CN109271734A - A kind of earthquake-induced site liquefaction potential evaluation method based on machine learning techniques - Google Patents
A kind of earthquake-induced site liquefaction potential evaluation method based on machine learning techniques Download PDFInfo
- Publication number
- CN109271734A CN109271734A CN201811184759.4A CN201811184759A CN109271734A CN 109271734 A CN109271734 A CN 109271734A CN 201811184759 A CN201811184759 A CN 201811184759A CN 109271734 A CN109271734 A CN 109271734A
- Authority
- CN
- China
- Prior art keywords
- length
- attribute
- turns
- err
- equal
- 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
Links
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Geometry (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Structural Engineering (AREA)
- Computational Mathematics (AREA)
- Civil Engineering (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Architecture (AREA)
- Investigation Of Foundation Soil And Reinforcement Of Foundation Soil By Compacting Or Drainage (AREA)
Abstract
The present invention discloses a kind of earthquake-induced site liquefaction potential evaluation method based on machine learning techniques, includes the following steps: S1: starting;S2: input: including the data acquisition system of n reason attribute variable and 1 tag variable, the broken number k of cross validation;S3: data are defined;S4: enabling i=0, turns S5;S5: when i is not more than length-FnWhen, enable length-FiFor FiIn include attribute number turn S6 with season j=1;If i is greater than length-FnWhen, directly arrive S12;S6: enabling j=1, turns S7;S7: when the numerical value of j is not more than length-Fn, while i be not equal to 0 when, define fjFor FiIn j-th of attribute, enable Fj=Fi\fj;Data acquisition system is divided into k parts simultaneously, turns S8;When i is equal to 0, data acquisition system is directly divided into k parts, turns S8;When the numerical value of j is greater than length-FnWhen, turn S11;S9: judging the size of s and k repeatedly, carries out cross validation;The reliability for improving model, improves the generalization ability of model, has further pushed the practical application of the Liquefaction Potential of Foundation evaluation method based on machine learning techniques.
Description
Technical field
The invention belongs to seismic resistance field more particularly to a kind of earthquake-induced site liquefaction potential evaluation sides based on machine learning techniques
Method.
Background technique
It is well known to those skilled in the art, soil can generate state sharply and change and lose under geological process
Strength and stiffness, lead to the destruction of above ground structure, and this phenomenon is referred to as liquefaction phenomenon.Liquefaction is to cause building in earthquake
In it is unstable so that destroy principal element and seismic study and ground Aseismic Design importance.Liquefaction potential is exactly full
The excess pore water pressure generated under geological process with sand or saturation silt, makes the effective shearing strength of the soil body reduce or disappear,
To lead to a kind of trend of soil layer sand boil or soil body slipping unstability.Liquefaction potential is evaluated, can determine whether ground in earthquake
Under effect a possibility that liquefaction, thus to take anti-liquefaction measure to provide foundation in architectural design.
Summary of the invention
Technical problem to be solved by the present invention lies in the earthquake-induced site liquefaction potential provided based on machine learning techniques evaluations
Method, evaluation of liquefaction classifier generalization ability to be improved.
Technical scheme is as follows: a kind of earthquake-induced site liquefaction potential evaluation method based on machine learning techniques, packet
Include following steps:
S1: start;
S2: input: including the data acquisition system of n reason attribute variable and 1 tag variable, the broken number k of cross validation;
S3: definition: FnFor all attribute sets, length-FnFor FnIn include attribute number, FiFor FnSubset, and
Initialization: Fi=Fn;
S4: enabling i=0, turns S5;
S5: when i is not more than length-FnWhen, enable length-FiFor FiIn include attribute number, with season j=1, turn
S6;
If i is greater than length-FnWhen, directly arrive S12;
S6: enabling j=1, turns S7;
S7: when the numerical value of j is not more than length-Fn, while i be not equal to 0 when, define fjFor FiIn j-th of attribute, enable Fj
=Fi\fj;Data acquisition system is divided into k parts simultaneously, turns S8;
When i is equal to 0, data acquisition system is directly divided into k parts, turns S8;
When the numerical value of j is greater than length-FnWhen, turn S11;
S8: enabling s=1, turns S9;
S9: judging the size of s and k repeatedly, carries out cross validation;
Cross validation is completed, and is had k error amount, is enabled errijFor maximum value in k error amount, err is recordedij, judge simultaneously
Whether i is equal to 0, if being equal to 0, enables erri=errij, turn S10, if i is not equal to 0, enable j=j+1, and return to S7;
S10: enabling i=i+1, returns to S5;
S11: it enablesEnable FiFor erriCorresponding attribute set, judges erriWhether it is greater than or equal to
erri-1,
If then exporting Fi-1, turn S12;
If otherwise returning to S10;
S12: terminate.
Preferably, cross validation is embodied in step S9 are as follows: when s is not more than k, with FjFor the property set of model
It closes, s parts of data are used to classification of assessment device, other k-1 parts is used to train classifier;The value of s+1 is assigned to simultaneously new
s;
When s is greater than k, cross validation is completed.
Preferably, in S1 advance line number Data preprocess, attribute variable's data are standardized, standardized calculation formula is such as
Under:
Using above-mentioned formula, attribute variable's data are mapped to section [0,1].As a result become
Measure yi∈ { -1,1 }, yi=-1 indicates non-liquefied foundation, yi=1 indicates Liquefaction Foundation.
Above-mentioned step realizes that process is all to complete under the operation of calculator memory, rather than human activity or intelligence are advised
Then.
The utility model has the advantages that optimisation strategy is one by one to delete attribute using the worst error of cross validation as optimization object, find
It can make the smallest attribute set of cross validation worst error;It can be effectively reduced the fluctuation of category of model accuracy rate, and improve
The reliability of model, improves the generalization ability of model, has further pushed the Liquefaction Potential of Foundation based on machine learning techniques
The practical application of evaluation method.
Detailed description of the invention
Fig. 1 is inventive algorithm flow chart;
Fig. 2 is cross validation accuracy rate figure before and after Attributions selection;
Fig. 3 is classifier cross validation AUC figure after Attributions selection.
Specific embodiment
Present invention will be further explained below with reference to the attached drawings and examples:
As shown in Figure 1,
A kind of earthquake-induced site liquefaction potential evaluation method based on machine learning techniques, includes the following steps:
S1: start;
S2: input: including the data acquisition system of n reason attribute variable and 1 tag variable, the broken number k of cross validation;
S3: definition: FnFor all attribute sets, length-FnFor FnIn include attribute number, FiFor FnSubset, and
Initialization: Fi=Fn;
S4: enabling i=0, turns S5;
S5: when i is not more than length-FnWhen, enable length-FiFor FiIn include attribute number, with season j=1, turn
S6;
If i is greater than length-FnWhen, directly arrive S12;
S6: enabling j=1, turns S7;
S7: when the numerical value of j is not more than length-Fn, while i be not equal to 0 when, define fjFor FiIn j-th of attribute, enable Fj
=Fi\fj;Data acquisition system is divided into k parts simultaneously, turns S8;
When i is equal to 0, data acquisition system is directly divided into k parts, turns S8;
When the numerical value of j is greater than length-FnWhen, turn S11;
S8: enabling s=1, turns S9;
S9: judging the size of s and k repeatedly, carries out cross validation;
Cross validation is completed, and is had k error amount, is enabled errijFor maximum value in k error amount, err is recordedij, judge simultaneously
Whether i is equal to 0, if being equal to 0, enables erri=errij, turn S10, if i is not equal to 0, enable j=j+1, and return to S7;
S10: enabling i=i+1, returns to S5;
S11: it enablesEnable FiFor erriCorresponding attribute set, judges erriWhether it is greater than or equal to
erri-1,
If then exporting Fi-1, turn S12;
If otherwise returning to S10;
S12: terminate.
Further, cross validation is embodied in step S9 are as follows: when s is not more than k, with FjFor the property set of model
It closes, s parts of data are used to classification of assessment device, other k-1 parts is used to train classifier;The value of s+1 is assigned to simultaneously new
s;Circulation comparison can be carried out compared with k to s, can quickly carry out cross validation.
When s is greater than k, cross validation is completed;
Further, in S1 advance line number Data preprocess, attribute variable's data are standardized, standardized calculation formula
It is as follows:
Using above-mentioned formula, attribute variable's data are mapped to section [0,1].As a result become
Measure yi∈ { -1,1 }, yi=-1 indicates non-liquefied foundation, yi=1 indicates Liquefaction Foundation.
As shown in Fig. 2, left side be Attributions selection before, right side be Attributions selection after, using support vector machines as classifier.Using
Originally cross validation accuracy rate that the method researched and proposed obtains and quasi- to be applicable in the cross validation that the method originally researched and proposed obtains
True rate such as Fig. 2, the attribute before Attributions selection includes soil depth (Depth), measures circular cone resistance (qc), frictional ratio (Rf), effectively
Vertical stress (σ 'v), total vertical stress (σv), peak ground horizontal acceleration (αmax) and moment magnitude (Mw), by calculating, obtain
Attribute set is { Depth, qc,Rf,σ'v,αmax,Mw}。
In 10 folding cross validations, the worst error before Attributions selection is 0.1304, and corresponding accuracy rate is 0.8696;Belong to
Property select after worst error be 0.1053, corresponding accuracy rate be 0.8947.Meanwhile there are 3 accuracy to be before Attributions selection
100%, having 5 accuracy after Attributions selection is 100%. as it can be seen that after Attributions selection, and the calculating accuracy rate of classifier has bright
Aobvious raising.The coefficient of variation of cross validation accuracy before Attributions selection is 0.04959;Cross validation after Attributions selection is quasi-
The coefficient of variation of exactness is 0.0478.After Attributions selection, the accuracy of classifier increases, and generalization ability is more preferable, meter
The discreteness for calculating accuracy rate reduces, and calculated result is relatively reliable.
As shown in figure 3, showing that classifier has preferable differentiation performance, using 10 folding cross validations, benefit when AUC > 0.9
Classifier is trained with different data, and obtains 10 AUC value, as shown in figure 3, whole AUC value are all larger than 0.9, illustrates process
Data after selection can train the classifier with preferable performance, the common evaluation index of classifier in machine learning practice
It is exactly AUC, i.e. AUC (Area under Curve): Roc area under a curve, between 0.1 and 1.AUC can as numerical value
With the quality of intuitive classification of assessment device, value is the bigger the better.
Described in summary, the data set that this case uses includes 226 datas, wherein 133 are liquefaction, other 93 are non-
Liquefaction.These data are, the soil classes collected from 52 places in 6 earthquakes by cone penetration test (abbreviation CPT)
Type includes sand, silt and silt.Originally the method researched and proposed can be effectively reduced the fluctuation of category of model accuracy rate, and mention
The high reliability of model, improves the generalization ability of model, has further pushed the foundation liquefaction based on machine learning techniques
The practical application of gesture evaluation method.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not limitation with the present invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (3)
1. a kind of earthquake-induced site liquefaction potential evaluation method based on machine learning techniques, it is characterised in that include the following steps:
S1: start;
S2: input: including the data acquisition system of n reason attribute variable and 1 tag variable, the broken number k of cross validation;
S3: definition: FnFor all attribute sets, length-FnFor FnIn include attribute number, FiFor FnSubset, and it is initial
Change: Fi=Fn;
S4: enabling i=0, turns S5;
S5: when i is not more than length-FnWhen, enable length-FiFor FiIn include attribute number turn S6 with season j=1;
If i is greater than length-FnWhen, directly arrive S12;
S6: enabling j=1, turns S7;
S7: when the numerical value of j is not more than length-Fn, while i be not equal to 0 when, define fjFor FiIn j-th of attribute, enable Fj=Fi\
fj;Data acquisition system is divided into k parts simultaneously, turns S8;
When i is equal to 0, data acquisition system is directly divided into k parts, turns S8;
When the numerical value of j is greater than length-FnWhen, turn S11;
S8: enabling s=1, turns S9;
S9: judging the size of s and k repeatedly, carries out cross validation;
Cross validation is completed, and is had k error amount, is enabled errijFor maximum value in k error amount, err is recordedij, while whether judging i
Equal to 0, if being equal to 0, err is enabledi=errij, turn S10, if i is not equal to 0, enable j=j+1, and return to S7;
S10: enabling i=i+1, returns to S5;
S11: it enablesEnable FiFor erriCorresponding attribute set, judges erriWhether err is greater than or equal toi-1,
If then exporting Fi-1, turn S12;
If otherwise returning to S10;
S12: terminate.
2. a kind of earthquake-induced site liquefaction potential evaluation method based on machine learning techniques according to claim 1, feature
Be: cross validation is embodied in step S9 are as follows: when s is not more than k, with FjFor the attribute set of model, s parts of data
For classification of assessment device, other k-1 parts is used to train classifier;The value of s+1 is assigned to new s simultaneously;
When s is greater than k, cross validation is completed.
3. a kind of earthquake-induced site liquefaction potential evaluation method based on machine learning techniques according to claim 1, feature
It is: in S1 advance line number Data preprocess, attribute variable's data is standardized, standardized calculation formula is as follows:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811184759.4A CN109271734B (en) | 2018-10-11 | 2018-10-11 | Site earthquake liquefaction potential evaluation method based on machine learning technology |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811184759.4A CN109271734B (en) | 2018-10-11 | 2018-10-11 | Site earthquake liquefaction potential evaluation method based on machine learning technology |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109271734A true CN109271734A (en) | 2019-01-25 |
CN109271734B CN109271734B (en) | 2023-04-07 |
Family
ID=65195803
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811184759.4A Active CN109271734B (en) | 2018-10-11 | 2018-10-11 | Site earthquake liquefaction potential evaluation method based on machine learning technology |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109271734B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111400915A (en) * | 2020-03-17 | 2020-07-10 | 桂林理工大学 | Sand liquefaction discrimination method and device based on deep learning |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170169534A1 (en) * | 2015-12-09 | 2017-06-15 | One Concern, Inc. | Damage data propagation in predictor of structural damage |
CN107292406A (en) * | 2016-03-30 | 2017-10-24 | 中国石油化工股份有限公司 | Seismic properties method for optimizing based on vector regression and genetic algorithm |
-
2018
- 2018-10-11 CN CN201811184759.4A patent/CN109271734B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170169534A1 (en) * | 2015-12-09 | 2017-06-15 | One Concern, Inc. | Damage data propagation in predictor of structural damage |
CN107292406A (en) * | 2016-03-30 | 2017-10-24 | 中国石油化工股份有限公司 | Seismic properties method for optimizing based on vector regression and genetic algorithm |
Non-Patent Citations (1)
Title |
---|
唐小微;白旭;胡记磊;: "基于贝叶斯网络的自由场地震液化沉降评估" * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111400915A (en) * | 2020-03-17 | 2020-07-10 | 桂林理工大学 | Sand liquefaction discrimination method and device based on deep learning |
Also Published As
Publication number | Publication date |
---|---|
CN109271734B (en) | 2023-04-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104879103B (en) | A kind of seperated layer water injection effect analysis method | |
CN104615862B (en) | The method that high water cut oil field based on evolution algorithm determines well location | |
CN105389467B (en) | A kind of method and device obtaining interwell communication relationship | |
CN104730596A (en) | Discrete fracture modeling method based on multiscale factor restraint | |
US9460403B2 (en) | Methods and systems related to hydrocarbon recovery strategy development | |
CN105447762B (en) | A kind of calculation method of the low-permeability oil deposit water logging information of fluid replacement | |
CN106250984A (en) | The determination methods of the oil water relation pattern of oil well and device | |
KR102155101B1 (en) | System and method for predicting ground layer information, and a recording medium having computer readable program for executing the method | |
CN107895092A (en) | A kind of interwell communication quantitative evaluation method that modeling is adopted based on complex nonlinear note | |
CN106096081B (en) | The estimation method of reserve of fracture hole type bottom water reservoir | |
CN107944126A (en) | A kind of method and device of definite water cut in water flooding reservoir | |
CN103628816A (en) | Seismic attribute guidance horizontal-well-steering method and device thereof | |
CN103268423B (en) | A kind of method that geography phenomenon multi-point simulation spatial scale selects | |
CN112861890A (en) | Reservoir evaluation model construction method and reservoir identification method | |
CN109271734A (en) | A kind of earthquake-induced site liquefaction potential evaluation method based on machine learning techniques | |
CN106777117A (en) | A kind of automatic identifying method of horizontal stratum tectonic landform | |
CN105676288B (en) | A kind of sand body stacked relation method of discrimination based on reflectance factor analysis | |
KR102631470B1 (en) | System for predicting leakage and failure of bank | |
CN111611422B (en) | SVC-based method and system for automatically generating qualitative graph in earthquake disaster risk assessment | |
CN110309597B (en) | Method, device and storage medium for determining choked flow band model based on configuration interface | |
CN104834934B (en) | A kind of nucleome capture method for being used to identify reservoir fluid | |
Li et al. | Flow field characterization and evaluation method based on unsupervised machine learning | |
CN106150491B (en) | A kind of exploitation method and device of oil reservoir | |
WO2020055379A1 (en) | Method and system for reactively defining valve settings | |
CN110924908A (en) | Water drive reservoir injection-production parameter determination method and computer readable storage medium |
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 |