the local search method, calculating the gray value of the target point and adding the gray value into the processed high-resolution image comprises the following steps:

Specifically, in order to solve the pixel value after resampling, firstly, determining the new size of the image after resampling by the geographic coordinate set Z, then finding the minimum geographic coordinate Q (X, Y) in the geographic coordinate set Z, and starting from the point Q, solving the gray value one by one, wherein the stepping value is the resolution of the image after correction, and the specific solving steps are as follows: setting an image point of a gray value to be solved as a point K, finding the geographical coordinates of the point K in a geographical coordinate set Z, setting a value R according to actual needs, and determining a search area D by taking the point K as a center and taking R as a search radius.

When the local search method is adopted, the most critical point is the determination of the search area D. After the conjugate point Q (X, Y) is obtained by inverse calculation by using the correction model, the upper left corner Q (X-R, Y-R) and the lower right corner Q (X+R, Y+R) of the region D are determined by the defined search radius R. Because the geographic coordinate correction results are sequentially stored in an array one by one, the coordinates of the upper left corner and the lower right corner of the area D in the geographic coordinate set Z can be found according to the known image sizes, distances from the coordinate points to conjugate points are compared one by one until four pixel points closest to the conjugate points are found, the gray values of the four pixel points are obtained after searching, and the gray value D _Q of the conjugate points Q (X, Y) is obtained through distance weight calculation.

And S3, integrating the acquired real-time monitoring data with the processed high-resolution image, and carrying out space-time scale fusion on the integrated data source to extract relevant characteristic variables.

The method comprises the steps of integrating acquired real-time monitoring data with processed high-resolution images, carrying out space-time scale fusion on integrated data sources, and extracting relevant characteristic variables, wherein the method comprises the following steps of:

S4, constructing a disaster early warning model based on evaluation function training, and calculating damage index scores of the slope states by using the disaster early warning model.

The construction of the disaster early warning model based on the evaluation function training and the calculation of the damage index score of the slope state by using the disaster early warning model comprises the following steps:

S41, acquiring a history instance sample set related to landslide disasters, and constructing an error cost function according to the distribution of the history instance sample set.

Specifically, the expression of the error cost function is:

g _i,g_j each represents a category;

d _i represents the Euclidean distance of the center of category g _i from the center of the total sample set;

d _j represents the Euclidean distance of the center of category g _j from the center of the total sample set;

j represents the number of historical instance sample sets;

ζ represents a regularization parameter.

s43, extracting a sample set from the historical instance sample set by adopting an Adaboost algorithm, and forming a sample subset and a corresponding out-of-bag data set.

Specifically, extracting a sample set from a historical instance sample set by adopting an Adaboost algorithm, and forming a sample subset and a corresponding out-of-bag data set, wherein the method comprises the following steps of:

Step one: for landslide hazard prediction problems, a data set can be prepared, wherein each sample represents an observation value of a specific place, time period and monitoring parameter, and each sample also needs to have a binary mark to represent whether landslide hazard occurs or not;

Step two: initializing the weight of each sample so that the sum of the weights of all samples is equal to 1;

Step three: classifying samples according to the defined weak classifier, and calculating a classification error rate;

step four: for classifiers with lower classification error rates, higher weights are given; for classifiers with higher error rates, lower weights are given;

Step five: updating the weight of each sample according to the alpha value obtained by calculation, wherein the error-divided sample weight gradually increases along with the increase of the iteration times, and the predicted correct sample weight decreases in the next iteration:

step six: randomly extracting a subset of the data set according to the corresponding sample weight, and performing subsequent classification learning processing;

step seven: the extracted subset is removed from the historical sample set of instances and the remaining portion can be modeled as an out-of-bag dataset.

S44, generating a decision tree model for the sample subset;

Wherein the generating a decision tree model for the subset of samples comprises the steps of:

Specifically, for the new dataset, predictions are made based on all decision trees (with respective weights). For example, in predicting problems based on landslide hazard, a prediction process involves using a set of historical data to predict the probability of landslide occurrence for each particular location, time period, and monitored parameter combination. The decision tree integration treats each parameter combination as a single sample and calculates a weighted probability average based on the classification results of all decision trees. And predicting samples in the test set, comparing the prediction result with an actual label, and calculating the accuracy and other performance indexes of the disaster early warning model. If the model is not effective, the model parameters may be re-optimized or features re-selected, and the process repeated until a more efficient model is obtained.

S5, acquiring a history instance related to landslide disasters, and combining a high slope landslide generation mechanism to formulate a corresponding slope state early warning threshold.

Specifically, the history examples include the following:

time and place of landslide hazard: the method comprises the basic information of landslide hazard occurrence time, place, scale and the like.

Monitoring data of landslide hazard precursors: the method comprises monitoring data such as terrain change, ground water level change, rainfall change and the like before landslide occurrence.

Landslide induced effects: including traffic jams, house collapse, casualties, etc. caused by landslide.

The cause and mechanism of landslide hazard: including the cause and mechanism of landslide occurrence, such as geologic structures, seismic activity, etc.

Landslide hazard prevention measures and effects: including the implementation process, effect evaluation, etc. of landslide control measures.

And S6, if the damage index score exceeds the slope state early warning threshold value, a corresponding warning signal instruction is sent out, and rescue measures are taken according to a preset early warning rule.

