CN111144636A - Slope deformation prediction method - Google Patents

Slope deformation prediction method Download PDF

Info

Publication number
CN111144636A
CN111144636A CN201911332358.3A CN201911332358A CN111144636A CN 111144636 A CN111144636 A CN 111144636A CN 201911332358 A CN201911332358 A CN 201911332358A CN 111144636 A CN111144636 A CN 111144636A
Authority
CN
China
Prior art keywords
sample
value
sum
data
derivatives
Prior art date
Legal status (The legal status 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 status listed.)
Granted
Application number
CN201911332358.3A
Other languages
Chinese (zh)
Other versions
CN111144636B (en
Inventor
许建聪
许辉鹏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tongji University
Original Assignee
Tongji University
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 Tongji University filed Critical Tongji University
Priority to CN201911332358.3A priority Critical patent/CN111144636B/en
Publication of CN111144636A publication Critical patent/CN111144636A/en
Application granted granted Critical
Publication of CN111144636B publication Critical patent/CN111144636B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
    • Y02A10/23—Dune restoration or creation; Cliff stabilisation

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • Development Economics (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Medical Informatics (AREA)
  • Evolutionary Computation (AREA)
  • Data Mining & Analysis (AREA)
  • Game Theory and Decision Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Pit Excavations, Shoring, Fill Or Stabilisation Of Slopes (AREA)
  • Testing Or Calibration Of Command Recording Devices (AREA)

Abstract

The invention relates to a slope deformation prediction method, which comprises the following steps: s1, encoding the monitoring data; s2, constructing a data set according to the coded data; s3, establishing an XGboost prediction model of slope deformation according to the data set; s4, optimizing the calculation parameters of the XGBoost prediction model; and S5, predicting the slope deformation inclination measurement value through the XGBoost prediction model after the parameters are optimized. Compared with the prior art, the prediction method provided by the invention can reflect the actual situation of slope deformation under complex working conditions more appropriately, and can effectively predict the slope and foundation pit deformation inclination measurement values, so that the possible dangerous situation can be predicted in advance, and valuable time is provided for timely making the best safety measures.

Description

Slope deformation prediction method
Technical Field
The invention relates to the field of slope and foundation pit engineering, in particular to a slope deformation prediction method.
Background
In order to ensure the safe construction and operation of the slope engineering and the safe construction of the deep foundation pit engineering adjacent to the existing tunnel, the inclination measurement is taken as an important monitoring means, and the change conditions of the slope and the foundation pit in the earthwork excavation can be reflected in real time. A certain number of inclinometer pipes are arranged in a deep foundation pit supported by a side slope and a pile anchor, the inclinometer data during construction is recorded periodically, the change trend of the inclinometer data is comprehensively analyzed, and the deformation conditions of the side slope and the foundation pit can be mastered. However, after the inclination survey data is obtained, if the deformation of the side slope and the foundation pit reaches the threshold value of occurrence of disaster, a dangerous situation occurs, and then the measures are taken possibly late, so that safety measures can be taken in time only by predicting the degree of the deformation to be achieved in advance, and the occurrence of the disaster is avoided. At present, methods such as a neural network and a support vector machine are commonly adopted to predict deformation. However, neural networks plus regularization mechanisms often assume that parameters conform to gaussian and laplacian distributions, and these models do not take into account temporal factors and are sensitive to outliers and missing values.
Disclosure of Invention
The invention aims to provide a method for predicting a slope deformation inclinometer value, which aims to overcome the defect that the prior art is sensitive to abnormal values and missing values.
The purpose of the invention can be realized by the following technical scheme:
a slope deformation prediction method comprises the following steps:
s1, encoding the monitoring data;
s2, constructing a data set according to the data coded in the step S1;
s3, establishing an XGBoost prediction model of slope deformation according to the data set constructed in the step S2;
s4, optimizing the calculation parameters of the XGBoost prediction model established in the step S3;
and S5, predicting the slope deformation inclination measurement value through the XGBoost prediction model established in the step S3 by adopting the calculation parameters optimized in the step S4.
Preferably, the specific process of the encoding process of the monitoring data in step S1 is as follows:
s11, time is directly expressed by integral days. When new data is predicted, the difference value of the current monitoring time minus the most initial monitoring time is used as the time code of the current data;
s12, taking the average value of the previous two times of data as the characteristic of the current data, and carrying out prior data coding;
and S13, coding the monitoring point numbers of all the monitoring points from 0 according to integers by adopting a one-hot coding mode.
S14, establishing a wide and universal model by specially coding the monitoring time, the monitoring data and the monitoring point number through S11, S12 and S13 respectively.
Preferably, in step S2, the constructing a data set from the encoded data includes:
s21, 16 feature dimensions are shared by one sample point, 13 dimensions are shared by monitoring point number codes, and three dimensions including time dimension, depth of a measuring point, prior information and the like are included. The y-label is the output value of the model, which is the inclinometer value.
S22, analyzing all data tables by adopting the 16 characteristic dimensions and the y labels of S21 to obtain total data;
s23, dividing the whole data set acquired in S22 into a training set, a verification set and a test set in a random mode. The training set accounts for 90% of the total data, the verification set accounts for 4.8-5.0% of the total data, and the test set accounts for 5.0-5.2% of the total data.
Preferably, the specific process of establishing the XGboost prediction model of slope deformation according to the constructed data set in step S3 is as follows:
s31, calculating the loss function L of the samples in the current round according to the training set samples I distributed in S23t：
Figure BDA0002329991390000021
In the formula, LtFor the loss function, n is the number of training samples,
Figure BDA0002329991390000022
for the loss of the ith sample, yiIn order to train the true label value of the sample,
Figure BDA0002329991390000023
for the strong learner predictor, f, for the ith sample at the t-1 iterationt(xi) For training sample xiA weak learner function in the T-th iterative training, gamma and lambda are coefficients set manually, T is the number of leaf nodes, wjIs the jth leaf node value.
S32, according to the training set sample I distributed in S23, calculating the empirical loss function L of the current sample in the current round based on the first order partial derivative and the second order partial derivative of the previous learner, and summing the first order partial derivative and the second order partial derivative:
Figure BDA0002329991390000024
in the formula, LtFor the loss function (equivalent to the previous expression, transformed by Taylor expansion), IjSet of training samples (sample index set) that fall into the jth leaf node, giIs the first partial derivative of the loss function of the ith sample on the predicted value of the ith sample, hiThe second partial derivative of the loss function of the ith sample to the predicted value of the ith sample is calculated, gamma and lambda are coefficients set manually, T is the number of leaf nodes, w is the number of leaf nodesjFor the jth leaf node value, the empirical loss function L is part of the loss function, i.e., the remaining part of the regularization is removed.
S33, the step of building the XGboost prediction model through iteration with a round number T of 1, 2.
S331, calculating the ith sample (i ═ 1,2, …, m) based on f in the current round loss function Lt-1(xi) First derivative g oftiSecond derivative htiCalculating the sum of the first derivatives of all samples
Figure BDA0002329991390000031
Sum of second derivatives of all samples
Figure BDA0002329991390000032
S332, attempting to split the decision tree based on the current node, wherein the default maximum score is 0, and the score is calculated for the feature sequence number K of 1,2, …, K, as follows:
s3321, order Gt＝0、Ht＝0；
S3322, arranging the samples from small to large according to the characteristic k, sequentially taking out the ith sample, and sequentially calculating the sum of the first derivatives and the sum of the second derivatives of the left subtree and the right subtree after the current sample is placed in the left subtree:
GL←GL+gti，GR←G-GL
HL←HL+hti，HR←H-hL
equation (c) ° n denotes value update, GLIs the sum of all sample first derivatives of the left sub-tree, GRIs the sum of all sample first order derivatives of the right subtree, G is the sum of all sample first order derivatives, HLIs the sum of all sample first derivatives of the left sub-tree, HRIs the sum of all sample second derivatives of the right subtree and H is the sum of all sample second derivatives.
S3323, updating the maximum score:
Figure BDA0002329991390000033
and S333, splitting the subtree based on the corresponding division characteristic and characteristic value of the maximum score.
S334, if the maximum score is 0, the current decision tree is established, and w of all leaf areas is calculatedtjGet weak learner ht(x) Updating strong learner ft(x) And entering the next weak learner iteration. If it is determined that the maximum score is not 0, go to step S332 to continue attempting to split the decision tree.
Preferably, the optimizing according to the calculation parameters of the established XGboost prediction model in step S4 includes:
and S41, introducing K-fold cross validation during XGboost algorithm training, estimating classifier generalization error, and evaluating the classification accuracy of the XGboost algorithm under different parameters when only having training samples.
K-fold verification: the sample is divided into K sub-samples, a single sub-sample is reserved as data of the verification model, and the other K-1 samples are used for training. Cross validation was repeated K times, once for each subsample, and the results averaged K times to obtain a single estimate.
S42, adopting a grid search algorithm to search multidimensional arrays from different growth directions in parallel, and determining grid search parameters as follows:
s421, determining the search range of the parameters (assumed to be p and q) to be searched according to experience;
s422, setting the search step length of the searched parameters;
s423, calculating the classification accuracy of the support vector machine according to a cross validation method for each group of p values and q values on the grid;
and S424, drawing the classification accuracy of each group by using contour lines to obtain a contour map, and accordingly determining the optimal p value and q value.
Preferably, the step S5 of predicting the slope deformation inclinometer value through the XGboost prediction model established in the step S3 according to the calculation parameters optimized in the step S4 includes: and inputting the coded data into the established XGBoost prediction model to obtain a predicted value of the inclinometer value.
Compared with the prior art, the invention has the following advantages:
(1) xgboost is an integrated learning model taking a decision tree as a basic model, has no assumed distribution on data, is insensitive to noise, and can encode time factors in the modeling process and input the time factors into the model. The XGboost prediction method adopts a 'divide-and-conquer, popular and popular' strategy, completes a learning task by constructing and combining a plurality of machine learners, more closely reflects the actual situation of slope deformation under complex working conditions, and can effectively predict the deformation slope-measuring values of the slope and the foundation pit, thereby predicting the dangerous case which possibly occurs in advance and providing precious time for making the best safety measure in time.
(2) The invention establishes a widely universal XGboost prediction model by respectively carrying out special coding on the monitoring point number, the monitoring data and the time, takes the time factor into consideration, and can effectively predict the slope deformation inclinometer value, thereby predicting the possible dangerous case in advance and providing precious time for making the best safety measure in time.
(3) The method can accurately predict according to the measured data of few time steps, so that the relative error between the predicted value and the measured value can quickly reach 6%, and the speed, the precision and the reliability of early warning and forecasting of dangerous cases in the slope construction process are improved.
Drawings
FIG. 1 is a schematic flow chart of the present invention;
FIG. 2 is a flow chart of the Xgboost algorithm modeling;
FIG. 3 is a grid search algorithm;
fig. 4 is a flowchart of a grid search algorithm.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
Examples
As shown in fig. 1, a method for predicting slope deformation includes the following steps:
s1, encoding the monitoring data;
s2, constructing a data set according to the data coded in the step S1;
s3, establishing an XGBoost prediction model of slope deformation according to the data set constructed in the step S2;
s4, optimizing the calculation parameters of the XGBoost prediction model established in the step S3;
and S5, predicting the slope deformation inclination measurement value through the XGBoost prediction model established in the step S3 by adopting the calculation parameters optimized in the step S4.
Wherein, step S1 includes:
s11, time is directly expressed by integral days. When new data is predicted, the difference value of the current monitoring time minus the most initial monitoring time is used as the time code of the current data;
s12, taking the average value of the previous two times of data as the characteristic of the current data, and carrying out prior data coding;
and S13, coding the monitoring point numbers of all the monitoring points from 0 according to integers by adopting a one-hot coding mode.
