CN108446413A - A kind of optimization assay method of injection forming club-footed pile stake diameter - Google Patents
A kind of optimization assay method of injection forming club-footed pile stake diameter Download PDFInfo
- Publication number
- CN108446413A CN108446413A CN201711276442.9A CN201711276442A CN108446413A CN 108446413 A CN108446413 A CN 108446413A CN 201711276442 A CN201711276442 A CN 201711276442A CN 108446413 A CN108446413 A CN 108446413A
- Authority
- CN
- China
- Prior art keywords
- base expanding
- pile
- injection forming
- club
- stake diameter
- 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]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Abstract
Method for measuring is optimized to injection forming club-footed pile stake diameter the present invention relates to a kind of, is included the following steps：Step 1：Influence the determination of the principal element of club-footed pile anti-pulling capacity；Step 2：The determination of the optimal expanding ratio of injection forming club-footed pile；Step 3：The determination of injection forming base expanding and base expanding pile bearing capacity BP neural network model；Step 4：The determination of injection forming base expanding and base expanding pile bearing capacity BP neural network model parameter；Step 5：The determination of injection forming club-footed pile anti-pulling capacity and stake diameter relation curve；Step 6：The determination of injection forming club-footed pile stake diameter.The method of the present invention is capable of the load bearing mechanism of definite simulation stake, and the bearing capacity obtained and measured value error are small, and scientific rationality foundation is provided for the design of stake diameter, and then can carry out accurate stake diameter design.
Description
Technical field
The invention belongs to the optimizations of injection forming club-footed pile stake diameter to measure field, and in particular to one kind is expanded based on injection forming
Bottom stake anti-pulling capacity BP neural network prediction model carries out the assay method of stake diameter optimization.
Background technology
With the development of economic construction of China, land resource growing tension, urban skyscraper is increased rapidly, right in recent years
The competition of land resource gradually turns to underground by ground, and the utilization of the underground space are as the important interior of current engineering design
Hold, especially municipal works field, since land acquisition, removal difficulty constantly increase, various underground installations (hand over by city rail
Logical, river-crossing tunnel, sunk type square and all kinds of water tank purification facilities etc.) development and utilization be even more to become that alleviate soil rare
Contradictory effective means.But in the coastal riparian area of south China, since groundwater level is generally higher, hypogee is being constructed
Or service stage often bears the effect of larger buoyancy force of underground water, so that become in underground space development, design must be through for Anti-floating design
The problem often considered.The considerations of being typically based on waterproof, geology and drawing anchor durability, mostly uses uplift pile to help in anti-floating
Help anti-floating.Uplift pile since stability is high, easily construction, applied widely and be widely adopted, thus the resistance to plucking problem of pile foundation at
For a new focus of engineering circles.And injection forming club-footed pile is a kind of pile-type that load-carrying efficiency is high, it is expanded by stake bottom
Major part increases effective bearing area, improves bearing capacity of pile foundation, and then significantly improves material use efficiency, therefore by adopting
Withdrawal resistance is improved as one of the effective scheme of pile foundation resistance to plucking is solved the problems, such as, so how to slip casting with injection forming club-footed pile
It is molded club-footed pile stake diameter and carries out Exact Design with important engineering application value.
But since the factor for influencing injection forming base expanding and base expanding pile bearing capacity is many and unstable, how reasonably to set
It counts stake diameter and accurately determines the bearing capacity of stake, give full play to the technical economic benefit of pile foundation, be engineering design personnel always
Very concern.Complexity due to its interaction mechanism between by engineering geological condition, stake and pile peripheral earth etc. is all
The Carrying Capacity of multifactor influence, stake is never well solved, therefore it is reasonable to cause the design of a diameter to lack
Property foundation.The design of uplift pile calculates all without very specific regulation in each specification, and designer often rule of thumb and understands
It is calculated, the design result difference of different personnel is very big.
The design of current uplift pile calculates the design experiences for relying primarily on designer, using limit equilibrium theory, cunning
Move the calculating that the numerical computation methods such as field theory, pile soil common action theory, finite element carry out bearing capacity of single pile, but these methods
Have limitation, cannot definite simulation stake load bearing mechanism, the bearing capacity obtained has certain error with actual measurement so that stake diameter
Design lack scientific rationality foundation.How stake diameter and bearing capacity established into corresponding correlativity, and relatively more accurate in turn
Progress stake diameter a great problem for being designed to face at present for designer.
Invention content
For existing for above-mentioned traditional design method cannot definite simulation stake load bearing mechanism, the bearing capacity obtained and reality
Survey has certain error, cannot carry out the problem of accurate stake diameter designs, this patent proposes a kind of to injection forming club-footed pile
The new method that stake diameter optimization measures, i.e., predicted and optimized to injection forming base expanding and base expanding pile bearing capacity with BP neural network modeling
The method of stake diameter design.
It is a kind of that method for measuring is optimized to injection forming club-footed pile stake diameter, specifically include following steps：
Step 1：Influence the determination of the principal element of club-footed pile anti-pulling capacity；
Step 2：The determination of the optimal expanding ratio of injection forming club-footed pile；
Step 3：The determination of injection forming base expanding and base expanding pile bearing capacity BP neural network model；
Step 4：The determination of injection forming base expanding and base expanding pile bearing capacity BP neural network model parameter；
Step 5：The determination of injection forming club-footed pile anti-pulling capacity and stake diameter relation curve；
Step 6：The determination of injection forming club-footed pile stake diameter.
Further, the principal element that the step 1 influences club-footed pile anti-pulling capacity is respectively soil layer liquidity index
IL, Pile side soil effective angle of inner friction weighted averageThe deformation modulus E of expanding part soil layerS, the long l of stake, non-base expanding and base expanding part stake
The extension of the ratio between diameter d, base expanding and base expanding height h, base expanding and base expanding part stake diameter and non-base expanding and base expanding part stake diameter D/d and enlarged footing upper diameter transition
Angle beta,；D/d is indicated with λ.
Further, injection forming club-footed pile is optimal expanding than being determined by following formula in the step 2,
Wherein, ρtFor the expanding more expanding single pile more corresponding than λ=t-0.2 of single pile Ultimate Up-lift Bearing Capacity more corresponding than λ=t
The growth rate of Ultimate Up-lift Bearing Capacity；TtFor expanding single pile Ultimate Up-lift Bearing Capacity more corresponding than λ=t；Tt-0.2For it is expanding than λ=
The corresponding single pile Ultimate Up-lift Bearing Capacities of t-0.2；It is corresponding when single pile Ultimate Up-lift Bearing Capacity raising efficiency reaches maximum value
It is expanding than be it is optimal expanding than λ '.
Further, the step 3 BP neural network structural model is made of input layer, hidden layer and output layer.
Further, the input layer is 8 nodes, by stake long vector group [l1,l2,Λ,ln], non-base expanding and base expanding part stake diameter
Vector Groups [d1,d2,Λ,dn], base expanding and base expanding height vector group [h1,h2,Λ,hn], base expanding and base expanding part stake diameter and non-base expanding and base expanding part stake diameter it
Than Vector Groups [λ1,λ2,Λ,λn], soil layer liquidity index Vector Groups [IL1,IL2,Λ,ILn], Pile side soil effective angle of inner friction weighting
Average value vector groupDeformation modulus Vector Groups [the E of expanding part soil layers1,Es2,Λ,Esn], enlarged footing top
Extended corner Vector Groups [the β of Diameter Gradual Change section1,β2,Λ,βn] eight factors indicate；Hidden layer is single hidden layer；Output layer unit
Number is 1, by injection forming club-footed pile anti-pulling capacity Vector GroupsIt indicates；Input layer is to hidden layer
Weights and threshold value be respectively wjkAnd θjk, the weights and threshold value of hidden layer to output layer are respectively vkjAnd γkt。
Further, the determination method of the step 4 BP neural network model parameter, specially：
(1) trial and error procedure determines hidden layer neuron nodal point number
First according to [2n-4,2n+6] number of influence factor (n be) determine hidden layer neuron nodal point number range [12,
22]；Sample data is inputted into the trial and error procedure training program worked out in Matlab, to hidden layer neuron nodal point number [12,22] into
Row tentative calculation；Selection mean square error hidden layer neuron nodal point number corresponding with when restraining step number minimum is the implicit of final network
Layer neuron nodal point number；
(2) determination of BP artificial neural networks weights and threshold value
1) sample size is determined according to N >=2n, and the quantity of training sample and test samples is determined by formula (2), (3)：
N-- is the total quantity of sample；T1-- the quantity of training sample；T2-- the quantity of the sample of inspection；[] indicates rounding
Operation；
2) random value initializes weights and threshold value in (0,1) section first, then to training sample data into
Row normalized, then passes to hidden layer neuron by formula (4), and hidden layer neuron carries out defeated according to formula (5)
Go out：
bk=f (S (k))=1/ (1+e-S(k)) (k=1,2, Λ, p) (5)
Wherein, PjFor input vector group, j is the number of element in input vector group, and k is the number of implicit layer unit；
3) hidden layer neuron output valve passes to output layer neuron by formula (6), and output layer neuron is according to public affairs
Formula (7) is exported：
4) network error e is calculated according to formula (8)：
Wherein, t is the number of element in output vector group；
As network global error e ＜ 0.0006, meet required precision, terminates network training；When network error e >=
When 0.0006, be unsatisfactory for required precision, need according to the following steps 5), 6) weights of network and threshold value are modified；
5) according to desired output TUKtWith network reality outputThe correction of output layer neuron is calculated using formula (9)
Error dt：
The correction error ek of hidden layer neuron is calculated according to formula (10)：
6) hidden layer is corrected to the connection weight v of output layer according to formula (11), (12)ktWith the threshold value of output layer neuron
γkt, wherein α is learning rate, 0 ＜ α ＜ 1
vkt(i+1)=vkt(i)+α·dt·bk(k=1,2, Λ, p；T=1,2, Λ, n) (11)
γkt(i+1)=γkt(i)+α·dt(t=1,2, Λ, n) (12)
Input layer is corrected to the connection weight W of hidden layer according to formula (13), (14)jkWith the threshold value of hidden layer neuron
θjk, wherein β is learning rate, 0 ＜ β ＜ 1
wjk(i+1)=wjk(i)+β·ek·Pj(j=1,2, Λ, n；K=1,2, Λ, p) (13)
θjk(i+1)=θjk(i)+β·ek(k=1,2, Λ, p) (14)
7) randomly select next mode of learning return the 3) step continue to train, until network error e meets precision
It is required that terminating network training, the weights and threshold value of the neural network are determined；
(3) establishment of neural network procedure
According to above-mentioned determining BP neural network model, and Matlab Neural Network Toolbox is used, to injection forming base expanding and base expanding
Stake anti-pulling capacity neural network prediction program is worked out.
Further, the determination method of the step 5 injection forming club-footed pile anti-pulling capacity and stake diameter relation curve has
Body is：
It is reconnoitred by field geology conditions, determines soil layer property parameter；According to the achievement in research of club-footed pile bearer properties and
《Code for design of building》(GB50007-2011) design code of pile foundation plinth determines the long l of stake, base expanding and base expanding height h=
The ratio between 0.6D=1.2d, base expanding and base expanding part stake diameter and non-base expanding and base expanding part stake diameter λ ', enlarged footing upper diameter transition extended corner β=
90°；According to《Code for design of building》(GB50007-2011) the non-base expanding and base expanding part stake diameter d ' of Preliminary design, and input
The neural network prediction model that training is completed carries out anti-pulling capacity prediction, according to the mistake of anti-pulling capacity predicted value and design value
Difference selects rational spacingIn non-base expanding and base expanding part, stake diameter Preliminary design value d ' neighborhoods carry out value (d ' -10
+ 10 μ of μ, Λ, d '-μ, d ', d '+μ, Λ, d '), and input and the neural network prediction model completed has been trained to carry out anti-pulling capacity
Prediction draws injection forming according to the relationship of the anti-pulling capacity of neural network prediction model output and non-base expanding and base expanding part stake diameter
The non-club-footed pile diameter of club-footed pile and anti-pulling capacity relation curve.
Further, the corresponding non-base expanding and base expanding of injection forming club-footed pile anti-pulling capacity design value is determined using linear interpolation method
Stake diameter d in part determines base expanding and base expanding part stake diameter D according to injection forming club-footed pile non-base expanding and base expanding part stake diameter d and optimal expanding than λ '：
D=λ ' d (16).
Further, the linear interpolation method the specific steps are：
(1) it according to injection forming club-footed pile anti-pulling capacity design value determined by the curve graph of step 5 drafting, determines
Go out the section [d '+k μ, d '+(k+1) μ] (k=0,1,2,3,4,5,6,7,8,9) where corresponding non-base expanding and base expanding part stake diameter d；
(2) according to bent in formula (15) determination section [d '+k μ, d '+(k+1) μ] (k=0,1,2,3,4,5,6,7,8,9)
The corresponding anti-pulling capacity linear equation of line,
The respectively corresponding anti-pulling capacities of non-base expanding and base expanding part stake diameter d '+k μ, the d '+corresponding resistance to pluckings of (k+1) μ
Bearing capacity；
(3) by injection forming club-footed pile anti-pulling capacity design value TUKIt substitutes into formula (15) and calculates corresponding non-base expanding and base expanding part
Stake diameter d, and non-base expanding and base expanding part stake diameter d input neural network prediction models are obtained into its anti-pulling capacity predicted value, utilize formulaIts relative error with design value is calculated, if error, within 5%, which, which meets design, wants
It asks；If error is less than 0 or more than 5%, which does not meet design requirement, appropriate to adjust stake diameter d, repeats
It inputs neural network prediction model and carries out error-tested, until error meets the requirements, determine the non-base expanding and base expanding of injection forming club-footed pile
Part stake diameter d.
The principle of the present invention is as follows：
Principle one：
MATLAB training programs based on trial and error procedure are as follows：
Principle two：
Input data normalization pretreatment and output data post-processing
Input data is normalized using formula (17), it is made to be distributed in section [0,1].
Wherein,For the data after normalization；X is initial data；Xmax is the maximum value of x；xminIt is the minimum value of x.
For the high stake diameter data use of similarity degree surely than diminution mode, whole divided by 1 000 makes them also all be distributed
Between [0,1].
Anti-normalization processing is carried out to neural network output data using formula (18)：
X=x ' × (xmax-xmin)+xmin (18)
Principle three：
BP neural network Prediction program
Defining input sample is
Defining target vector is
Compared with prior art, the present invention predicts mould by establishing injection forming club-footed pile anti-pulling capacity BP neural network
Type, and BP neural network prediction model being trained using field test field data, in conjunction with BP neural network it is adaptive,
The advantages that precision is high, highly practical establishes prediction result and the more identical prediction model of actual carrying capacity, finally by BP nerves
Network Prediction Model draws the relation curve of injection forming club-footed pile stake diameter and anti-pulling capacity, and utilizes the method for function approximation
And then determine the non-base expanding and base expanding part stake diameter d of injection forming club-footed pile, the optimization design to injection forming club-footed pile stake diameter is realized, is
Designer provides a kind of injection forming club-footed pile design method of practicality.The method of the present invention is capable of definite simulation stake
Load bearing mechanism, the bearing capacity obtained and measured value error are small, and scientific rationality foundation is provided for the design of stake diameter, and then can be into
The accurate stake diameter design of row.
Description of the drawings
Fig. 1 is flow chart of the present invention；
Fig. 2 is that 1 single pile Ultimate Up-lift Bearing Capacity growth rate of embodiment compares graph of relation with expanding；
Fig. 3 is embodiment 1BP neural network model figures；
Fig. 4 is embodiment 1BP neural computing flow charts；
Fig. 5 is 1 anti-pulling capacity of embodiment and non-club-footed pile diameter graph of relation；
Fig. 6 is 2 anti-pulling capacity of embodiment and non-club-footed pile diameter graph of relation.
Specific implementation mode
The present invention is described in further details with reference to specific embodiment and attached drawing.
Embodiment 1
Certain pile foundation is located at the areas certain A, and the design of resistance to plucking engineering pile uses injection forming club-footed pile.
Step 1：Determining influences the principal element of club-footed pile anti-pulling capacity
According to《Technical code for building pile foundation》(JGJ94-2008)、《Building pile foundation inspection specifications》(JGJ106-
2014) and injection forming club-footed pile practical engineering experience, and the anti-pulling capacity characteristic of research injection forming club-footed pile with
On the basis of load transfer mechanism, it is determined that the influence factor of injection forming club-footed pile anti-pulling capacity, i.e. soil layer liquidity index
IL, Pile side soil effective angle of inner friction weighted averageThe deformation modulus E of expanding part soil layerS, the long l of stake, non-base expanding and base expanding part stake
The ratio between diameter d, base expanding and base expanding height h, base expanding and base expanding part stake diameter and non-base expanding and base expanding part stake diameter D/d (being indicated with λ), enlarged footing upper diameter gradual change
Eight factors of extended corner β of section are to influence the principal element of injection forming club-footed pile anti-pulling capacity.
Step 2：The determination of the optimal expanding ratio of injection forming club-footed pile
According to《Architecture foundation pile inspection specifications》(JGJ106-2014) the unit for single-pile vertical anti-pulling static test regulation in,
The single pile Ultimate Up-lift Bearing Capacity for measuring same geological conditions area injection forming club-footed pile takes the long l=37m of stake, non-base expanding and base expanding portion
Dividing stake diameter d=300mm, base expanding and base expanding height h=360mm, expanding ratio is respectively 1,1.2,1.4,1.6,1.8,2.0,2.2,2.4,
2.6,2.8,3.0, the growth rate ρ of injection forming club-footed pile single pile Ultimate Up-lift Bearing Capacity is calculated according to formula (1)t, as a result see
Table 1.
The growth rate of 1 single pile Ultimate Up-lift Bearing Capacity of table
Many experimental results show that the growth rate of single pile Ultimate Up-lift Bearing Capacity is expanding than in first increases and then decreases with its
Trend (see Fig. 2), therefore corresponding expanding ratio is as optimal when single pile Ultimate Up-lift Bearing Capacity raising efficiency reaches maximum value
It is expanding than λ ', it is optimal expanding than λ '=2 as shown in Table 1.
Step 3：The determination of injection forming base expanding and base expanding pile bearing capacity BP neural network structural model
According to BP neural network prediction technique modeling principle, this patent determines that BP neural network model is made of 3 parts, packet
Input layer, hidden layer and output layer are included, it is final to determine that input layer is 8 nodes, by stake long vector group [l1,l2,Λ,ln], non-expansion
Stake radius vector group [d is divided in bottom1,d2,Λ,dn], base expanding and base expanding height vector group [h1,h2,Λ,hn], the stake of base expanding and base expanding part diameter and non-base expanding and base expanding
The ratio between part stake diameter Vector Groups [λ1,λ2,Λ,λn], soil layer liquidity index Vector Groups [IL1,IL2,Λ,ILn], Pile side soil effectively in
Angle of friction weighted average Vector GroupsDeformation modulus Vector Groups [the E of expanding part soil layers1,Es2,Λ,Esn]、
Extended corner Vector Groups [the β of enlarged footing upper diameter transition1,β2,Λ,βn] eight factors indicate；Hidden layer is single hidden layer；
Output layer unit number is 1, by injection forming club-footed pile anti-pulling capacity Vector GroupsIt indicates；Input
The weights and threshold value of layer to hidden layer are respectively wjkAnd θjk, the weights and threshold value of hidden layer to output layer are respectively vkjAnd γkt。
The BP neural network model established is shown in attached drawing 3.
Step 4：The determination of injection forming base expanding and base expanding pile bearing capacity BP neural network model parameter
It determines that sample size is 26 according to N >=2n, determines that training samples number is 15 and examines by formula (2), (3)
Sample size is 11.Sample data is shown in Table 2.
2 sample data of table
Training sample data are normalized using principle two, and input trial and error procedure training program to [12,22]
Different hidden neuron nodal point numbers are trained in range, and network can export corresponding network mean square error and convergence step number, tool
Body value is listed in table 3.
The corresponding error of the different hidden layer nodes of table 3 and convergence step number
Pass through analysis " mean square error " and " convergence step number " two aspect factors, the BP neural network model of the pile foundation engineering
Optimal hidden layer node number be 19.
By the injection forming club-footed pile anti-pulling capacity nerve net of the sample data principle of substitution three Jing Guo normalized
Network Prediction program is trained and examines, and continues to optimize weights and threshold value by recycling learning training, establishes high-precision note
Slurry molding club-footed pile anti-pulling capacity neural network prediction program.
Step 5：The determination of injection forming club-footed pile anti-pulling capacity and stake diameter relation curve
It is reconnoitred by field geology conditions, determines soil layer liquidity index IL=0.97, Pile side soil effective angle of inner friction weights
Average valueThe deformation modulus E of expanding part soil layerS=68MPa；According to the achievement in research of a large amount of club-footed pile bearer properties and
《Code for design of building》(GB50007-2011) design code of pile foundation plinth, determine the long l=37m of rational stake,
The ratio between base expanding and base expanding height h=0.6D=1.2d, base expanding and base expanding part stake diameter and non-base expanding and base expanding part stake diameter λ '=2, enlarged footing upper diameter are gradually
Become extended corner β=90 ° of section.
According to《Code for design of building》(GB50007-2011) Preliminary design non-base expanding and base expanding part stake diameter d '=
600mm, and input and the neural network prediction model completed has been trained to show that anti-pulling capacity predicted value is 1900kN, according to resistance to plucking
It is -8% that design ultimate bearing capacity 2700kN, which calculates its error, with 30mm spacing in the non-base expanding and base expanding part stake right neighbours of diameter Preliminary design value d '
Domain carries out value (630,660,690,720,750,780,810,840,870,900), and inputs the nerve net trained and completed
Network prediction model carries out anti-pulling capacity prediction, the results are shown in Table 4.
4 anti-pulling capacity predicted value of table
According to the relationship of anti-pulling capacity and non-base expanding and base expanding part the stake diameter of neural network prediction model output, draw slip casting at
The non-club-footed pile diameter of type club-footed pile and anti-pulling capacity relation curve, are shown in Fig. 5.
Step 6：The determination of injection forming club-footed pile stake diameter
(1) it according to injection forming club-footed pile anti-pulling capacity design value determined by the curve graph of step 5 drafting, determines
Go out the section [690,720] where corresponding non-base expanding and base expanding part stake diameter d.
(2) corresponding according to curve in formula (15) determination section (690,720) (k=0,1,2,3,4,5,6,7,8,9)
Anti-pulling capacity linear equation：
Y=6.743x-2124.2
(3) by injection forming club-footed pile anti-pulling capacity design value TUKIt substitutes into above formula and calculates corresponding non-base expanding and base expanding part stake diameter
D=715mm, and non-base expanding and base expanding part stake diameter d=715mm input neural network prediction models are obtained into its anti-pulling capacity predicted value
For 2714.2kN, formula is utilizedThe relative error for calculating itself and design value is 0.53%, is less than 5%, the non-base expanding and base expanding portion
Stake diameter d=715mm is divided to meet design requirement.
(4) it according to injection forming club-footed pile non-base expanding and base expanding part stake diameter d and optimal expanding than λ ', can determine using formula (16)
Base expanding and base expanding part stake diameter D=λ ' d=1430mm.
Embodiment 2
Certain pile foundation is located at the areas certain B, and the design of resistance to plucking engineering pile uses injection forming club-footed pile.
Step 1：Determining influences the principal element of club-footed pile anti-pulling capacity
According to《Technical code for building pile foundation》(JGJ94-2008)、《Building pile foundation inspection specifications》(JGJ106-
2014) and injection forming club-footed pile practical engineering experience, and the anti-pulling capacity characteristic of research injection forming club-footed pile with
On the basis of load transfer mechanism, it is determined that the influence factor of injection forming club-footed pile anti-pulling capacity, i.e. soil layer liquidity index
IL, Pile side soil effective angle of inner friction weighted averageThe deformation modulus E of expanding part soil layerS, the long l of stake, non-base expanding and base expanding part stake
The ratio between diameter d, base expanding and base expanding height h, base expanding and base expanding part stake diameter and non-base expanding and base expanding part stake diameter D/d (being indicated with λ), enlarged footing upper diameter gradual change
Eight factors of extended corner β of section are to influence the principal element of injection forming club-footed pile anti-pulling capacity.
Step 2：The determination of the optimal expanding ratio of injection forming club-footed pile
According to《Architecture foundation pile inspection specifications》(JGJ106-2014) the unit for single-pile vertical anti-pulling static test regulation in,
The single pile Ultimate Up-lift Bearing Capacity for measuring same geological conditions area injection forming club-footed pile takes the long l=25m of stake, non-base expanding and base expanding portion
Dividing stake diameter d=300mm, base expanding and base expanding height h=360mm, expanding ratio is respectively 1,1.2,1.4,1.6,1.8,2.0,2.2,2.4,
2.6,2.8,3.0, the growth rate ρ of injection forming club-footed pile single pile Ultimate Up-lift Bearing Capacity is calculated according to formula (1)t, as a result see
Table 5.
The growth rate of 5 single pile Ultimate Up-lift Bearing Capacity of table
Many experimental results show that the growth rate of single pile Ultimate Up-lift Bearing Capacity is expanding than in first increases and then decreases with its
Trend (see Fig. 2), therefore corresponding expanding ratio is as optimal when single pile Ultimate Up-lift Bearing Capacity raising efficiency reaches maximum value
It is expanding than λ ', it is optimal expanding than λ '=2 as shown in Table 1.
Step 3：The determination of injection forming base expanding and base expanding pile bearing capacity BP neural network structural model
According to BP neural network prediction technique modeling principle, this patent determines that BP neural network model is made of 3 parts, packet
Input layer, hidden layer and output layer are included, it is final to determine that input layer is 8 nodes, by stake long vector group [l1,l2,Λ,ln], non-expansion
Stake radius vector group [d is divided in bottom1,d2,Λ,dn], base expanding and base expanding height vector group [h1,h2,Λ,hn], the stake of base expanding and base expanding part diameter and non-base expanding and base expanding
The ratio between part stake diameter Vector Groups [λ1,λ2,Λ,λn], soil layer liquidity index Vector Groups [IL1,IL2,Λ,ILn], Pile side soil effectively in
Angle of friction weighted average Vector GroupsDeformation modulus Vector Groups [the E of expanding part soil layers1,Es2,Λ,Esn]、
Extended corner Vector Groups [the β of enlarged footing upper diameter transition1,β2,Λ,βn] eight factors indicate；Hidden layer is single hidden layer；
Output layer unit number is 1, by injection forming club-footed pile anti-pulling capacity Vector GroupsIt indicates；Input
The weights and threshold value of layer to hidden layer are respectively wjkAnd θjk, the weights and threshold value of hidden layer to output layer are respectively vkjAnd γkt。
The BP neural network model established is shown in attached drawing 3.
Step 4：The determination of injection forming base expanding and base expanding pile bearing capacity BP neural network model parameter
It determines that sample size is 26 according to N >=2n, determines that training samples number is 15 and examines by formula (2), (3)
Sample size is 11.Sample data is shown in Table 6.
6 sample data of table
Training sample data are normalized using principle two, and input trial and error procedure training program to [12,22]
Different hidden neuron nodal point numbers are trained in range, and network can export corresponding network mean square error and convergence step number, tool
Body value is listed in table 7.
The corresponding error of the different hidden layer nodes of table 7 and convergence step number
Pass through analysis " mean square error " and " convergence step number " two aspect factors, the BP neural network model of the pile foundation engineering
Optimal hidden layer node number be 19.
By the injection forming club-footed pile anti-pulling capacity nerve net of the sample data principle of substitution three Jing Guo normalized
Network Prediction program is trained and examines, and continues to optimize weights and threshold value by recycling learning training, establishes high-precision note
Slurry molding club-footed pile anti-pulling capacity neural network prediction program.
Step 5：The determination of injection forming club-footed pile anti-pulling capacity and stake diameter relation curve
It is reconnoitred by field geology conditions, determines soil layer liquidity index IL=0.97, Pile side soil effective angle of inner friction weights
Average valueThe deformation modulus E of expanding part soil layerS=68MPa；According to the achievement in research of a large amount of club-footed pile bearer properties and
《Code for design of building》(GB50007-2011) design code of pile foundation plinth, determine the long l=20m of rational stake,
The ratio between base expanding and base expanding height h=0.6D=1.2d, base expanding and base expanding part stake diameter and non-base expanding and base expanding part stake diameter λ '=2, enlarged footing upper diameter are gradually
Become extended corner β=90 ° of section.
According to《Code for design of building》(GB50007-2011) Preliminary design non-base expanding and base expanding part stake diameter d '=
300mm, and input and the neural network prediction model completed has been trained to show that anti-pulling capacity predicted value is 2040kN, according to resistance to plucking
It is -7.3% that design ultimate bearing capacity 2200kN, which calculates its error, right in non-base expanding and base expanding part stake diameter Preliminary design value d ' with 30mm spacing
Neighborhood carries out value (330,360,390,420,450,480,510,540,570,600), and inputs the nerve trained and completed
Network Prediction Model carries out anti-pulling capacity prediction, the results are shown in Table 8.
8 anti-pulling capacity predicted value of table
According to the relationship of anti-pulling capacity and non-base expanding and base expanding part the stake diameter of neural network prediction model output, draw slip casting at
The non-club-footed pile diameter of type club-footed pile and anti-pulling capacity relation curve, are shown in Fig. 6.
Step 6：The determination of injection forming club-footed pile stake diameter
(1) it according to injection forming club-footed pile anti-pulling capacity design value determined by the curve graph of step 5 drafting, determines
Go out the section [390,420] where corresponding non-base expanding and base expanding part stake diameter d.
(2) corresponding according to curve in formula (15) determination section (390,420) (k=0,1,2,3,4,5,6,7,8,9)
Anti-pulling capacity linear equation：
Y=2.67x+1098.6
(4) by injection forming club-footed pile anti-pulling capacity design value TUKIt substitutes into above formula and calculates corresponding non-base expanding and base expanding part stake diameter
D=413mm, and non-base expanding and base expanding part stake diameter d=413mm input neural network prediction models are obtained into its anti-pulling capacity predicted value
For 2281kN, formula is utilizedThe relative error for calculating itself and design value is 3.68%, is less than 5%, the non-base expanding and base expanding part
Stake diameter d=413mm meets design requirement.
(5) it according to injection forming club-footed pile non-base expanding and base expanding part stake diameter d and optimal expanding than λ ', can determine using formula (16)
Base expanding and base expanding part stake diameter D=λ ' d=2 × 413=826mm.
Claims (9)
1. a kind of optimizing method for measuring to injection forming club-footed pile stake diameter, which is characterized in that specifically include following steps：
Step 1：Influence the determination of the principal element of club-footed pile anti-pulling capacity；
Step 2：The determination of the optimal expanding ratio of injection forming club-footed pile；
Step 3：The determination of injection forming base expanding and base expanding pile bearing capacity BP neural network model；
Step 4：The determination of injection forming base expanding and base expanding pile bearing capacity BP neural network model parameter；
Step 5：The determination of injection forming club-footed pile anti-pulling capacity and stake diameter relation curve；
Step 6：The determination of injection forming club-footed pile stake diameter.
2. according to claim 1 optimize method for measuring to injection forming club-footed pile stake diameter, which is characterized in that the step
Rapid one principal element for influencing club-footed pile anti-pulling capacity is respectively soil layer liquidity index IL, Pile side soil effective angle of inner friction weighting
Average valueThe deformation modulus E of expanding part soil layerS, the long l of stake, non-base expanding and base expanding part stake diameter d, base expanding and base expanding height h, base expanding and base expanding part stake diameter
With non-base expanding and base expanding part stake the ratio between the diameter D/d and extended corner β of enlarged footing upper diameter transition；D/d is indicated with λ.
3. according to claim 2 optimize method for measuring to injection forming club-footed pile stake diameter, which is characterized in that the step
Injection forming club-footed pile is optimal expanding than being determined by following formula in rapid two,
Wherein, ρtFor the expanding more expanding single pile limit more corresponding than λ=t-0.2 of single pile Ultimate Up-lift Bearing Capacity more corresponding than λ=t
The growth rate of anti-pulling capacity；TtFor expanding single pile Ultimate Up-lift Bearing Capacity more corresponding than λ=t；Tt-0.2To be expanding than λ=t-
0.2 corresponding single pile Ultimate Up-lift Bearing Capacity；When single pile Ultimate Up-lift Bearing Capacity improves corresponding expansion when efficiency reaches maximum value
Diameter is than being optimal expanding than λ '.
4. according to claim 3 optimize method for measuring to injection forming club-footed pile stake diameter, which is characterized in that the step
Rapid three BP neural networks structural model is made of input layer, hidden layer and output layer.
5. according to claim 4 optimize method for measuring to injection forming club-footed pile stake diameter, which is characterized in that described defeated
It is 8 nodes to enter layer, by stake long vector group [l1,l2,Λ,ln], non-base expanding and base expanding part stake radius vector group [d1,d2,Λ,dn], base expanding and base expanding
Height vector group [h1,h2,Λ,hn], the ratio between base expanding and base expanding part stake diameter and non-base expanding and base expanding part stake diameter Vector Groups [λ1,λ2,Λ,λn], soil
Layer liquidity index Vector Groups [IL1,IL2,Λ,ILn], Pile side soil effective angle of inner friction weighted average Vector Groups
Deformation modulus Vector Groups [the E of expanding part soil layers1,Es2,Λ,Esn], the extended corner Vector Groups of enlarged footing upper diameter transition
[β1,β2,Λ,βn] eight factors indicate；Hidden layer is single hidden layer；Output layer unit number is 1, by injection forming club-footed pile
Anti-pulling capacity Vector GroupsIt indicates；The weights and threshold value of input layer to hidden layer are respectively wjkWith
θjk, the weights and threshold value of hidden layer to output layer are respectively vkjAnd γkt。
6. according to claim 5 optimize method for measuring to injection forming club-footed pile stake diameter, which is characterized in that the step
The determination method of rapid four BP neural networks model parameter, specially：
(1) trial and error procedure determines hidden layer neuron nodal point number
Hidden layer neuron nodal point number range [12,22] is determined first；Sample data is inputted to the trial and error procedure worked out in Matlab
Training program carries out tentative calculation to hidden layer neuron nodal point number [12,22]；It is corresponding when selecting mean square error with convergence step number minimum
Hidden layer neuron nodal point number be final network hidden layer neuron nodal point number；
(2) determination of BP artificial neural networks weights and threshold value
1) sample size is determined according to N >=2n, and the quantity of training sample and test samples is determined by formula (2), (3)：
N-- is the total quantity of sample；T1-- the quantity of training sample；T2-- the quantity of the sample of inspection；[] indicates rounding operation；
2) random value initializes weights and threshold value in (0,1) section first, then returns to training sample data
One change is handled, and then passes to hidden layer neuron by formula (4), hidden layer neuron is exported according to formula (5)：
bk=f (S (k))=1/ (1+e-S(k)) (k=1,2, Λ, p) (5)
Wherein, PjFor input vector group, j is the number of element in input vector group, and k is the number of implicit layer unit；
3) hidden layer neuron output valve passes to output layer neuron by formula (6), and output layer neuron is according to formula (7)
It is exported：
4) network error e is calculated according to formula (8)：
Wherein, t is the number of element in output vector group；
As network global error e ＜ 0.0006, meet required precision, terminates network training；When network error e >=0.0006
When, be unsatisfactory for required precision, need according to the following steps 5), 6) weights of network and threshold value are modified；
5) according to desired output TUKtWith network reality outputThe correction error of output layer neuron is calculated using formula (9)
dt：
The correction error ek of hidden layer neuron is calculated according to formula (10)：
6) hidden layer is corrected to the connection weight v of output layer according to formula (11), (12)ktWith the threshold gamma of output layer neuronkt,
Wherein α is learning rate, 0 ＜ α ＜ 1
vkt(i+1)=vkt(i)+α·dt·bk(k=1,2, Λ, p；T=1,2, Λ, n) (11)
γkt(i+1)=γkt(i)+α·dt(t=1,2, Λ, n) (12)
Input layer is corrected to the connection weight W of hidden layer according to formula (13), (14)jkWith the threshold θ of hidden layer neuronjk,
Middle β is learning rate, 0 ＜ β ＜ 1
wjk(i+1)=wjk(i)+β·ek·Pj(j=1,2, Λ, n；K=1,2, Λ, p) (13)
θjk(i+1)=θjk(i)+β·ek(k=1,2, Λ, p) (14)
7) randomly select next mode of learning return the 3) step continue to train, until network error e meets required precision,
Network training is terminated, the weights and threshold value of the neural network are determined；
(3) establishment of neural network procedure
According to above-mentioned determining BP neural network model, and Matlab Neural Network Toolbox is used, it is anti-to injection forming club-footed pile
Bearing capacity neural network prediction program is pulled out to be worked out.
7. according to claim 6 optimize method for measuring to injection forming club-footed pile stake diameter, which is characterized in that the step
The determination method of rapid five injection formings club-footed pile anti-pulling capacity and stake diameter relation curve is specially：
It is reconnoitred by field geology conditions, determines soil layer property parameter；According to the achievement in research of club-footed pile bearer properties and《Building
Basic design specification of the foundation》(GB50007-2011) design code of pile foundation plinth determines the long l of stake, base expanding and base expanding height h=0.6D=
Extended corner β=90 ° of the ratio between 1.2d, base expanding and base expanding part stake diameter and non-base expanding and base expanding part stake diameter λ ', enlarged footing upper diameter transition；Root
According to《Code for design of building》(GB50007-2011) the non-base expanding and base expanding part stake diameter d ' of Preliminary design, and input and trained
At neural network prediction model carry out anti-pulling capacity prediction, according to the error of anti-pulling capacity predicted value and design value, choosing
Select rational spacingNon- base expanding and base expanding part stake diameter Preliminary design value d ' neighborhoods carry out value (d ' -10 μ, Λ,
+ 10 μ of d '-μ, d ', d '+μ, Λ, d '), and input and the neural network prediction model completed has been trained to carry out anti-pulling capacity prediction,
According to the relationship of the anti-pulling capacity of neural network prediction model output and non-base expanding and base expanding part stake diameter, injection forming club-footed pile is drawn
Non- club-footed pile diameter and anti-pulling capacity relation curve.
8. according to claim 7 optimize method for measuring to injection forming club-footed pile stake diameter, which is characterized in that utilize line
Property interpolation method determine the corresponding non-base expanding and base expanding part stake diameter d of injection forming club-footed pile anti-pulling capacity design value, according to injection forming
Club-footed pile non-base expanding and base expanding part stake diameter d and optimal expanding than λ ', determines base expanding and base expanding part stake diameter D：
D=λ ' d (16).
9. according to claim 8 optimize method for measuring to injection forming club-footed pile stake diameter, which is characterized in that the line
Property interpolation method the specific steps are：
(1) injection forming club-footed pile anti-pulling capacity design value determined by the curve graph drawn according to step 5, is determined pair
Section [d '+k μ, d '+(k+1) μ] (k=0,1,2,3,4,5,6,7,8,9) where the non-base expanding and base expanding part stake diameter d answered；
(2) according to curve pair in formula (15) determination section [d '+k μ, d '+(k+1) μ] (k=0,1,2,3,4,5,6,7,8,9)
The anti-pulling capacity linear equation answered,
The respectively corresponding anti-pulling capacities of non-base expanding and base expanding part stake diameter d '+k μ, the corresponding resistance to plucking carryings of d '+(k+1) μ
Power；
(3) by injection forming club-footed pile anti-pulling capacity design value TUKIt substitutes into formula (15) and calculates corresponding non-base expanding and base expanding part stake diameter
D, and non-base expanding and base expanding part stake diameter d input neural network prediction models are obtained into its anti-pulling capacity predicted value, utilize formulaIts relative error with design value is calculated, if error, within 5%, which, which meets design, wants
It asks；If error is less than 0 or more than 5%, which does not meet design requirement, appropriate to adjust stake diameter d, repeats
It inputs neural network prediction model and carries out error-tested, until error meets the requirements, determine the non-base expanding and base expanding of injection forming club-footed pile
Part stake diameter d.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711276442.9A CN108446413B (en) | 2017-12-06 | 2017-12-06 | Method for optimally measuring pile diameter of grouting-formed club-footed pile |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711276442.9A CN108446413B (en) | 2017-12-06 | 2017-12-06 | Method for optimally measuring pile diameter of grouting-formed club-footed pile |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108446413A true CN108446413A (en) | 2018-08-24 |
| CN108446413B CN108446413B (en) | 2021-07-13 |
Family
ID=63190636
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711276442.9A Active CN108446413B (en) | 2017-12-06 | 2017-12-06 | Method for optimally measuring pile diameter of grouting-formed club-footed pile |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108446413B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111428297A (en) * | 2020-03-23 | 2020-07-17 | 交通运输部公路科学研究所 | BP neural network-based pile foundation P-S curve determination method |
| CN113123330A (en) * | 2021-03-02 | 2021-07-16 | 汕头大学 | Method for determining diameter of jet grouting pile based on long and short memory neural network |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106988772A (en) * | 2017-02-15 | 2017-07-28 | 河北钢铁集团矿业有限公司 | Underground mine arch suspension bridge support bearing method |
| CN107330232A (en) * | 2017-08-11 | 2017-11-07 | 上海岩土工程勘察设计研究院有限公司 | Ultimate bearing capacity evaluation method for pile-end post-grouting ultra-long pile |
-
2017
- 2017-12-06 CN CN201711276442.9A patent/CN108446413B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106988772A (en) * | 2017-02-15 | 2017-07-28 | 河北钢铁集团矿业有限公司 | Underground mine arch suspension bridge support bearing method |
| CN107330232A (en) * | 2017-08-11 | 2017-11-07 | 上海岩土工程勘察设计研究院有限公司 | Ultimate bearing capacity evaluation method for pile-end post-grouting ultra-long pile |
Non-Patent Citations (3)
| Title |
|---|
| BASHAR TARAWNEH、RANA IMAM: "Regression versus Artifi from Empirical Datacial Neural Networks: Predicting Pile Setup", 《KSCE JOURNAL OF CIVIL ENGINEERING》 * |
| 王志辉等: "桩端压力注浆桩承载力径向基遗传神经网络研究", 《山东交通学院学报》 * |
| 蒋建平: "基于BP神经网络的横向受荷桩承载力预测 ", 《水运工程》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111428297A (en) * | 2020-03-23 | 2020-07-17 | 交通运输部公路科学研究所 | BP neural network-based pile foundation P-S curve determination method |
| CN113123330A (en) * | 2021-03-02 | 2021-07-16 | 汕头大学 | Method for determining diameter of jet grouting pile based on long and short memory neural network |
| CN113123330B (en) * | 2021-03-02 | 2022-07-22 | 汕头大学 | Method for determining diameter of jet grouting pile based on long and short memory neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108446413B (en) | 2021-07-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107045568B (en) | River course roughness inversion method based on dynamic programming successive approximation method | |
| CN103093114A (en) | Distributed-type river basin water deficit calculating method based on terrain and soil characteristics | |
| CN112100927A (en) | Prediction method for slope deformation and soft soil foundation settlement based on GA-BP neural network | |
| CN105279361B (en) | Gao Tushi not overflow cofferdam slope instability relative risk detection methods | |
| Qiu et al. | Seepage monitoring models study of earth-rock dams influenced by rainstorms | |
| CN103530700B (en) | Urban distribution network saturation loading Comprehensive Prediction Method | |
| CN108446413A (en) | A kind of optimization assay method of injection forming club-footed pile stake diameter | |
| CN108763797B (en) | Lake and underground water stable flow effect simulation method based on underground water model | |
| CN115130772A (en) | Ground surface settlement prediction method based on PSO (particle swarm optimization) optimization BP (back propagation) neural network | |
| CN112884198B (en) | Method for predicting dam crest settlement of panel dam by combining threshold regression and improved support vector machine | |
| Xu et al. | Inversion analysis of seepage parameters based on improved shuffled frog leaping algorithm | |
| Legrand et al. | Unstructured, anisotropic mesh generation for the Northwestern European continental shelf, the continental slope and the neighbouring ocean | |
| CN106651054A (en) | Method for identifying roughness of water conveyance channel of long-distance water transfer project based on genetic algorithm | |
| CN108763798B (en) | Method for simulating unsteady flow effect of lake and underground water | |
| Qi et al. | Application of a groundwater modelling system in groundwater environmental impact assessment of river and lake connection in western Jilin region. | |
| CN109241579A (en) | A kind of fast modeling method considering attached fluid mass hyperspace transient characteristic | |
| CN115759203A (en) | Offshore pile foundation erosion prediction method under multi-factor coupling condition | |
| CN109117517A (en) | A kind of base pit dewatering optimum design method based on genetic algorithm and groundwater model | |
| Lian et al. | Evaluation and applicability study on prediction methods of water inflow in mines | |
| CN106650293A (en) | AM nest sampling algorithm based groundwater model evaluation method | |
| CN114595642A (en) | Simplified analysis method for soil slope stability considering rainfall infiltration influence | |
| CN109063306B (en) | Soil infiltration capacity space dispersion method of gridding Hebei model | |
| Jianbin et al. | An ANN model for predicting level ultimate bearing capacity of PHC pipe pile | |
| CN106595798B (en) | A method of aquifer parameter is sought automatically using high frequency level of ground water data | |
| Wu | Protection of water resources in Dongsha scenic area of south sea |
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 |