When the damage index score exceeds the slope state early warning threshold, a corresponding warning signal instruction is sent out, and rescue measures are taken according to a preset early warning rule, and the method comprises the following steps:

And S61, comparing the damage index score with a slope state early warning threshold value, and sending a corresponding alarm signal instruction to a management department if the damage index score exceeds the slope state early warning threshold value.

Specifically, the corresponding alarm signal instruction sent to the management department comprises the modes of audible and visual alarm, short message prompt, vehicle broadcasting and the like, and early warning information is sent to the traffic management department and drivers and passengers. This will enable traffic management and drivers to obtain early warning information in the first time, optimize road management control and vehicle scheduling, and ensure driving safety.

S62, starting an emergency plan, and dispatching a professional team to rescue according to the scene disaster condition.

The emergency plan comprises emergency equipment material preparation, personnel and property evacuation, emergency allocation plan and emergency instruction signaling.

Specifically, emergency equipment supplies are prepared: including emergency equipment, emergency medical equipment, communication equipment, protective equipment, etc. The devices should be configured and stored according to actual conditions so as to ensure on-site rescue and investigation work when an emergency occurs.

Evacuating people and property: when a high slope landslide disaster occurs, traffic safety and personnel life safety are ensured. Thus, emergency plans require evacuation plans to be formulated, while taking into account measures relating to property protection and temporary storage.

Emergency deployment planning: including personnel and material scheduling, resource integration, point protection, etc. The emergency plan needs to determine the allocation plan, command organization and responsibilities and action schemes of each member unit so as to organize coordination and quick response.

Emergency instruction signaling: namely, the establishment and the guarantee of an emergency communication system. The emergency plan needs to define the channel and mode of information transmission, including internal and external communication, disaster release, etc.

S63, constructing a risk assessment model based on landslide disasters, and assessing influence and diffusion degree caused by the disasters by using the risk assessment model.

The construction of the landslide disaster-based risk assessment model and the assessment of the influence and the diffusion degree caused by the occurrence of the disaster by using the risk assessment model comprises the following steps:

S632, determining the weight of each disaster evaluation index by using an AHP algorithm, and acquiring a relative weight matrix among the influence factors.

Specifically, the AHP (ANALYTIC HIERARCHY Process) analytic hierarchy Process is a calculation method based on personal judgment, is a decision problem analysis method and a comprehensive evaluation method, has qualitative and quantitative characteristics, and can be used for determining the relative importance among a plurality of factors.

Specifically, assuming that there are p factors, w1, w2, w3, w4, … …, wd, the weights of which are respectively denoted by t1, t2, t3, t4, … …, td, and the vector set is taken as a basis to obtain a relative weight matrix between the influencing factors, and the expression is as follows:

S64, control and management of the highway high slope area are enhanced.

Specifically, constructing a three-dimensional perspective view of a highway high side slope region and providing visualization includes the steps of:

Collecting data: and collecting related data of the expressway high-slope area, including DEM (digital elevation model), geological information, slope vegetation, risk assessment data and the like.

Establishing a three-dimensional model: and establishing a three-dimensional model of the expressway high-side slope area according to the DEM data and the corresponding functional area division by using modeling software. Meanwhile, high slope landslide risk point labels are added, and objects such as slopes, rock mass, communication towers and the like are marked.

Rendering a model: and adding details such as illumination, texture and the like into the three-dimensional model by using a three-dimensional rendering technology so as to improve the visual effect, and adjusting and amplifying and highlighting the scheme item by adding the related effect. And finally forming a complete three-dimensional perspective view.

The visual display can adopt various modes, the three-dimensional image is uploaded to a platform for display, or the AR/VR and other augmented reality technologies are used, and the user interactivity and the dynamic effect are added to the model.

In summary, by means of the technical scheme, the acquired real-time monitoring data and the processed high-resolution image are integrated, and the integrated data sources are subjected to space-time scale fusion, so that valuable data can be widely collected on time and space levels, further, the change rule and unknown information of the expressway high slope area can be comprehensively and rapidly analyzed and researched, hidden danger points in the expressway high slope area are identified and classified, and the acquired remote sensing image is geometrically corrected, so that the image quality is improved, the identification precision is improved, and the deviation of an early warning result is avoided.

According to the invention, through constructing the disaster early warning model based on evaluation function training, various sensor data, remote sensing image information, geological data and the like can be input into the model in the model training stage, a proper disaster risk evaluation index is formed according to data characteristics and related rules, the set index coefficient rule is learned, and the optimal detection and early warning discrimination parameters are obtained, so that not only can cause analysis be more accurately carried out, but also quantitative treatment can be carried out on the prediction mechanism flow, and through the landslide disaster early warning model trained by the evaluation function, the possible landslide risk can be rapidly and accurately found and early warned, the practical guarantee is provided for the manager to formulate a more accurate coping scheme, and the traffic safety can be improved.

According to the invention, the risk assessment model based on landslide disasters is constructed, and the influence and the diffusion degree caused by the disasters are assessed by using the risk assessment model, so that the influence range and the diffusion degree after the disasters occur can be comprehensively assessed, and the influence and the diffusion degree thereof which are possibly caused by the landslide disasters of the high slope can be analyzed and explored, so that early warning personnel can refine design and implement precautionary measures according to the prediction result of the risk assessment model, and meanwhile, the method plays a certain guiding and promoting role in high-quality, high-efficiency and safe construction of highways.

The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.

Claims

1. The highway high slope landslide hazard early warning method based on evaluation function training is characterized by comprising the following steps of: