CN107905270A - A kind of Deformation Prediction in Deep Foundation Pit method - Google Patents
A kind of Deformation Prediction in Deep Foundation Pit method Download PDFInfo
- Publication number
- CN107905270A CN107905270A CN201711213401.5A CN201711213401A CN107905270A CN 107905270 A CN107905270 A CN 107905270A CN 201711213401 A CN201711213401 A CN 201711213401A CN 107905270 A CN107905270 A CN 107905270A
- Authority
- CN
- China
- Prior art keywords
- components
- deformation
- particle
- prediction
- foundation pit
- 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.)
- Pending
Links
Classifications
-
- E—FIXED CONSTRUCTIONS
- E02—HYDRAULIC ENGINEERING; FOUNDATIONS; SOIL SHIFTING
- E02D—FOUNDATIONS; EXCAVATIONS; EMBANKMENTS; UNDERGROUND OR UNDERWATER STRUCTURES
- E02D33/00—Testing foundations or foundation structures
-
- E—FIXED CONSTRUCTIONS
- E02—HYDRAULIC ENGINEERING; FOUNDATIONS; SOIL SHIFTING
- E02D—FOUNDATIONS; EXCAVATIONS; EMBANKMENTS; UNDERGROUND OR UNDERWATER STRUCTURES
- E02D1/00—Investigation of foundation soil in situ
Landscapes
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Mining & Mineral Resources (AREA)
- Paleontology (AREA)
- Civil Engineering (AREA)
- General Engineering & Computer Science (AREA)
- Structural Engineering (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Soil Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of Deformation Prediction in Deep Foundation Pit method, including:Foundation pit deformation monitoring time sequence is subjected to LMD decomposition, tries to achieve PF components;The prediction of subsequent time deformation of deep excavation is carried out to each PF components using least square method supporting vector machine;Subsequent time deformation of deep excavation corresponding to each PF components is superimposed as subsequent time Deformation Prediction in Deep Foundation Pit value.By the way that foundation pit deformation monitoring time sequence samples are carried out LMD decomposition, the information self-adapting of complicated non-stationary signal foundation its own signal is resolved into several production function component PF with actual physical meaning, reflect the signal distributions rule on space scale, then each PF components are respectively adopted the prediction that least square method supporting vector machine carries out subsequent time deformation of deep excavation;And the parameter in least square method supporting vector machine is optimized using particle swarm optimization algorithm, the subsequent time deformation of deep excavation corresponding to for minimum support vector machines each PF components obtained from of each PF components is obtained, superposition obtains final result.
Description
Technical field
The invention belongs to base pit engineering technical field, more particularly to a kind of Deformation Prediction in Deep Foundation Pit method.
Background technology
Foundation pit deformation is the key factor for influencing base pit engineering safe construction, grasp foundation pit deformation feelings that can be promptly and accurately
Condition is to realize the effective way of deep pit monitor early warning.Excavation of foundation pit is the process of a soil body off-load, native during off-load
The stress balance of body is destroyed, this is an important factor for causing foundation pit deformation.In addition, can be subject in the construction process construction because
The influence of element, environmental factor and time factor, the prediction work to foundation pit deformation bring some difficult.At present, base pit engineering is more
Numerical digit is complicated in Adjacent Buildings, densely populated area, once accident, which occurs, will result in imponderable consequence.Therefore,
The research of Deep Foundation Distortion Forecast forecast is particularly important.
In recent decades, scholars to Deep Foundation Distortion Forecast study and acquire a great achievement.When being usually used in monitoring
Between Series Modeling predict method mainly have:The methods of neutral net, support vector machines, time series analysis.But these methods
It is theoretical with go back Shortcomings in application in place of, as grey forecasting model is in most cases more coarse, model predictive error
It is larger;Neutral net obtains more application in recent years, but the generalization ability of model is not high;Support vector machines kernel functional parameter
Choose and have a certain impact to prediction result tool.It is therefore desirable to propose a kind of model that can reduce prediction error, accomplish more
Nearly accurate foundation pit side shape prediction, lays the foundation for the safe construction of base pit engineering.
The content of the invention
In view of the deficiencies of the prior art, the present invention provides a kind of Deformation Prediction in Deep Foundation Pit method.
Technical solution is as follows:
A kind of Deformation Prediction in Deep Foundation Pit method, including:
Foundation pit deformation monitoring time sequence is subjected to LMD decomposition, tries to achieve PF components;
The prediction of subsequent time deformation of deep excavation is carried out to each PF components using least square method supporting vector machine;
Subsequent time deformation of deep excavation corresponding to each PF components is superimposed as subsequent time Deformation Prediction in Deep Foundation Pit value.
The prediction for carrying out subsequent time deformation of deep excavation to each PF components using least square method supporting vector machine, bag
Include:
The deformation of deep excavation data being utilized respectively in each PF components, the minimum support vector machines corresponding to each PF components of training
Model;
The kernel function in each minimum supporting vector machine model and penalty coefficient are optimized using particle swarm optimization algorithm;
Each PF components are predicted using the minimum supporting vector machine model after being optimized using particle swarm optimization algorithm,
Obtain the prediction result of each PF components, i.e., the subsequent time deformation of deep excavation corresponding to each PF components.
It is described that the kernel function in each minimum supporting vector machine model and penalty coefficient are carried out using particle swarm optimization algorithm
Optimization, including:
Population is initialized;
One group of kernel function, penalty coefficient are subjected to population iteration as a particle;
The current fitness function value of each particle is calculated, if it is optimal suitable that the current fitness function value of particle is less than its
Response functional value, then using the position of the particle corresponding to adaptive optimal control degree functional value as the current location of particle, otherwise continue
Iteration;
With the adaptive optimal control degree functional value of each particle compared with the adaptive optimal control degree functional value of all particles, look for
Go out the optimal location of particle in population, and update particle rapidity, position and inertia weight;
Judge whether to meet iterations or the end condition of fitness function value:If satisfied, then iteration optimizing terminates, obtain
To optimal kernel function, penalty coefficient, substitute into minimum supporting vector machine model, after being utilized particle swarm optimization algorithm optimization
Minimum supporting vector machine model, otherwise continue iteration.
Beneficial effect:
The present invention provides a kind of Deformation Prediction in Deep Foundation Pit method, by the way that foundation pit deformation monitoring time sequence samples are carried out
LMD is decomposed, and the information self-adapting of complicated non-stationary signal foundation its own signal is resolved into several with actual physical
The production function component PF (Production Function) of meaning, so that reflect the signal distributions rule on space scale,
Carry out the prediction of subsequent time deformation of deep excavation respectively to each PF components again, improve precision of prediction;By monitoring data after LMD decomposition
Some Small Sample Databases have been divided into it, have been predicted using minimum support vector machines, and the parameter in minimum support vector machines has been used
After particle swarm optimization algorithm optimization, the thus obtained minimum support vector machines for each PF components is all optimal, and then
Subsequent time deformation of deep excavation accuracy corresponding to each PF components arrived is high, then is superimposed to obtain final prediction result, due to
Prediction result is accurate, therefore can instruct the safe construction of base pit engineering well.
Brief description of the drawings
Fig. 1 is a kind of Deformation Prediction in Deep Foundation Pit method flow diagram of the present invention;
Fig. 2 is different Forecasting Methodology Comparative result curves in the embodiment of the present invention.
Embodiment
Below in conjunction with the accompanying drawings and specific embodiment elaborates technical scheme.
The method of the present invention synthesis uses LMD (Local Mean Decomposition), LSSVM (Least Squares
Support Vector Machines) and PSO (Particle Swarm Optimization) methods model and pre- depth measurement
Foundation pit deformation, is decomposed into multiple PF components with LMD methods by monitoring time sequence first, then passes through particle swarm optimization algorithm pair
Minimum supporting vector machine model parameter optimizes, and carries out rolling forecast to each component with the model after optimization, then pass through
Superposition reconstructs each component prediction result, obtains the accurate prediction result of foundation pit deformation.
A kind of Deformation Prediction in Deep Foundation Pit method as shown in Figure 1, including:
(1) foundation pit deformation monitoring time sequence is set as u (t), t=1,2 ... n, i.e., the deep base gathered in different time points
Deformation data is cheated, u (t) is subjected to LMD decomposition, tries to achieve PF (Production Function) component, and each PF components include
The effective information of sample data u (t).
(2) prediction of subsequent time deformation of deep excavation is carried out to each PF components using least square method supporting vector machine;
Specific Forecasting Methodology includes:
(2.1) the deformation of deep excavation data being utilized respectively in each PF components, the minimum support corresponding to each PF components of training
Vector machine model;
(2.2) using particle swarm optimization algorithm to the kernel function σ in each minimum supporting vector machine model and penalty coefficient d into
Row optimization.
Specific optimization method, including:
1. population is initialized:Operating parameter is set, including:Particle random site and speed, iterations, population
Scale, fitness function;Determine Optimal Parameters --- kernel function σ, the value range of penalty coefficient d.
Using one group of kernel function σ, penalty coefficient d as a particle, it is m=30 to set particle populations scale, iterations
For 200 times, the value range for determining σ, d is respectively { 3,5;100,300 }.
2. calculate the current fitness function value f (x of each particlei), wherein xiThe position of i-th of particle is represented, by particle
Current fitness function value f (xi) and its adaptive optimal control degree functional value f (pbesti) be compared:If f (xi) < f
(pbesti), then xi=pbesti, i.e.,:Present bit using the position of the particle corresponding to adaptive optimal control degree functional value as particle
Put;Otherwise, iteration is continued;
3. with the adaptive optimal control degree functional value of each particle compared with the adaptive optimal control degree functional value of all particles,
The optimal location of particle in population is found out, and updates particle rapidity, position and inertia weight.
4. judge whether to meet iterations or the end condition of fitness function value:If satisfied, then iteration optimizing terminates,
Optimal kernel function σ, penalty coefficient d are obtained, substitutes into minimum supporting vector machine model, it is excellent to be utilized particle swarm optimization algorithm
Minimum supporting vector machine model-PSO-LSSVM (Particle Swarm Optimization-Least Squares after change
Support Vector Machine) model.
(2.3) each PF components are carried out using the minimum supporting vector machine model after being optimized using particle swarm optimization algorithm
Prediction, obtains the prediction result of each PF components, i.e., the subsequent time deformation of deep excavation corresponding to each PF components.
(3) the subsequent time deformation of deep excavation corresponding to each PF components is superimposed as subsequent time Deformation Prediction in Deep Foundation Pit
Value.
Analysis is predicted to the foundation pit deformation of Fuxin base pit engineering project using the above method.Excavation of foundation pit depth is about
11.8m, the support pattern of this engineering is support pile+prestress anchorage cable.Miscellaneous fill, silt, coarse sand and gravel is distributed with place underground
Sand etc..It is right in the construction process in order to ensure the safety of foundation pit and Adjacent Buildings since Adjacent Buildings are nearer apart from foundation pit
Foundation pit deformation has carried out monitoring.Present embodiment is predicted exemplified by choosing the measured data of W-12 monitoring points.Choose
25 days-July 5 April in 2013, the monitoring result of totally 43 times carried out analysis prediction.
Table 1 utilizes the prediction result of the minimum supporting vector machine model after particle swarm optimization algorithm optimization
The different model prediction results of table 2
The different model consensus forecast results of table 3
It can be seen that by table 1-3 and try to achieve predicted value result with the method for the present invention:Maximum relative error is 0.4%, average
Relative error is 0.1039%, and maximum absolute error is that 0.0990mm mean absolute errors are 0.0427mm,.And without LMD
Decompose the predicted value result directly obtained with PSO-LSSVM models:Maximum relative error is 0.736%, average to miss relatively
Difference is 0.3892%, maximum absolute error 0.1987mm, mean absolute error 0.0942mm,.Shown by comparing result,
In the method for the present invention using LMD decompose after each PF components are respectively adopted be superimposed after PSO-LSSVM model predictions reconstruct it is pre-
Survey result precision higher.PSO-LSSVM models are larger to the fluctuation of nonlinear Deep Foundation Distortion Forecast result, and pass through LMD
Decompose so that the method for the present invention possesses preferable capability of fitting and generalization ability, foundation pit deformation can be reduced by reflecting LMD decomposition
Influence of the nonlinear characteristic to prediction result.
Claims (3)
- A kind of 1. Deformation Prediction in Deep Foundation Pit method, it is characterised in that including:Foundation pit deformation monitoring time sequence is subjected to LMD decomposition, tries to achieve PF components;The prediction of subsequent time deformation of deep excavation is carried out to each PF components using least square method supporting vector machine;Subsequent time deformation of deep excavation corresponding to each PF components is superimposed as subsequent time Deformation Prediction in Deep Foundation Pit value.
- 2. according to the method described in claim 1, it is characterized in that, described use least square method supporting vector machine to each PF components The prediction of subsequent time deformation of deep excavation is carried out, including:The deformation of deep excavation data being utilized respectively in each PF components, the minimum support vector machines mould corresponding to each PF components of training Type;The kernel function in each minimum supporting vector machine model and penalty coefficient are optimized using particle swarm optimization algorithm;Each PF components are predicted using the minimum supporting vector machine model after being optimized using particle swarm optimization algorithm, are obtained The prediction result of each PF components, i.e., the subsequent time deformation of deep excavation corresponding to each PF components.
- 3. according to the method described in claim 1, it is characterized in that, it is described using particle swarm optimization algorithm each minimum is supported to Kernel function and penalty coefficient in amount machine model optimize, including:Population is initialized;One group of kernel function, penalty coefficient are subjected to population iteration as a particle;The current fitness function value of each particle is calculated, if the current fitness function value of particle is less than its adaptive optimal control degree Functional value, then using the position of the particle corresponding to adaptive optimal control degree functional value as the current location of particle, otherwise continue iteration;With the adaptive optimal control degree functional value of each particle compared with the adaptive optimal control degree functional value of all particles, kind is found out The optimal location of particle in group, and update particle rapidity, position and inertia weight;Judge whether to meet iterations or the end condition of fitness function value:If satisfied, then iteration optimizing terminates, obtain most Excellent kernel function, penalty coefficient, substitute into minimum supporting vector machine model, are utilized after particle swarm optimization algorithm optimization most Small supporting vector machine model, otherwise continues iteration.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711213401.5A CN107905270A (en) | 2017-11-28 | 2017-11-28 | A kind of Deformation Prediction in Deep Foundation Pit method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711213401.5A CN107905270A (en) | 2017-11-28 | 2017-11-28 | A kind of Deformation Prediction in Deep Foundation Pit method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107905270A true CN107905270A (en) | 2018-04-13 |
Family
ID=61848813
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711213401.5A Pending CN107905270A (en) | 2017-11-28 | 2017-11-28 | A kind of Deformation Prediction in Deep Foundation Pit method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107905270A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110042874A (en) * | 2019-05-13 | 2019-07-23 | 燕山大学 | A kind of deep basal pit safety detection method and system |
CN113139228A (en) * | 2021-04-22 | 2021-07-20 | 南京智慧岩土工程技术研究院有限公司 | Monitoring point arrangement optimization method for large-span foundation pit complex support system structure |
CN115169243A (en) * | 2022-07-28 | 2022-10-11 | 中铁三局集团有限公司 | GA-PSO-GLSSVM algorithm-based soil-rock composite stratum deep foundation pit deformation time sequence prediction method |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWM370768U (en) * | 2009-06-03 | 2009-12-11 | Qing-Chang Weng | Acceleration device for particle swarm optimization (PSO) |
CN102305891A (en) * | 2011-07-04 | 2012-01-04 | 武汉大学 | On-line monitoring method of low-frequency oscillation of power system |
CN102663412A (en) * | 2012-02-27 | 2012-09-12 | 浙江大学 | Power equipment current-carrying fault trend prediction method based on least squares support vector machine |
CN104952226A (en) * | 2014-03-31 | 2015-09-30 | 中铁西北科学研究院有限公司深圳南方分院 | Wireless testing method, testing device and testing system for deformation of deep pit |
CN105862935A (en) * | 2016-04-12 | 2016-08-17 | 陕西理工学院 | Damage recognition method used for retaining wall structural system |
CN106407581A (en) * | 2016-09-28 | 2017-02-15 | 华中科技大学 | Intelligent prediction method for ground surface settlement induced by subway tunnel construction |
-
2017
- 2017-11-28 CN CN201711213401.5A patent/CN107905270A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWM370768U (en) * | 2009-06-03 | 2009-12-11 | Qing-Chang Weng | Acceleration device for particle swarm optimization (PSO) |
CN102305891A (en) * | 2011-07-04 | 2012-01-04 | 武汉大学 | On-line monitoring method of low-frequency oscillation of power system |
CN102663412A (en) * | 2012-02-27 | 2012-09-12 | 浙江大学 | Power equipment current-carrying fault trend prediction method based on least squares support vector machine |
CN104952226A (en) * | 2014-03-31 | 2015-09-30 | 中铁西北科学研究院有限公司深圳南方分院 | Wireless testing method, testing device and testing system for deformation of deep pit |
CN105862935A (en) * | 2016-04-12 | 2016-08-17 | 陕西理工学院 | Damage recognition method used for retaining wall structural system |
CN106407581A (en) * | 2016-09-28 | 2017-02-15 | 华中科技大学 | Intelligent prediction method for ground surface settlement induced by subway tunnel construction |
Non-Patent Citations (4)
Title |
---|
张俊红等: "基于LMD 和SVM 的柴油机气门故障诊断", 《内燃机学报》 * |
罗亦泳等: "基于改进LMD的大坝变形特征提取分析", 《江西科学》 * |
蔡道勇: "大型机械加工设备轴承故障诊断方法研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
郑志成等: "基于混合核函数PSO-LSSVM 的边坡变形预测", 《岩土力学》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110042874A (en) * | 2019-05-13 | 2019-07-23 | 燕山大学 | A kind of deep basal pit safety detection method and system |
CN113139228A (en) * | 2021-04-22 | 2021-07-20 | 南京智慧岩土工程技术研究院有限公司 | Monitoring point arrangement optimization method for large-span foundation pit complex support system structure |
CN113139228B (en) * | 2021-04-22 | 2022-08-12 | 南京智慧岩土工程技术研究院有限公司 | Monitoring point arrangement optimization method for large-span foundation pit complex support system structure |
CN115169243A (en) * | 2022-07-28 | 2022-10-11 | 中铁三局集团有限公司 | GA-PSO-GLSSVM algorithm-based soil-rock composite stratum deep foundation pit deformation time sequence prediction method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Kordnaeij et al. | Prediction of recompression index using GMDH-type neural network based on geotechnical soil properties | |
CN101344389B (en) | Method for estimating tunnel surrounding rock displacement by neural network | |
CN107905270A (en) | A kind of Deformation Prediction in Deep Foundation Pit method | |
CN101847171A (en) | Back analysis method of slope displacement based on safety monitoring | |
CN105606063A (en) | Soil layer slope stability determining method based on orthogonal strain ratio | |
CN111382472A (en) | Method and device for predicting shield-induced proximity structure deformation by random forest fusion SVM (support vector machine) | |
CN109632016A (en) | Rock And Soil adit digging and surrouding rock stress, strain monitoring experimental rig and its method | |
CN111365051B (en) | Method for estimating stress of carbonaceous rock tunnel anchor rod based on transfer function of feedback algorithm | |
CN115659749A (en) | Foundation pit deformation prediction method and system, electronic equipment and storage medium | |
CN114548482A (en) | Creep type landslide kinetic energy change rate face-slip early warning method | |
CN105893329B (en) | A kind of tide gauge consistent correction method based on moon yardstick | |
CN112883478A (en) | Steel structure displacement prediction method and device, terminal equipment and system | |
CN117150925A (en) | Reverse analysis method for rock mass mechanical parameters of high-steep slope of hydropower engineering | |
CN111582634A (en) | Multi-factor safety grading method and system for underground large-space construction | |
CN114491730B (en) | Dynamic stability analysis iteration method and device for high-speed railway roadbed structure | |
Willems | Stochastic generation of spatial rainfall for urban drainage areas | |
CN115809598A (en) | Slope deformation prediction and control value correction method based on safety coefficient | |
CN106934729A (en) | Building Testing and appraisal method and device | |
He et al. | Estimation of unloading relaxation depth of Baihetan Arch Dam foundation using long-short term memory network | |
CN111797577A (en) | Method and system for evaluating adaptability of typical remediation engineering of estuary and river network | |
CN206627073U (en) | The quick monitoring and forecasting device of underground engineering construction based on 3D laser scannings | |
CN104951597B (en) | A kind of Forecasting Methodology of underwater sound signal | |
CN106640066B (en) | A kind of method of determining girdle flitting fully mechanized mining mode | |
Qian et al. | Damage Identification and Comprehensive Safety Evaluation of Artificial Neural Network for High-rise Buildings | |
CN116776707A (en) | Foundation pit structure digital numerical analysis method based on geological drilling |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180413 |
|
RJ01 | Rejection of invention patent application after publication |