WO2020151226A1 - 一种基于训练模型的滑坡预测方法及装置 - Google Patents

一种基于训练模型的滑坡预测方法及装置 Download PDF

Info

Publication number
WO2020151226A1
WO2020151226A1 PCT/CN2019/102982 CN2019102982W WO2020151226A1 WO 2020151226 A1 WO2020151226 A1 WO 2020151226A1 CN 2019102982 W CN2019102982 W CN 2019102982W WO 2020151226 A1 WO2020151226 A1 WO 2020151226A1
Authority
WO
WIPO (PCT)
Prior art keywords
landslide
random forest
model
perspective
view
Prior art date
Application number
PCT/CN2019/102982
Other languages
English (en)
French (fr)
Inventor
李岩山
周李
夏荣杰
刘瑜
王海鹏
谢维信
Original Assignee
深圳大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳大学 filed Critical 深圳大学
Publication of WO2020151226A1 publication Critical patent/WO2020151226A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation

Definitions

  • the invention relates to the technical field of landslide prediction, in particular to a landslide prediction method and device based on a training model.
  • Landslides are a common geological disaster that occurs in the natural world. They are very harmful and often cause heavy losses to people's lives and properties. As we all know, the prediction and forecast of landslides is not the result of subjective guesswork. It needs to be based on real-time deformation monitoring of landslides. At the same time, it is necessary to carry out accurate analysis and accurate judgment with the help of computer systems in accordance with the principles and methods of system engineering. Reasonable prediction results.
  • the current methods of landslide prediction mainly include engineering condition analysis, site condition analysis, rock and soil mechanics experiment, and mechanical model.
  • the current landslide prediction methods usually first carry out a detailed off-the-shelf engineering geological survey, establish a geological model, and then take samples to conduct mechanical tests in the laboratory or on-site, further establish the mechanical model, and then perform mechanical analysis, and finally obtain the landslide motion state data. It can be seen that the current landslide prediction methods not only have low automation and low efficiency, but also may have low prediction accuracy due to subjective factors. Therefore, how to improve the prediction efficiency and accuracy of the landslide motion state is a problem that needs to be solved urgently.
  • the technical problem to be solved by the present invention is to provide a landslide prediction method and device based on a training model, which can analyze landslide changes through a multi-view weight random forest model, which can not only improve the prediction efficiency and prediction accuracy of the landslide motion state, but also It can provide a quantitative evaluation basis for later analysis and evaluation of landslide stability, prediction and early warning of landslides and later prevention and control.
  • the first aspect of the embodiments of the present invention discloses a landslide prediction method, which includes:
  • the landslide training data includes at least one of sensor displacement training data, marker motion trajectory training data, and crack size training data ;
  • the multi-view weight random forest model is used to perform model evaluation on each of the landslide observation angles to obtain model evaluation results, and the model evaluation results are merged to obtain landslide warning classification results.
  • the use of a random forest model to construct a multi-view weight random forest model corresponding to all the landslide warning classification perspectives includes:
  • the perspective weights corresponding to the landslide warning classification perspectives based on the model scores are constructed a priori with each of the landslide training
  • the weight of the landslide warning perspective corresponding to the data includes:
  • the landslide warning perspective weight corresponding to the landslide training data is constructed prior to the Bayesian framework.
  • the calculation formula of the landslide viewing angle weight is:
  • ⁇ (i) represents the perspective index of the landslide warning classification perspective
  • W ⁇ (i) represents the landslide warning perspective weight
  • P i represents the perspective weight prior
  • Pvi represents the model score .
  • P i ⁇ P vi represents the posterior probability of the accuracy of the landslide warning classification perspective.
  • a random forest-based landslide prediction device in the second aspect of the embodiments of the present invention, includes:
  • An acquisition module for collecting multiple types of landslide training data where the landslide training data includes at least one of sensor displacement training data, marker movement trajectory training data, and fracture size training data;
  • the first construction module is used to separately construct the landslide warning classification perspective of the multiple types of landslide training data
  • the second construction module is configured to use the random forest model to construct a multi-view weight random forest model corresponding to all the landslide warning classification views;
  • the collection module is also used to collect multiple types of landslide test data
  • the first construction module is also used to separately construct the landslide observation angle of view of the multiple types of landslide test data;
  • An evaluation module configured to use the multi-view weight random forest model to perform model evaluation on each of the landslide observation views to obtain a model evaluation result
  • the fusion module is used to fuse the model evaluation results to obtain landslide warning classification results.
  • the second building module includes a learning sub-module and a building sub-module, wherein:
  • the learning sub-module is configured to use a random forest model to separately learn all the landslide warning classification perspectives to obtain a random forest model of all perspectives and a random forest model corresponding to each of the random forest models of all perspectives Model score
  • the construction sub-module is configured to a priori construct the landslide warning view weight corresponding to the landslide training data based on the model score and the view angle weight corresponding to the landslide warning classification view;
  • the construction sub-module is also used to construct a multi-view weight random forest model based on all the landslide warning view angle weights and all the view angle random forest models.
  • the construction sub-module constructs a priori construction and the landslide based on the model score and the view weight corresponding to the landslide warning classification view.
  • the method of weighting the landslide warning view angle corresponding to the training data is as follows:
  • the landslide warning perspective weight corresponding to the landslide training data is constructed prior to the Bayesian framework.
  • ⁇ (i) represents the perspective index of the landslide warning classification perspective
  • W ⁇ (i) represents the landslide warning perspective weight
  • P i represents the perspective weight prior
  • Pvi represents the model score.
  • P i ⁇ P vi represents the posterior probability of the accuracy of the landslide warning classification perspective.
  • the third aspect of the embodiment of the present invention discloses another landslide prediction device, the device includes:
  • a memory storing executable program codes
  • a processor coupled with the memory
  • the processor calls the executable program code stored in the memory to execute the landslide prediction method based on the random forest model disclosed in the first aspect of the embodiment of the present invention.
  • the fourth aspect of the embodiments of the present invention discloses a computer storage medium, the computer storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the random forest model-based Landslide prediction method.
  • the fifth aspect of the embodiments of the present invention discloses a computer program product, which when the computer program product runs on a computer, causes the computer to execute the random forest model-based landslide prediction method disclosed in the first aspect of the embodiments of the present invention.
  • the present invention has the following beneficial effects:
  • multiple types of landslide training data are collected, and a landslide warning classification perspective for each type of landslide training data is separately constructed; a random forest model is used to construct a multi-view weight random forest model for all landslide warning classification perspectives; multiple types of landslide tests are collected Landslide observation perspectives for each type of landslide test data; use the multi-perspective weighted random forest model to evaluate each landslide observation perspective to obtain model evaluation results, and integrate the model evaluation results to obtain landslide warning Classification results.
  • the implementation of the present invention can analyze the landslide changes through the multi-view weighted random forest model, not only can improve the prediction efficiency and prediction accuracy of the landslide movement state, but also can analyze and evaluate the stability of the landslide in the later stage, predict the early warning of the landslide and the later prevention and control Work provides quantitative evaluation basis.
  • Fig. 1 is a schematic flowchart of a landslide prediction method disclosed in an embodiment of the present invention
  • Figure 2 is a schematic structural diagram of a landslide prediction device disclosed in an embodiment of the present invention.
  • Figure 3 is a schematic structural diagram of another landslide prediction device disclosed in an embodiment of the present invention.
  • Fig. 4 is a schematic structural diagram of another landslide prediction device disclosed in an embodiment of the present invention.
  • the embodiment of the present invention discloses a landslide prediction method and device based on a training model, which can analyze landslide changes through a multi-view weighted random forest model, which can not only improve the prediction efficiency and prediction accuracy of the landslide motion state, but also can perform later Landslide stability analysis and evaluation, prediction and early warning of landslides and later prevention and control work provide a quantitative evaluation basis. Detailed descriptions are given below.
  • FIG. 1 is a schematic flowchart of a landslide prediction method disclosed in an embodiment of the present invention.
  • the landslide prediction method based on the random forest model described in FIG. 1 can be applied to a terminal device that monitors the movement state of a landslide, which is not limited in the embodiment of the present invention.
  • the landslide prediction method based on the random forest model may include the following steps:
  • the aforementioned landslide training data includes sensor displacement training data, marker movement trajectory training data, and crack size training data, which are not limited in the embodiment of the present invention.
  • step 101 may include:
  • t k represents the k-th moment
  • n represents that a total of n time units are monitored.
  • the three-dimensional spatial data (x, y, z) output by each sensor at each time constitutes the sensor displacement training data, and then the sensor displacement landslide warning classification perspective is constructed based on the sensor displacement training data.
  • step 101 may include:
  • the movement trajectory of the marker contains data of three dimensions (x, y, z) at each moment.
  • the set B i (t 1 ), B i (t 2 ), B i (t 3 ),..., B i (t n ), called the time series of the trajectory data of the landslide markers, is expressed as:
  • t k represents the k-th moment
  • n represents that a total of n time units are monitored.
  • the three-dimensional (x, y, z) motion trajectory data output by each marker at each time constitutes the marker motion trajectory training data, and then based on the marker motion trajectory training data, the landslide warning classification perspective of the marker motion trajectory is constructed.
  • step 101 may include:
  • the crack on the space position si is a series of time t 1 , t 2 ,..., t n (t is the time independent variable and t 1 ⁇ t 2 ⁇ ... ⁇ t n ).
  • C i (t 1 ),C i (t 2 ),C i (t 3 ),...,C i (t n ) called the time series of the fracture size data of the landslide body, the training data time of the fracture size of the landslide body
  • the calculation formula of the sequence is:
  • t k represents the k-th moment
  • n represents that a total of n time units are monitored.
  • using the random forest model to construct a multi-view weight random forest model corresponding to all landslide warning classification perspectives may include:
  • the landslide perspective weights corresponding to the landslide training data are constructed a priori;
  • the random forest classification model For example, after determining the random forest classification model for each perspective, according to the criterion of the smallest combination error, find the combination of all perspective random forest classification models with the smallest total error, and the random forest classification model based on all perspectives with the smallest error
  • the obtained multi-view weight random forest model is used as the optimal multi-view weight random forest model, and the calculation formulas for the optimal random forest classification model and the landslide warning classification view weight are:
  • N represents the number of training samples of each landslide training data
  • x ij represents the characteristics of the j-th sample of the i-th view
  • y j represents the landslide mark
  • W ⁇ (i) represents the perspective weight of the landslide warning classification perspective
  • P i ⁇ P vi represents the posterior probability of the accuracy of the landslide warning classification perspective.
  • a priori the landslide view weight corresponding to each landslide training data based on each model score and the view angle weight corresponding to each landslide warning classification view including :
  • the landslide perspective weight corresponding to the above landslide training data is constructed under the Bayesian framework.
  • the aforementioned landslide test data includes sensor displacement test data, marker movement trajectory test data, and crack size test data, which is not limited in the embodiment of the present invention.
  • the calculation formula for the above-mentioned landslide warning classification result is:
  • x ij represents the characteristics of the j-th sample of the i-th viewing angle
  • the forest model weights the coefficient on the test sample, and the weight coefficient is determined by the sample action coefficient I ij and the landslide warning classification perspective weight W ⁇ (i) , and the calculation formula of the sample action coefficient I ij is:
  • the weight coefficient of the test sample is equal to the landslide warning classification perspective weight W ⁇ (i)
  • the weight coefficient of the test sample is also 0, that is, the perspective random forest model may not be constructed at this time (for example: the perspective random forest model of the crack size is not constructed) or the perspective random
  • the value of the forest model is 0 (for example, the value of the perspective random forest model for constructing the crack size is 0), which is not limited in the embodiment of the present invention.
  • the foregoing landslide warning classification result includes at least one of a 0-level white warning, a first-level blue warning, a second-level yellow warning, and a third-level red warning.
  • the embodiment of the present invention does not limit it, and the higher the level is Indicates the greater the probability of landslide occurrence.
  • the landslide prediction method based on the random forest model may further include the following steps:
  • the emergency plan is to increase the frequency of observation; when the above-mentioned landslide warning classification result is a second-level yellow warning, the emergency plan is to strengthen prediction and forecasting; when the above-mentioned landslide warning classification result is Three-level red warning, the emergency plan is to send an emergency notice to the competent unit, so that the staff of the competent unit can issue early warning information for organizing the evacuation of relevant personnel on the spot.
  • this optional embodiment can ensure the safety of people's lives and property in time by formulating different emergency plans for different landslide warning classification results.
  • the implementation of the landslide prediction method based on the random forest model described in Figure 1 can analyze the landslide changes through the multi-view weight random forest model, which can not only improve the prediction efficiency and prediction accuracy of the landslide movement state, but also stabilize the landslide in the later stage. It provides quantitative evaluation basis for analysis and evaluation, forecasting early warning of landslides and post-control work.
  • it can also monitor the spatial distribution characteristics of landslide deformation in an all-round way, and analyze and determine the overall deformation trend and sliding direction of the landslide from the monitoring information; it can also formulate different emergency plans for different landslide warning classification results, and timely It protects the safety of people's lives and property; it can also identify the trend and development of landslides; it can also analyze the scale and formation mechanism of landslides, predict the development trend of landslides, and provide guidance for subsequent landslide treatment.
  • FIG. 2 is a schematic structural diagram of a landslide prediction device disclosed in an embodiment of the present invention.
  • the landslide prediction device based on the random forest model described in FIG. 2 is a terminal device for monitoring the motion state of the landslide, which is not limited in the embodiment of the present invention.
  • the landslide prediction device based on the random forest model includes an acquisition module 401, a first construction module 402, a second construction module 403, an evaluation module 404, and a fusion module 404, wherein:
  • the collection module 401 is used to collect multiple types of landslide training data.
  • the aforementioned landslide training data includes at least one of sensor displacement training data, marker motion trajectory training data, and fracture size training data, which is not limited in the embodiment of the present invention.
  • the first construction module 402 is used to separately construct the landslide warning classification perspectives of the above-mentioned multiple types of landslide training data.
  • the second construction module 403 is configured to use the random forest model to construct a multi-view weight random forest model corresponding to all the aforementioned landslide warning classification views.
  • the collection module 401 is also used to collect multiple types of landslide test data.
  • the first construction module 402 is also used to separately construct the landslide observation angle of view of the aforementioned multiple types of landslide test data.
  • the evaluation module 404 is configured to use the above-mentioned multi-view weighted random forest model to separately evaluate each landslide observation angle to obtain a model evaluation result.
  • the fusion module 405 is used to fuse the above model evaluation results to obtain the landslide warning classification result.
  • the implementation of the landslide prediction device based on the random forest model described in Figure 2 analyzes the landslide changes through the multi-view weighted random forest model, which can not only improve the prediction efficiency and prediction accuracy of the landslide motion state, but also provide landslide stability for the later stage. Analysis and evaluation, prediction and early warning of landslides and subsequent prevention and control work to provide quantitative evaluation basis.
  • the second construction module 403 may include a learning sub-module 4031 and a construction sub-module 4032.
  • the landslide prediction device based on the random forest model is shown in Figure 3.
  • Figure 3 is another landslide prediction device, in which:
  • the learning sub-module 4031 is configured to use the random forest model to separately learn all the above-mentioned landslide warning classification perspectives, to obtain all the perspective random forest models and the model scores corresponding to each perspective random forest model in the above-mentioned all perspective random forest models.
  • the construction sub-module 4032 is used to construct a priori the landslide warning view weight corresponding to the landslide training data based on the model score and the view weight corresponding to the landslide warning classification view.
  • the construction sub-module 4032 is also used to construct a multi-view weight random forest model based on all the aforementioned landslide warning view weights and all view random forest models.
  • the construction sub-module 4032 constructs a priori the landslide warning view corresponding to the above landslide training data based on the aforementioned model score and the view weight corresponding to the aforementioned landslide warning classification view.
  • the specific weighting method is:
  • the landslide warning perspective weights corresponding to the above landslide training data are constructed under the Bayesian framework.
  • ⁇ (i) represents the perspective index of the above-mentioned landslide warning classification perspective
  • W ⁇ (i) represents the above-mentioned landslide warning perspective weight
  • P i represents the above-mentioned perspective weight prior
  • P vi represents the above-mentioned model score.
  • P i ⁇ P vi represents the posterior probability of the accuracy of the landslide warning classification perspective.
  • FIG. 4 is a schematic structural diagram of another landslide prediction device disclosed in an embodiment of the present invention.
  • the landslide prediction device based on the random forest model may include:
  • a memory 401 storing executable program codes
  • a processor 402 coupled to the memory 401;
  • the processor 402 calls the executable program code stored in the memory 401 to execute the steps in the landslide prediction method based on the random forest model described in the first embodiment.
  • the embodiment of the present invention discloses a computer-readable storage medium that stores a computer program for electronic data exchange, wherein the computer program causes the computer to execute the method for landslide prediction based on the random forest model described in the first embodiment step.
  • the embodiment of the present invention discloses a computer program product.
  • the computer program product includes a non-transitory computer-readable storage medium storing the computer program, and the computer program is operable to cause the computer to execute the random-based operation described in the first embodiment. Steps in the landslide prediction method of the forest model.
  • the device embodiments described above are only illustrative.
  • the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in One place, or it can be distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the objectives of the solutions of the embodiments. Those of ordinary skill in the art can understand and implement without creative work.
  • each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be achieved by hardware.
  • the computer software product can be stored in a computer-readable storage medium, which includes a read-only memory.
  • Read-Only Memory ROM
  • RAM Random Access Memory
  • PROM Programmable Read-only Memory
  • EPROM Erasable Programmable Read-Only Memory
  • OTPROM One-time Programmable Read-Only Memory
  • EEPROM Electronically-Erasable Programmable Read-Only Memory
  • CD-ROM Compact Disc Read -Only Memory
  • CD-ROM Compact Disc Read -Only Memory

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Evolutionary Computation (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Testing Or Calibration Of Command Recording Devices (AREA)
  • Pit Excavations, Shoring, Fill Or Stabilisation Of Slopes (AREA)

Abstract

一种基于训练模型的滑坡预测方法及装置，该方法包括采集多类滑坡训练数据，并分别构建针对每类滑坡训练数据的滑坡预警分类视角（101）；使用随机森林模型构建针对所有滑坡预警分类视角的多视角权重随机森林模型（102）；采集多类滑坡测试数据，并分别构建针对每类滑坡测试数据的滑坡观测视角（103）；使用该多视角权重随机森林模型分别对每个滑坡观测视角进行模型评估，得到模型评估结果，并融合该模型评估结果，得到滑坡预警分类结果（104）。该方法能够通过多视角权重随机森林模型分析滑坡变动情况，不仅能够提高滑坡运动状态的预测效率和预测精准度，还能够为后期进行滑坡稳定性分析与评价，预测预警滑坡及后期防治工作提供定量化的评价依据。

Description

一种基于训练模型的滑坡预测方法及装置 技术领域
本发明涉及滑坡预测技术领域，尤其涉及一种基于训练模型的滑坡预测方法及装置。
背景技术
滑坡是发生在自然界中的一种常见的地质灾害，其危害巨大，经常对人们的生命和财产造成重大的损失。众所周知，对滑坡的预测预报并不是主观臆断的猜测结果，需建立在滑坡实时变形监测的基础上，同时必须按照系统工程的原理和方法，借助计算机系统来进行精确分析和准确判断，最终得出合理的预测结果。目前滑坡预测的方法主要包括工程条件分析、地址条件分析、岩土体力学实验、力学模型。然而，实践发现，目前的滑坡预测方法通常先进行详细的现成工程地质调查，建立地质模型，然后取样在实验室或现场进行力学试验，进一步建立力学模型，再进行力学分析，最后得到滑坡运动状态数据。可见，目前的滑坡预测方法不仅自动化程度低，效率低，而且还可能由于主观因素导致预测精度低。因此，如何提高滑坡运动状态的预测效率和预测精准度是当下急需解决的问题。
发明内容
本发明所要解决的技术问题在于，提供一种基于训练模型的滑坡预测方法及装置，能够通过多视角权重随机森林模型分析滑坡变动情况，不仅能够提高滑坡运动状态的预测效率和预测精准度，还能够为后期进行滑坡稳定性分析与评价，预测预警滑坡及后期防治工作提供定量化的评价依据。
为了解决上述技术问题，本发明实施例第一方面公开了一种滑坡预测方法，所述方法包括：
采集多类滑坡训练数据，并分别构建所述多类滑坡训练数据的滑坡预警分类视角,所述滑坡训练数据包括传感器位移训练数据、标志物运动轨迹训练数据、裂缝尺寸训练数据中的至少一种；
使用随机森林模型构建与所有所述滑坡预警分类视角相对应的多视角权重 随机森林模型；
采集多类滑坡测试数据，并分别构建所述多类滑坡测试数据的滑坡观测视角；
使用所述多视角权重随机森林模型分别对每个所述滑坡观测视角进行模型评估，得到模型评估结果，并融合所述模型评估结果，得到滑坡预警分类结果。
作为一种可选的实施方式，在本发明实施例第一方面中，所述使用随机森林模型构建与所有所述滑坡预警分类视角相对应的多视角权重随机森林模型，包括：
使用随机森林模型分别学习所有所述滑坡预警分类视角，得到所有视角随机森林模型以及与所述所有视角随机森林模型中的每个所述视角随机森林模型相对应的模型评分；
分别基于所述模型评分、与所述滑坡预警分类视角相对应的视角权重先验构建与所述滑坡训练数据相对应的滑坡预警视角权重；
基于所有所述滑坡预警视角权重和所有所述视角随机森林模型构建多视角权重随机森林模型。
作为一种可选的实施方式，在本发明实施例第一方面中，所述分别基于所述模型评分、与所述滑坡预警分类视角相对应的视角权重先验构建与每个所述滑坡训练数据相对应的滑坡预警视角权重，包括：
分别基于所述模型评分、与所述滑坡预警分类视角相对应的视角权重先验在贝叶斯框架下构建与所述滑坡训练数据相对应的滑坡预警视角权重。
作为一种可选的实施方式，在本发明实施例第一方面中，其特征在于，所述滑坡视角权重的计算公式为：
Figure PCTCN2019102982-appb-000001
式中：
θ (i)表示所述滑坡预警分类视角的视角索引，W θ(i)表示所述滑坡预警视角权重，P i表示所述视角权重先验，P vi表示所述模型评分 。在所述贝叶斯框架下，P i×P vi表示所述滑坡预警分类视角的准确度的后验概率。
本发明实施例第二方面公开了一种基于随机森林的滑坡预测装置，所述装置包括：
采集模块，用于采集多类滑坡训练数据,所述滑坡训练数据包括传感器位移训练数据、标志物运动轨迹训练数据、裂缝尺寸训练数据中的至少一种；
第一构建模块，用于分别构建所述多类滑坡训练数据的滑坡预警分类视角；
第二构建模块，用于使用随机森林模型构建与所有所述滑坡预警分类视角相对应的多视角权重随机森林模型；
所述采集模块，还用于采集多类滑坡测试数据；
所述第一构建模块，还用于分别构建所述多类滑坡测试数据的滑坡观测视角；
评估模块，用于使用所述多视角权重随机森林模型分别对每个所述滑坡观测视角进行模型评估，得到模型评估结果；
融合模块，用于融合所述模型评估结果，得到滑坡预警分类结果。
作为一种可选的实施方式，在本发明实施例第二方面中，所述第二构建模块包括学习子模块以及构建子模块，其中：
所述学习子模块，用于使用随机森林模型分别学习所有所述滑坡预警分类视角，得到所有视角随机森林模型以及与所述所有视角随机森林模型中的每个所述视角随机森林模型相对应的模型评分；
所述构建子模块，用于分别基于所述模型评分、与所述滑坡预警分类视角相对应的视角权重先验构建与所述滑坡训练数据相对应的滑坡预警视角权重；
所述构建子模块，还用于基于所有所述滑坡预警视角权重和所有所述视角随机森林模型构建多视角权重随机森林模型。
作为一种可选的实施方式，在本发明实施例第二方面中，所述构建子模块分别基于所述模型评分、与所述滑坡预警分类视角相对应的视角权重先验构建与所述滑坡训练数据相对应的滑坡预警视角权重的方式具体为：
分别基于所述模型评分、与所述滑坡预警分类视角相对应的视角权重先验在贝叶斯框架下构建与所述滑坡训练数据相对应的滑坡预警视角权重。
所述滑坡预警视角权重的计算公式为：
Figure PCTCN2019102982-appb-000002
式中：
θ (i)表示所述滑坡预警分类视角的视角索引，W θ(i)表示所述滑坡预警视角权重， P i表示所述视角权重先验，P vi表示所述模型评分。在所述贝叶斯框架下，P i×P vi表示所述滑坡预警分类视角的准确度的后验概率。
本发明实施例第三方面公开了另一种滑坡预测装置，所述装置包括：
存储有可执行程序代码的存储器；
与所述存储器耦合的处理器；
所述处理器调用所述存储器中存储的所述可执行程序代码，执行本发明实施例第一方面公开的基于随机森林模型的滑坡预测方法。
本发明实施例第四方面公开了一种计算机存储介质，所述计算机存储介质存储有计算机指令，所述计算机指令被调用时，用于执行本发明实施例第一方面公开的基于随机森林模型的滑坡预测方法。
本发明实施例第五方面公开了一种计算机程序产品，当所述计算机程序产品在计算机上运行时，使得所述计算机执行本发明实施例第一方面公开的基于随机森林模型的滑坡预测方法。
与现有技术相比，本发明具有以下有益效果：
本发明中，采集多类滑坡训练数据，并分别构建针对每类滑坡训练数据的滑坡预警分类视角；使用随机森林模型构建针对所有滑坡预警分类视角的多视角权重随机森林模型；采集多类滑坡测试数据，并分别构建针对每类滑坡测试数据的滑坡观测视角；使用该多视角权重随机森林模型分别对每个滑坡观测视角进行模型评估，得到模型评估结果，并融合该模型评估结果，得到滑坡预警分类结果。可见，实施本发明能够通过多视角权重随机森林模型分析滑坡变动情况，不仅能够提高滑坡运动状态的预测效率和预测精准度，还能够为后期进行滑坡稳定性分析与评价，预测预警滑坡及后期防治工作提供定量化的评价依据。
附图说明
为了更清楚地说明本发明实施例中的技术方案，下面将对实施例中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1是本发明实施例公开的一种滑坡预测方法的流程示意图；
图2是本发明实施例公开的一种滑坡预测装置的结构示意图；
图3是本发明实施例公开的另一种滑坡预测装置的结构示意图；
图4是本发明实施例公开的又一种滑坡预测装置的结构示意图。
具体实施方式
为了更好地理解和实施，下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
本发明实施例的术语“包括”和“具有”以及他们的任何变形，意图在于覆盖不排他的包含，例如，包含了一系列步骤或模块的过程、方法、系统、产品或设备不必限于清楚地列出的那些步骤或模块，而是可包括没有清楚地列出的或对于这些过程、方法、产品或设备固有的其它步骤或模块。此外，本发明的说明书和权利要求书中的术语“第一”、“第二”等仅是用于区别不同对象，对于本领域的普通技术人员而言，可以根据具体情况理解上述术语在本发明中的具体含义。
本发明实施例公开了一种基于训练模型的滑坡预测方法及装置，能够通过多视角权重随机森林模型分析滑坡变动情况，不仅能够提高滑坡运动状态的预测效率和预测精准度，还能够为后期进行滑坡稳定性分析与评价，预测预警滑坡及后期防治工作提供定量化的评价依据。以下分别进行详细说明。
实施例一
请参阅图1，图1是本发明实施例公开的一种滑坡预测方法的流程示意图。其中，图1所描述的基于随机森林模型的滑坡预测方法可以应用在监测山体滑坡运动状态的终端设备中，本发明实施例不做限定。如图1所示，该基于随机森林模型的滑坡预测方法可以包括以下步骤：
101、采集多类滑坡训练数据，并分别构建该多类滑坡训练数据的滑坡预警分类视角。
本发明实施例中，上述滑坡训练数据包括传感器位移训练数据、标志物运动轨迹训练数据、裂缝尺寸训练数据，本发明实施例不做限定。
举例来说，当上述滑坡训练数据为传感器位移训练数据时，步骤101可以 包括：
采集滑坡坡体不同部位的传感器监测信息，并形成滑坡坡体不同部位的传感监测时间序列。设空间位置s i上的传感器在一系列时刻t 1,t 2,...,t n(t为时间自变量且t 1＜t 2＜...＜t n)得到的传感序列集合A i(t 1),A i(t 2),...,A i(t n)，称为滑坡体传感器监测数据时间序列，并且该滑坡体传感器监测数据时间序列的计算公式为：
A si(t)＝{A si(t k),k＝1,2,3,...,n}
其中，t k表示第k个时刻，n表示总共监测了n个时间单位。
各传感器在每个时刻输出的三维空间数据(x,y,z)构成传感器位移训练数据，再基于该传感器位移训练数据构建传感器位移滑坡预警分类视角。
这样通过检测滑坡体传感器位移的变化情况，不仅能够全方位地监测滑坡变形在空间上的分布特征，还能够从监测信息中分析和判定滑坡总体变形趋势和滑动方向。
举例来说，当该滑坡训练数据为标志物运动轨迹训练数据时，步骤101可以包括：
标志物的运动轨迹每时刻包含三个维度(x,y,z)的数据。设空间位置s i上的标志物在一系列时刻t 1,t 2,...,t n(t为时间自变量且t 1＜t 2＜...＜t n)得到的运动轨迹序列集合B i(t 1),B i(t 2),B i(t 3),...,B i(t n)，称为滑坡体标志物运动轨迹数据时间序列，表示为：
B si(t)＝{B si(t k),k＝1,2,3,...,n}
其中，t k表示第k个时刻，n表示总共监测了n个时间单位。
各标志物在每个时刻输出的三维(x,y,z)运动轨迹数据构成标志物运动轨迹训练数据，再基于该标志物运动轨迹训练数据构建标志物运动轨迹滑坡预警分类视角。
这样通过监测滑坡体的标志物运动轨迹，能够识别滑坡运动趋势、发展状况。
举例来说，当该滑坡训练数据为裂缝尺寸训练数据时，步骤101可以包括：
设空间位置s i上的裂缝在一系列时刻t 1,t 2,...,t n(t为时间自变量且t 1＜t 2＜...＜t n)得到的尺寸大小序列集合C i(t 1),C i(t 2),C i(t 3),...,C i(t n)，称为滑坡体裂缝尺寸数据时间序列，该滑坡体裂缝尺寸训练数据时间序列的计算公式为：
C si(t)＝{C si(t k),k＝1,2,3,...,n}
其中，t k表示第k个时刻，n表示总共监测了n个时间单位。
计算各个空间位置每时刻输出的裂缝大小数据构成裂缝尺寸训练数据，再基于该裂缝尺寸训练数据构建裂缝尺寸滑坡预警分类视角。
这样通过监测不同尺寸的滑坡裂缝，能够对滑坡的规模、形成机制进行分析，预测滑坡发展趋势，为后续滑坡治理提供指导依据。
102、使用随机森林模型构建与所有滑坡预警分类视角相对应的多视角权重随机森林模型。
本发明实施例中，作为一种可选的实施方式，使用随机森林模型构建针与所有滑坡预警分类视角相对应的多视角权重随机森林模型，可以包括：
使用随机森林模型分别学习所有滑坡预警分类视角，得到所有视角随机森林模型以及与所述所有视角随机森林模型中的每个视角随机森林模型相对应的模型评分；
分别基于上述模型评分、与上述滑坡预警分类视角相对应的视角权重先验构建与滑坡训练数据相对应的滑坡视角权重；
基于所有滑坡视角权重和所有视角随机森林模型构建多视角权重随机森林模型。
举例来说，在确定了每个视角随机森林分类模型之后，按照组合误差最小的准则，找出总误差最小的所有视角随机森林分类模型组合，以及将基于该误差最小的所有视角随机森林分类模型得到的多视角权重随机森林模型作为最优多视角权重随机森林模型，以及该最优随机森林分类模型和滑坡预警分类视角权重的计算公式分别为：
Figure PCTCN2019102982-appb-000003
Figure PCTCN2019102982-appb-000004
式中：N表示每个滑坡训练数据的训练样本个数，x ij表示第i个视角第j个样本的特征，y j表示滑坡标记，θ(i)(i＝1,2,3)表示滑坡预警分类视角的视角索引，H θ(i)(i＝1,2,3)表示视角随机森林模型，W θ(i)表示滑坡预警分类视角权重，P i(i＝1,2,3)表示视角权重先验，P vi(i＝1,2,3)表示模型评分。在贝叶斯框架下，P i×P vi表示该滑坡预警分类视角的准确度的后验概率。
在该可选的实施方式中，进一步可选的，分别基于每个模型评分、与每个滑坡预警分类视角相对应的视角权重先验构建与每个滑坡训练数据相对应的滑坡视角权重，包括：
分别基于上述模型评分、与上述滑坡预警分类视角相对应的视角权重先验在贝叶斯框架下构建与上述滑坡训练数据相对应的滑坡视角权重。
103、采集多类滑坡测试数据，并分别构建该多类滑坡测试数据的滑坡观测视角。
本发明实施例中，上述滑坡测试数据包括传感器位移测试数据、标志物运动轨迹测试数据、裂缝尺寸测试数据，本发明实施例不做限定。
104、使用上述多视角权重随机森林模型分别对每个滑坡观测视角进行模型评估，得到模型评估结果，并融合该模型评估结果，得到滑坡预警分类结果。
本发明实施例中，上述滑坡预警分类结果的计算公式为：
Figure PCTCN2019102982-appb-000005
式中x ij表示第i个视角第j个样本的特征，H θ(i)(i＝1,2,3)表示视角随机森林模型，r ij(i＝1,2,3)表示视角随机森林模型在测试样本上权重系数，并且该权重系数由样本作用系数I ij和滑坡预警分类视角权重W θ(i)决定，以及该样本作用系数I ij的计算公式为：
r 1j＝I ijW θ1
r 2j＝I ijW θ2
r 3j＝I ijW θ3
Figure PCTCN2019102982-appb-000006
由该公式可知，当视角随机森林模型H θ(i)(x ij)的值大于0时，测试样本的权重系数等于滑坡预警分类视角权重W θ(i)，当视角随机森林模型H θ(i)(x ij)的值等于0时，测试样本的权重系数也为0，即此时可能没有构建该视角随机森林模型(例如：没有构建裂缝尺寸的视角随机森林模型)或者构建该视角随机森林模型的值为0(例如：构建裂缝尺寸的视角随机森林模型的值为0)，本发明实施例不做限定。
本发明实施例中，上述滑坡预警分类结果包括0级白色预警、一级蓝色预警、二级黄色预警、三级红色预警中的至少一种，本发明实施例不做限定，并 且等级越高表示滑坡发生的概率就越大。
作为一个可选的实施例，执行完步骤104之后，该基于随机森林模型的滑坡预测方法还可以包括以下步骤：
制定与上述滑坡预警分类结果相对应应急方案。
举例来说，当上述滑坡预警分类结果为一级蓝色预警，应急方案为增加观测频次；当上述滑坡预警分类结果为二级黄色预警，应急方案为加强预测预报；当上述滑坡预警分类结果为三级红色预警，应急方案为向主管单位发送紧急通知，以使得该主管单位的工作人员发布及时组织有关人员撤离现场预警信息。
可见，该可选的实施例通过针对不同的滑坡预警分类结果制定不同的应急方案，能够及时在保障人们生命、财产安全。
可见，实施图1所描述的基于随机森林模型的滑坡预测方法能够通过多视角权重随机森林模型分析滑坡变动情况，不仅能够提高滑坡运动状态的预测效率和预测精准度，还能够为后期进行滑坡稳定性分析与评价，预测预警滑坡及后期防治工作提供定量化的评价依据。此外，还能够全方位地监测滑坡变形在空间上的分布特征，以及从监测信息中分析和判定滑坡总体变形趋势和滑动方向；还能够通过针对不同的滑坡预警分类结果制定不同的应急方案，及时在保障人们生命、财产安全；还能够识别滑坡运动趋势、发展状况；还能够对滑坡的规模、形成机制进行分析，预测滑坡发展趋势，为后续滑坡治理提供指导依据。
实施例二
请参阅图2，图2是本发明实施例公开的一种滑坡预测装置的结构示意图。其中，图2所描述的基于随机森林模型的滑坡预测装置为监测山体滑坡运动状态的终端设备，本发明实施例不做限定。如图2所示，该基于随机森林模型的滑坡预测装置包括采集模块401、第一构建模块402、第二构建模块403评估模块404以及融合模块404，其中：
采集模块401，用于采集多类滑坡训练数据。
本发明实施例中，上述滑坡训练数据包括传感器位移训练数据、标志物运动轨迹训练数据、裂缝尺寸训练数据中的至少一种，本发明实施例不做限定。
第一构建模块402，用于分别构建上述多类滑坡训练数据的滑坡预警分类视角。
第二构建模块403，用于使用随机森林模型构建与所有上述滑坡预警分类视角相对应的多视角权重随机森林模型。
采集模块401，还用于采集多类滑坡测试数据。
第一构建模块402，还用于分别构建上述多类滑坡测试数据的滑坡观测视角。
评估模块404，用于使用上述多视角权重随机森林模型分别对每个滑坡观测视角进行模型评估，得到模型评估结果。
融合模块405，用于融合上述模型评估结果，得到滑坡预警分类结果。
可见，实施图2所描述的基于随机森林模型的滑坡预测装置通过多视角权重随机森林模型分析滑坡变动情况，不仅能够提高滑坡运动状态的预测效率和预测精准度，还能够为后期进行滑坡稳定性分析与评价，预测预警滑坡及后期防治工作提供定量化的评价依据。
本发明实施了中，作为一种可选的实施方式，第二构建模块403可以包括学习子模块4031、构建子模块4032。此时，该基于随机森林模型的滑坡预测装置如图3所示，图3为另一种滑坡预测装置，其中：
学习子模块4031，用于使用随机森林模型分别学习所有上述滑坡预警分类视角，得到所有视角随机森林模型以及与上述所有视角随机森林模型中的每个视角随机森林模型相对应的模型评分。
构建子模块4032，用于分别基于上述模型评分、与上述滑坡预警分类视角相对应的视角权重先验构建与上述滑坡训练数据相对应的滑坡预警视角权重。
构建子模块4032，还用于基于所有上述滑坡预警视角权重和所有视角随机森林模型构建多视角权重随机森林模型。
其中，在该可选的实施方式中，可选的，构建子模块4032分别基于上述模型评分、与上述滑坡预警分类视角相对应的视角权重先验构建与上述滑坡训练数据相对应的滑坡预警视角权重的方式具体为：
分别基于上述模型评分、与上述滑坡预警分类视角相对应的视角权重先验在贝叶斯框架下构建与上述滑坡训练数据相对应的滑坡预警视角权重。
其中，上述滑坡预警视角权重的计算公式为：
Figure PCTCN2019102982-appb-000007
式中：θ (i)表示上述滑坡预警分类视角的视角索引，，W θ(i)表示上述滑坡预警 视角权重，P i表示上述视角权重先验，P vi表示上述模型评分。在贝叶斯框架下，P i×P vi表示该滑坡预警分类视角的准确度的后验概率。
实施例三
请参阅图4，图4是本发明实施例公开的又一种滑坡预测装置的结构示意图。如图4所示，该基于随机森林模型的滑坡预测装置可以包括：
存储有可执行程序代码的存储器401；
与存储器401耦合的处理器402；
处理器402调用存储器401中存储的可执行程序代码，执行实施例一中所描述的基于随机森林模型的滑坡预测方法中的步骤。
实施例四
本发明实施例公开了一种计算机可读存储介质，其存储用于电子数据交换的计算机程序，其中，该计算机程序使得计算机执行实施例一中所描述的基于随机森林模型的滑坡预测方法中的步骤。
实施例五
本发明实施例公开了一种计算机程序产品，该计算机程序产品包括存储了计算机程序的非瞬时性计算机可读存储介质，且该计算机程序可操作来使计算机执行实施例一中所描述的基于随机森林模型的滑坡预测方法中的步骤。
以上所描述的装置实施例仅是示意性的，其中所述作为分离部件说明的模块可以是或者也可以不是物理上分开的，作为模块显示的部件可以是或者也可以不是物理模块，即可以位于一个地方，或者也可以分布到多个网络模块上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。本领域普通技术人员在不付出创造性的劳动的情况下，即可以理解并实施。
通过以上的实施例的具体描述，本领域的技术人员可以清楚地了解到各实施方式可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件。基于这样的理解，上述技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品可以存储在计算机可读存储介质中,存储介质包括只读存储器(Read-Only Memory，ROM)、随机存储器(Random Access Memory，RAM)、可编程只读存储器(Programmable Read-only Memory，PROM)、可擦除可编程只读存储器(Erasable Programmable Read Only Memory，EPROM)、一次可编程只读存储器(One-time Programmable Read-Only Memory，OTPROM)、电子抹除式可复写只读存储器(Electrically-Erasable  Programmable Read-Only Memory，EEPROM)、只读光盘(Compact Disc Read-Only Memory，CD-ROM)或其他光盘存储器、磁盘存储器、磁带存储器、或者能够用于携带或存储数据的计算机可读的任何其他介质。
最后应说明的是：本发明实施例公开的基于随机森林模型的滑坡预测方法及装置所揭露的仅为本发明较佳实施例而已，仅用于说明本发明的技术方案，而非对其限制；尽管参照前述实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解；其依然可以对前述各项实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或替换，并不使相应的技术方案的本质脱离本发明各项实施例技术方案的精神和范围。

Claims (10)

  1. 一种滑坡预测方法，其特征在于，所述方法包括：
    采集多类滑坡训练数据，并分别构建所述多类滑坡训练数据的滑坡预警分类视角,所述滑坡训练数据包括传感器位移训练数据、标志物运动轨迹训练数据、裂缝尺寸训练数据中的至少一种；
    使用随机森林模型构建与所有所述滑坡预警分类视角相对应的多视角权重随机森林模型；
    采集多类滑坡测试数据，并分别构建所述多类滑坡测试数据的滑坡观测视角；
    使用所述多视角权重随机森林模型分别对每个所述滑坡观测视角进行模型评估，得到模型评估结果，并融合所述模型评估结果，得到滑坡预警分类结果。
  2. 根据权利要求1所述的基于随机森林模型的滑坡预测方法，其特征在于，所述使用随机森林模型构建与所有所述滑坡预警分类视角相对应的多视角权重随机森林模型，包括：
    使用随机森林模型分别学习所有所述滑坡预警分类视角，得到所有视角随机森林模型以及与所述所有视角随机森林模型中的每个所述视角随机森林模型相对应的模型评分；
    分别基于所述模型评分、与所述滑坡预警分类视角相对应的视角权重先验构建与所述滑坡训练数据相对应的滑坡预警视角权重；
    基于所有所述滑坡预警视角权重和所有所述视角随机森林模型构建多视角权重随机森林模型。
  3. 根据权利要求2所述的基于随机森林模型的滑坡预测方法，其特征在于，所述分别基于所述模型评分、与所述滑坡预警分类视角相对应的视角权重先验构建与所述滑坡训练数据相对应的滑坡预警视角权重，包括：
    分别基于所述模型评分、与所述滑坡预警分类视角相对应的视角权重先验在贝叶斯框架下构建与所述滑坡训练数据相对应的滑坡预警视角权重。
  4. 根据权利要求3所述的基于随机森林模型的滑坡预测方法，其特征在于，所述滑坡视角权重的计算公式为：
    Figure PCTCN2019102982-appb-100001
    式中：
    θ (i)表示所述滑坡预警分类视角的视角索引，W θ(i)表示所述滑坡预警视角权重，P i表示所述视角权重先验，P vi表示所述模型评分。在所述贝叶斯框架下，P i×P vi表示所述滑坡预警分类视角的准确度的后验概率。
  5. 一种滑坡预测装置，其特征在于，所述装置包括：
    采集模块，用于采集多类滑坡训练数据,所述滑坡训练数据包括传感器位移训练数据、标志物运动轨迹训练数据、裂缝尺寸训练数据中的至少一种；
    第一构建模块，用于分别构建所述多类滑坡训练数据的滑坡预警分类视角；
    第二构建模块，用于使用随机森林模型构建与所有所述滑坡预警分类视角相对应的多视角权重随机森林模型；
    所述采集模块，还用于采集多类滑坡测试数据；
    所述第一构建模块，还用于分别构建所述多类滑坡测试数据的滑坡观测视角；
    评估模块，用于使用所述多视角权重随机森林模型分别对每个所述滑坡观测视角进行模型评估，得到模型评估结果；
    融合模块，用于融合所述模型评估结果，得到滑坡预警分类结果。
  6. 根据权利要求5所述的基于随机森林模型的滑坡预测装置，其特征在于，所述第二构建模块包括学习子模块以及构建子模块，其中：
    所述学习子模块，用于使用随机森林模型分别学习所有所述滑坡预警分类视角，得到所有视角随机森林模型以及与所述所有视角随机森林模型中的每个所述视角随机森林模型相对应的模型评分；
    所述构建子模块，用于分别基于所述模型评分、与所述滑坡预警分类视角相对应的视角权重先验构建与所述滑坡训练数据相对应的滑坡预警视角权重；
    所述构建子模块，还用于基于所有所述滑坡预警视角权重和所有所述视角随机森林模型构建多视角权重随机森林模型。
  7. 根据权利要求6所述的基于随机森林模型的滑坡预测装置，其特征在于，所述构建子模块分别基于所述模型评分、与所述滑坡预警分类视角相对应的视角权重先验构建与所述滑坡训练数据相对应的滑坡预警视角权重的方式具体为：
    分别基于所述模型评分、与所述滑坡预警分类视角相对应的视角权重先验在贝叶斯框架下构建与所述滑坡训练数据相对应的滑坡预警视角权重。
  8. 根据权利要求7所述的基于随机森林模型的滑坡预测装置，其特征在于， 所述滑坡预警视角权重的计算公式为：
    Figure PCTCN2019102982-appb-100002
    式中：
    θ (i)表示所述滑坡预警分类视角的视角索引，，W θ(i)表示所述滑坡预警视角权重，P i表示所述视角权重先验，P vi表示所述模型评分。在所述贝叶斯框架下，P i×P vi表示所述滑坡预警分类视角的准确度的后验概率。
  9. 一种滑坡预测装置，其特征在于，所述装置包括：
    存储有可执行程序代码的存储器；
    与所述存储器耦合的处理器；
    所述处理器调用所述存储器中存储的所述可执行程序代码，执行如权利要求1-4任一项所述的基于随机森林模型的滑坡预测方法。
  10. 一种计算机存储介质，其特征在于，所述计算机存储介质存储有计算机指令，所述计算机指令被调用时，用于执行如权利要求1-4任一项所述的基于随机森林模型的滑坡预测方法。
PCT/CN2019/102982 2019-01-25 2019-08-28 一种基于训练模型的滑坡预测方法及装置 WO2020151226A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910073688.9 2019-01-25
CN201910073688.9A CN109783967B (zh) 2019-01-25 2019-01-25 基于随机森林模型的滑坡预测方法、装置及存储介质

Publications (1)

Publication Number Publication Date
WO2020151226A1 true WO2020151226A1 (zh) 2020-07-30

Family

ID=66502409

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/102982 WO2020151226A1 (zh) 2019-01-25 2019-08-28 一种基于训练模型的滑坡预测方法及装置

Country Status (2)

Country Link
CN (1) CN109783967B (zh)
WO (1) WO2020151226A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112686192A (zh) * 2021-01-06 2021-04-20 电子科技大学 一种基于精细地形特征的滑坡稳定性分类方法
US11880015B2 (en) 2021-06-07 2024-01-23 International Business Machines Corporation Modeling tropical cyclone surface fields for impact assessment

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109783967B (zh) * 2019-01-25 2020-02-21 深圳大学 基于随机森林模型的滑坡预测方法、装置及存储介质
CN110223490B (zh) * 2019-05-28 2021-03-23 成都理工大学 一种基于预警等级判断岩质斜坡稳定性的方法
CN111667125B (zh) * 2020-08-10 2020-11-20 成都嘉捷信诚信息技术有限公司 一种滑坡位移预测方法、滑坡位移预测装置及存储介质
CN114078140B (zh) * 2020-08-19 2022-08-26 香港理工大学深圳研究院 一种基于滑坡边界多边形和坡向图的滑坡轨迹提取方法
CN113486575A (zh) * 2021-06-21 2021-10-08 武汉科技大学 一种地表变形破坏预测预警方法、装置及存储介质
CN113723446B (zh) * 2021-07-15 2024-05-24 杭州鲁尔物联科技有限公司 地灾监测预警方法、装置、计算机设备及存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102819023A (zh) * 2012-07-27 2012-12-12 中国地质大学（武汉） 基于LiDAR的复杂地质背景区滑坡识别的方法及系统
US20150205856A1 (en) * 2014-01-21 2015-07-23 Decision Makers - Learning & Research Simulations Ltd. Dynamic brownian motion with density superposition for abnormality detection
CN109003422A (zh) * 2018-08-02 2018-12-14 北京大学深圳研究生院 用于山体滑坡的监测数据处理方法和山体滑坡预报方法
CN109783967A (zh) * 2019-01-25 2019-05-21 深圳大学 一种滑坡预测方法及系统

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107292186B (zh) * 2016-03-31 2021-01-12 阿里巴巴集团控股有限公司 一种基于随机森林的模型训练方法和装置
CN107092798A (zh) * 2017-04-28 2017-08-25 成都理工大学 滑坡预测模型的稳定性评价方法及装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102819023A (zh) * 2012-07-27 2012-12-12 中国地质大学（武汉） 基于LiDAR的复杂地质背景区滑坡识别的方法及系统
US20150205856A1 (en) * 2014-01-21 2015-07-23 Decision Makers - Learning & Research Simulations Ltd. Dynamic brownian motion with density superposition for abnormality detection
CN109003422A (zh) * 2018-08-02 2018-12-14 北京大学深圳研究生院 用于山体滑坡的监测数据处理方法和山体滑坡预报方法
CN109783967A (zh) * 2019-01-25 2019-05-21 深圳大学 一种滑坡预测方法及系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
WU XIAOQING, CHENGGUANG LAI, XIAOHONG CHEN, XIUWEN REN: "A Landslide Hazard Assessment based on Random Forest Weight: a Case Study in the Dongjiang River Basin", JOURNAL OF NATURAL DISASTERS, vol. 26, no. 5, 28 October 2017 (2017-10-28), pages 119 - 129, XP055723594, ISSN: 1004-4574, DOI: 10.13577/j.jnd.2017.0514 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112686192A (zh) * 2021-01-06 2021-04-20 电子科技大学 一种基于精细地形特征的滑坡稳定性分类方法
CN112686192B (zh) * 2021-01-06 2022-05-31 电子科技大学 一种基于精细地形特征的滑坡稳定性分类方法
US11880015B2 (en) 2021-06-07 2024-01-23 International Business Machines Corporation Modeling tropical cyclone surface fields for impact assessment

Also Published As

Publication number Publication date
CN109783967B (zh) 2020-02-21
CN109783967A (zh) 2019-05-21

Similar Documents

Publication Publication Date Title
WO2020151226A1 (zh) 一种基于训练模型的滑坡预测方法及装置
CN108257114A (zh) 一种基于深度学习的输电设备缺陷自动识别方法
CN109979541A (zh) 基于胶囊网络的药物分子药代动力学性质和毒性预测方法
Le-Hoai et al. Estimating time performance for building construction projects in Vietnam
Dogan et al. Detection of damages caused by earthquake and reinforcement corrosion in RC buildings with Deep Transfer Learning
CN109490072A (zh) 一种土木工程建筑用检测系统及其检测方法
KR102379472B1 (ko) 시공간 피해특성을 고려한 다종 데이터 융합 분석방법
CN112149750A (zh) 一种供水管网爆管识别数据驱动方法
CN116384837B (zh) 用于工程质量检测的信息处理方法、系统和检测员终端
Catbas Structural health monitoring: applications and data analysis
CN115879654A (zh) 一种地下巷道稳定性预测和可靠性评价方法
CN114611372A (zh) 一种基于物联网边缘计算的工业设备健康预测方法
Matos et al. Building condition indicators analysis for BIM-FM integration
Ye et al. A deep learning-based method for automatic abnormal data detection: Case study for bridge structural health monitoring
Amuda Yusuf et al. Barriers to building information modelling adoption in Nigeria
CN109656904A (zh) 一种案件风险检测方法及系统
CN116884192A (zh) 一种电力生产作业风险预警方法、系统和设备
Liu et al. Quantifying the reliability of defects located by bridge inspectors through human observation behavioral analysis
CN117252305A (zh) 一种房屋风险评估方法、装置、设备及介质
CN117191147A (zh) 一种泄洪大坝水位监测预警方法及系统
CN117032165A (zh) 一种工业设备故障诊断方法
US20230169681A1 (en) Quantitative evaluation method and system for prediction result of remote sensing inversion
CN115795052A (zh) 一种产业链地图构建方法、装置及电子设备
CN108304809A (zh) 基于震后航拍影像的近实时震损评估方法
CN115168958A (zh) 基于建筑结构节点强度分析的快速推演系统及其方法

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 27.01.2022)

122 Ep: pct application non-entry in european phase

Ref document number: 19911293

Country of ref document: EP

Kind code of ref document: A1