S14, establishing a wide and universal model by specially coding the monitoring time, the monitoring data and the monitoring point number through S11, S12 and S13 respectively.
Step S2 includes constructing a data set from the encoded data, specifically including:
s21, 16 feature dimensions are shared by one sample point, 13 dimensions are shared by monitoring point number codes, and three dimensions including time dimension, depth of a measuring point, prior information and the like are included. The y-label is the output value of the model, which is the inclinometer value.
S22, analyzing all data tables by adopting the 16 characteristic dimensions and the y labels of S21 to obtain total data;
s23, dividing the whole data set acquired in S22 into a training set, a verification set and a test set in a random mode. The training set accounts for 90% of the total data, the verification set accounts for 4.8-5.0% of the total data, and the test set accounts for 5.0-5.2% of the total data.
Step S3 includes: s31, calculating the loss function L of the samples in the current round according to the training set samples I distributed in S23t：
Figure BDA0002329991390000061
In the formula, LtFor the loss function, n is the number of training samples,
Figure BDA0002329991390000062
for the loss of the ith sample, yiIn order to train the true label value of the sample,
Figure BDA0002329991390000063
for the strong learner predictor, f, for the ith sample at the t-1 iterationt(xi) For training sample xiA weak learner function in the T-th iterative training, gamma and lambda are coefficients set manually, T is the number of leaf nodes, wjIs the jth leaf node value.
S32, according to the training set sample I distributed in S23, calculating the empirical loss function L of the current sample in the current round based on the first order partial derivative and the second order partial derivative of the previous learner, and summing the first order partial derivative and the second order partial derivative:
Figure BDA0002329991390000064
in the formula, LtAs a loss function, IjSet of training samples (sample index set) that fall into the jth leaf node, giIs the first partial derivative of the loss function of the ith sample on the predicted value of the ith sample, hiThe second partial derivative of the loss function of the ith sample to the predicted value of the ith sample is calculated, gamma and lambda are coefficients set manually, T is the number of leaf nodes, w is the number of leaf nodesjIs the jth leaf node value.
S33, through iteration with a round number T being 1, 2.. T, as shown in fig. 2, the process of specifically establishing the XGboost prediction model in this embodiment is as follows:
s331, calculating the i-th sample (i ═ 1,2, …, m) based on f in the current round empirical loss function Lt-1(xi) First derivative g oftiSecond derivative htiCalculating the sum of the first derivatives of all samples
Figure BDA0002329991390000065
Sum of second derivatives of all samples
Figure BDA0002329991390000066
S332, attempting to split the decision tree based on the current node, where the default score is 0, and the feature sequence number K is 1,2, …, and K, and the step of calculating the score is:
s3321, order Gt＝0、Ht＝0；
S3322, arranging the samples from small to large according to the characteristic k, sequentially taking out the ith sample, and sequentially calculating the sum of the first derivatives and the sum of the second derivatives of the left subtree and the right subtree after the current sample is placed in the left subtree:
GL←GL+gti，GR←G-GL
HL←HL+hti，HR←H-hL
in the formula, GLIs the sum of all sample first derivatives of the left sub-tree, GRIs the sum of all sample first order derivatives of the right subtree, G is the sum of all sample first order derivatives, HLIs the sum of all sample first derivatives of the left sub-tree, HRIs the sum of all sample second derivatives of the right subtree and H is the sum of all sample second derivatives.
S3323, updating the maximum score:
Figure BDA0002329991390000071
and S333, splitting the subtree based on the corresponding division characteristic and characteristic value of the maximum score.
S334, if the maximum score is 0, the current decision tree is established, and w of all leaf areas is calculatedtjGet weak learner ht(x) Updating strong learner ft(x) And entering the next weak learner iteration. If it is determined that the maximum score is not 0, go to step S332 to continue attempting to split the decision tree.
Step S4 includes optimizing according to the calculation parameters of the established XGboost prediction model, which specifically includes:
and S41, introducing K-fold cross validation during XGboost algorithm training, estimating classifier generalization error, and evaluating the classification accuracy of the XGboost algorithm under different parameters when only having training samples.
K-fold verification: the sample is divided into K sub-samples, a single sub-sample is reserved as data of the verification model, and the other K-1 samples are used for training. Cross validation was repeated K times, once for each subsample, and the results averaged K times to obtain a single estimate.
S42, using a grid search algorithm to search the multidimensional arrays from different growth directions in parallel, as shown in fig. 3 and 4, the process of determining the grid search parameters in this embodiment specifically includes:
s421, determining the search range of the parameters (assumed to be p and q) to be searched according to experience;
s422, setting the search step length of the searched parameters;
s423, calculating the classification accuracy of the support vector machine according to a cross validation method for each group of p values and q values on the grid;
and S424, drawing the classification accuracy of each group by using contour lines to obtain a contour map, and accordingly determining the optimal p value and q value.
Step S5 includes: predicting a slope deformation inclinometry value through the XGBoost prediction model established in the step S3 according to the calculation parameters optimized in the step S4, wherein the slope deformation inclinometry value comprises the following steps: and inputting the coded data into the established XGBoost prediction model to obtain a predicted value of the measuring point deformation inclinometry value.

Claims (9)

1. A method for predicting slope deformation is characterized by comprising the following steps:
s1, encoding the monitoring data;
s2, constructing a data set according to the coded data;
s3, establishing an XGboost prediction model of slope deformation according to the data set;
s4, optimizing the calculation parameters of the XGBoost prediction model;
and S5, predicting the slope deformation inclination measurement value through the XGBoost prediction model after the parameters are optimized.
2. The method for predicting slope deformation according to claim 1, wherein the step S1 specifically comprises the steps of:
s11, taking the difference value of the current monitoring time minus the most initial monitoring time as the time code of the current data;
s12, taking the average value of the previous two monitoring data as the characteristic of the current monitoring data, and carrying out prior data coding;
and S13, coding the monitoring point numbers of all the monitoring points from 0 by adopting a one-hot coding mode.
S14, through coding the monitoring time, the monitoring data and the monitoring point number respectively through S11, S12 and S13, a widely universal model is established.
3. The method for predicting slope deformation according to claim 1, wherein the step S2 specifically comprises the steps of:
s21, analyzing all data tables by adopting 16 characteristic dimensions and y labels of sample points to obtain a total data set, wherein the 16 characteristic dimensions of the sample points comprise 13 dimensions of monitoring point number codes, time dimensions, depth of a measuring point and 3 dimensions of prior information, the y label is an output value of a model, and the output value of the model is an inclinometry value;
and S22, dividing the total data set into a training set, a verification set and a test set in a random mode.
4. The method for predicting slope deformation according to claim 3, wherein the step S3 specifically comprises the following steps:
s31, calculating the loss function L of the sample I in the current round according to the sample I of the training sett：
Figure FDA0002329991380000011
In the formula, LtFor the loss function, n is the number of training samples,
Figure FDA0002329991380000012
for the ith training sample xiLoss of (y)iIn order to train the true label value of the sample,
Figure FDA0002329991380000013
for the strong learner predictor, f, for the ith sample at the t-1 iterationt(xi) For the ith training sample xiA weak learner function in the T-th iterative training, gamma and lambda are coefficients set manually, T is the number of leaf nodes, wjIs the jth leaf node value;
s32, according to the training set sample I, calculating the empirical loss function L of the current sample based on the first order partial derivative and the second order partial derivative of the previous learner, and calculating the sum of the first order partial derivative and the second order partial derivative:
Figure FDA0002329991380000021
in the formula, LtAs a loss function, IjFor the training sample set that falls into the jth leaf node, giIs the first partial derivative of the loss function of the ith sample on the predicted value of the ith sample, hiThe second partial derivative of the loss function of the ith sample to the predicted value of the ith sample is obtained, gamma and lambda are manually set coefficients, and the empirical loss function L is part of the loss function, namely the rest part of regularization is removed;
s33, building an XGboost prediction model through iteration with a round number T ═ 1, 2.. T.
5. The method for predicting slope deformation according to claim 4, wherein the step S33 specifically comprises the following steps:
s331, calculating the i-th sample in an empirical loss function L based on ft-1(xi) First derivative g oftiAnd second derivative htiCalculating the sum of the first derivatives of all samples
Figure FDA0002329991380000022
Sum of second derivatives of all samples
Figure FDA0002329991380000023
i＝1，2，…，m；
S332, attempting to split the decision tree based on the current node, and calculating the maximum score for all the features k;
s333, splitting sub-trees based on the division characteristics and the characteristic values corresponding to the maximum score;
s334, if the maximum score is 0, the current decision tree is established, and w of all leaf areas is calculatedtj，wtjTraining the value of the jth leaf node for the t round to obtain a weak learner ht(x) Updating strong learning device ft(x) And entering the next weak learner iteration, and if the maximum score is not 0, going to step S332 to continue trying to split the decision tree.
6. The method according to claim 5, wherein the step S332 specifically comprises the following steps:
s3321, let initial maximum score 0, Gt＝0、Ht＝0；
S3322, arranging the samples from small to large according to the characteristic k, sequentially taking out the ith sample, and sequentially calculating the sum of the first derivatives and the sum of the second derivatives of the left subtree and the right subtree after the current sample is placed in the left subtree:
GL←GL+gti，GR←G-GL
HL←HL+hti，HR←H-hL
equation (c) ° n denotes value update, GLIs the sum of all sample first derivatives of the left sub-tree, GRIs the sum of all sample first order derivatives of the right subtree, G is the sum of all sample first order derivatives, HLIs the sum of all sample first derivatives of the left sub-tree, HRIs the sum of all sample second derivatives of the right subtree and H is the sum of all sample second derivatives.
S3323, updating the maximum score:
Figure FDA0002329991380000031
7. the method according to claim 1, wherein the step S4 includes the steps of:
s41, introducing K-fold cross validation during XGboost algorithm training, estimating classifier generalization error, and evaluating classification accuracy of the XGboost algorithm under different parameters when only having training samples;
s42, adopting a grid search algorithm to search the multidimensional arrays in parallel from different growth directions, and determining the parameters of grid search.
8. The method for predicting slope deformation according to claim 7, wherein said step S42 includes the steps of:
s421, determining the search range of the parameters p and q to be searched according to experience;
s422, setting the search step length of the searched parameters;
s423, calculating the classification accuracy of the support vector machine according to a cross validation method for each group of p values and q values on the grid;
and S424, drawing the classification accuracy of each group by using contour lines to obtain a contour map, and accordingly determining the optimal p value and q value.
9. The method according to claim 7, wherein the step S5 includes: and inputting the coded data into the established XGBoost prediction model to obtain a predicted value of the inclinometer value.
CN201911332358.3A 2019-12-22 2019-12-22 Slope deformation prediction method Active CN111144636B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911332358.3A CN111144636B (en) 2019-12-22 2019-12-22 Slope deformation prediction method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911332358.3A CN111144636B (en) 2019-12-22 2019-12-22 Slope deformation prediction method

Publications (2)

Publication Number Publication Date
CN111144636A true CN111144636A (en) 2020-05-12
CN111144636B CN111144636B (en) 2022-08-16

Family

ID=70519276

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911332358.3A Active CN111144636B (en) 2019-12-22 2019-12-22 Slope deformation prediction method

Country Status (1)

Country Link
CN (1) CN111144636B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111784070A (en) * 2020-07-09 2020-10-16 中国地质环境监测院 Intelligent landslide short-term early warning method based on XGboost algorithm
CN112287595A (en) * 2020-09-16 2021-01-29 同济大学 Method for predicting shield tunnel wall back grouting thickness based on ground penetrating radar detection and machine learning
CN113326660A (en) * 2021-06-17 2021-08-31 广西路桥工程集团有限公司 Tunnel surrounding rock extrusion deformation prediction method based on GA-XGboost model

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105139585A (en) * 2015-07-29 2015-12-09 同济大学 Intelligent early warning and prediction method for soil slope danger
CN108229726A (en) * 2017-12-15 2018-06-29 同济大学 A kind of shiver stone side slope catastrophe dangerous situation early alarming and forecasting method
CN108830012A (en) * 2018-06-29 2018-11-16 上海勘察设计研究院(集团)有限公司 A kind of malformation prediction technique excavated based on total factor
WO2019094729A1 (en) * 2017-11-09 2019-05-16 Strong Force Iot Portfolio 2016, Llc Methods and systems for the industrial internet of things
CN110348561A (en) * 2019-07-13 2019-10-18 东北大学 A kind of Slope Displacement Prediction early warning system based on improvement artificial fish-swarm algorithm

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105139585A (en) * 2015-07-29 2015-12-09 同济大学 Intelligent early warning and prediction method for soil slope danger
WO2019094729A1 (en) * 2017-11-09 2019-05-16 Strong Force Iot Portfolio 2016, Llc Methods and systems for the industrial internet of things
CN108229726A (en) * 2017-12-15 2018-06-29 同济大学 A kind of shiver stone side slope catastrophe dangerous situation early alarming and forecasting method
CN108830012A (en) * 2018-06-29 2018-11-16 上海勘察设计研究院(集团)有限公司 A kind of malformation prediction technique excavated based on total factor
CN110348561A (en) * 2019-07-13 2019-10-18 东北大学 A kind of Slope Displacement Prediction early warning system based on improvement artificial fish-swarm algorithm

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
GAO HEBIN: "Comprehensive Application of Slope Monitoring and Early-Warning Techniques in Landslide Monitoring", 《2015 8TH INTERNATIONAL CONFERENCE ON INTELLIGENT COMPUTATION TECHNOLOGY AND AUTOMATION (ICICTA)》 *
林楠等: "基于随机森林模型的季节性冻区深基坑变形预测", 《地下空间与工程学报》 *
胡添翼等: "基于随机森林分类算法的边坡稳定预测模型", 《人民黄河》 *
邵良杉等: "基于RF-ELM模型的边坡稳定性预测研究", 《中国安全生产科学技术》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111784070A (en) * 2020-07-09 2020-10-16 中国地质环境监测院 Intelligent landslide short-term early warning method based on XGboost algorithm
CN112287595A (en) * 2020-09-16 2021-01-29 同济大学 Method for predicting shield tunnel wall back grouting thickness based on ground penetrating radar detection and machine learning
CN112287595B (en) * 2020-09-16 2023-05-23 同济大学 Prediction method of post-grouting thickness of shield tunnel wall based on ground penetrating radar detection and machine learning
CN113326660A (en) * 2021-06-17 2021-08-31 广西路桥工程集团有限公司 Tunnel surrounding rock extrusion deformation prediction method based on GA-XGboost model

Also Published As

Publication number Publication date
CN111144636B (en) 2022-08-16

Similar Documents

Publication Publication Date Title
CN113919448B (en) Method for analyzing influence factors of carbon dioxide concentration prediction at any time-space position
CN108701274B (en) Urban small-scale air quality index prediction method and system
CN111144636B (en) Slope deformation prediction method
CN108009674A (en) Air PM2.5 concentration prediction methods based on CNN and LSTM fused neural networks
CN107153874B (en) Water quality prediction method and system
CN109145464B (en) Structural damage identification method integrating multi-target ant lion optimization and trace sparse regularization
CN105740991A (en) Climate change prediction method and system for fitting various climate modes based on modified BP neural network
Eyoh et al. Modelling and predicting future urban expansion of Lagos, Nigeria from remote sensing data using logistic regression and GIS
CN110738355A (en) urban waterlogging prediction method based on neural network
CN111242351A (en) Tropical cyclone track prediction method based on self-encoder and GRU neural network
CN111325403A (en) Method for predicting remaining life of electromechanical equipment of highway tunnel
CN105678417A (en) Prediction method and device for tunnel face water inflow of construction tunnel
CN114997534A (en) Similar rainfall forecasting method and equipment based on visual features
CN114169434A (en) Load prediction method
CN115238850A (en) Mountain slope displacement prediction method based on MI-GRA and improved PSO-LSTM
CN116050670A (en) Road maintenance decision method and system based on data driving
CN115948964A (en) Road flatness prediction method based on GA-BP neural network
CN115204032A (en) ENSO prediction method and device based on multi-channel intelligent model
CN110852415B (en) Vegetation index prediction method, system and equipment based on neural network algorithm
JP4486004B2 (en) River pollution load estimation system, method, and program
Tewari et al. Application of LSTMs and HAND in rapid flood inundation mapping
CN114676887A (en) River water quality prediction method based on graph convolution STG-LSTM
CN116362714B (en) Prediction method and device for pavement maintenance period
CN116188921B (en) Tunnel gas concentration prediction method based on multi-graph fusion
CN117454762B (en) Coal-penetrating tunnel face gas concentration prediction method of Markov-neural network

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant