Deep cement mixing pile construction method
By determining the distribution ratio and stirring resistance benchmark of sprinkler irrigation cement slurry group in the construction of deep cement mixing piles, collecting and analyzing pile deep section parameters in real time, establishing a prediction training model to adjust the stirring parameters, solving the problem of pile formation stability under different geological and stratigraphic conditions, and achieving high-quality and stable deep cement mixing pile construction.
Patent Information
- Application Number
- CN202510002061.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-02
- Publication Date
- 2025-04-22
AI Technical Summary
The existing deep cement mixing pile construction methods cannot guarantee the stability of pile formation under different geological and stratigraphic conditions, especially the soil layer samples in different depths of the same cement mixing pile are different.
By determining the mixing ratio of the sprinkler cement slurry components, and measuring the stirring resistance when the soil layer sample and the cement slurry are mixed uniformly as the reference resistance in a stirring state. The mixing speed, spraying amount, drill bit movement speed and mixing resistance of different pile depth sections are collected in real time, and the corresponding relationship is constructed, and a prediction training model is established to adjust the mixing parameters to ensure uniform mixing of the soil layer and cement slurry.
The pile-forming quality and stability of deep cement mixing piles are improved, suitable for different geological and stratigraphic conditions, ensuring full mixing of sprinkler cement slurry and soil layers, and improving the reliability and efficiency of construction.
Smart Images
Figure CN119859994A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of civil construction, and in particular to a method for constructing deep cement mixing piles. Background Art
[0002] Deep cement mixing piles are a method of using cement as a curing agent to forcefully mix the soft soil, silt, etc. in the stratum with concrete spray cement slurry in the foundation through deep mixing machinery, so as to harden the soft foundation and improve the foundation strength.
[0003] In the existing construction process of deep cement mixing piles, due to the constraints of different construction geology, the soil layer samples of the strata where the mixing piles are located cannot be uniform. Therefore, the stability of the cement mixing piles in the strata where they are located cannot be guaranteed, especially for the strata at different depths of the same cement mixing pile, the soil layer samples may also be different. Therefore, it is necessary to provide a deep cement mixing pile construction method to ensure the stability of the deep cement mixing piles. Summary of the invention
[0004] The purpose of this application is to provide a deep cement mixing pile construction method, which can improve the quality of pile formation, is suitable for construction in different geological and strata, and ensures the stability of pile formation.
[0005] In order to achieve the above object, the present invention provides a deep cement mixing pile construction method, comprising the following steps:
[0006] Determine the mix ratio of the sprinkler cement slurry components, mix the target soil layer sample with the determined sprinkler cement slurry sample, and at the same time, measure the stirring resistance that the stirring paddle can receive when the soil layer sample and the sprinkler cement slurry are evenly mixed under the stirring state, and use the detected stirring resistance as the reference resistance under the evenly mixed state;
[0007] When the slurry spraying pile machine and the mortar pump are turned on at the same time, the stirring speed, spraying volume, movement speed of the spraying drill bit and the stirring resistance that the drill pipe stirring blade can withstand in different pile depth sections are collected in real time, and the corresponding relationship between the collected stirring resistance of the drill pipe stirring blade and the reference resistance is established;
[0008] The grouting pile machine is lifted at the same speed as the drilling process to collect the mixing speed, grouting volume, grouting drill bit movement speed and the mixing resistance that the drilling rod mixing blade can withstand in real time during different pile depths. After the drilling is lifted, the data is reviewed.
[0009] A prediction training model is established, and data with large differences before and after the review process are eliminated. According to the corresponding relationship between the stirring resistance and the reference resistance of the mixing blade of the drill rod in the pile depth section with the same soil layer composition as the largest proportion, the model input data of each variable parameter is restored, and the model input data is input into the prediction training model to obtain the output results. Based on the output results, the correlation between different soil layer compositions in different pile depth sections and the stirring speed, shotcrete volume, and the moving speed of the shotcrete drill bit is determined;
[0010] Repeat the sinking twice, and adjust the operating parameters of the mixing speed, shotcrete volume and shotcrete drill bit moving speed in real time according to the correlation results between the different soil layer compositions in different pile depth sections and the mixing speed, shotcrete volume and shotcrete drill bit moving speed;
[0011] The slurry spraying pile machine is lifted at the same speed as the secondary repeated sinking to carry out the drilling and complete the construction of the mixing pile.
[0012] In an optional embodiment, the mix ratio of the sprinkler cement slurry components is determined during the pile construction survey and test stage. During the pile construction survey and test stage, soil layer samples at different pile depths are collected and corresponding soil layer data are obtained.
[0013] The soil layer data obtained in the pile investigation and testing stage are used to determine the soil layer composition with the largest proportion at different pile depths, and the soil layer composition with the largest proportion is calibrated as the target soil layer sample. The mix ratio of the sprinkler cement slurry components is determined based on the composition of the target soil layer sample.
[0014] In an optional embodiment, the corresponding relationship between the collected stirring resistance of the drill pipe stirring blade and the reference resistance is constructed during the pile forming test stage. During the pile forming test stage, the slurry spraying pile machine is used to spray the soil from top to bottom, and the mortar pump is turned on to spray the mortar with the determined proportion of the cement slurry components into the formation until the bottom of the pile while rotating along with the operation of the drill bit. At the same time, during the drilling process, the stirring speed, spraying amount, movement speed of the spraying drill bit and the stirring resistance that the drill pipe stirring blade can withstand during different pile depth sections are collected in real time. When the drill bit reaches the pile depth section with the same soil layer composition as the largest proportion, the stirring resistance that the drill pipe stirring blade can withstand is obtained, and the corresponding relationship between the collected stirring resistance of the drill pipe stirring blade and the reference resistance is constructed.
[0015] In an optional embodiment, the model input data for restoring each variable parameter is based on the correlation analysis coefficient between the stirring resistance of the drill rod stirring blade and the benchmark resistance in a uniformly mixed state when the shotcrete drill bit reaches the pile depth section with the same composition as the soil layer with the largest proportion, and the input data of the prediction training model for each variable parameter is restored according to the correlation analysis coefficient.
[0016] In an optional embodiment, during the training process, the reviewed data before and after are cleaned, and the data is standardized or normalized before being input into the prediction training model. The stirring resistance that the stirring paddle part can receive when the soil layer sample and the sprinkler cement slurry are evenly mixed under the soil layer composition state with the largest proportion is used as the target variable parameter, and model training with different variable parameters is performed.
[0017] In an optional embodiment, the output results of the prediction training model include the mixing speed, shotcrete volume and shotcrete drill bit movement speed corresponding to different soil layer compositions in different pile depth sections, and based on this, the operating parameters of the mixing speed, shotcrete volume and shotcrete drill bit movement speed are adjusted during the secondary repeated sinking process.
[0018] In an optional implementation, the variable parameter prediction training is performed through a multi-layer perceptron training model, which mainly includes forward propagation, loss calculation, back propagation, weight update and iterative training.
[0019] In an optional embodiment, during the forward propagation process, the input signal generates an output through the hidden nodes of each network layer, and the forward propagation formula of the multilayer perceptron includes:
[0020] a (l+1) =g l (W l a l +b l )
[0021] Among them, a l represents the activation value or output value of the lth layer; W l represents the weight matrix of the lth layer; b l represents the bias vector of the lth layer; g l represents the activation function of the lth layer; a (l+1) It represents the input of the l+1th layer, or the output of the lth layer after being processed by the activation function.
[0022] In an optional embodiment, back propagation is used to calculate the gradient of the loss function with respect to the hidden layer and update the weights by gradient descent. The back propagation formula includes:
[0023]
[0024] Among them, z 2 is the weighted input of the hidden layer, that is, W 2 a 1 +b 2 , is the derivative of the activation function of the hidden layer, is the weight matrix W 2 The transpose of
[0025] Weight updates include:
[0026]
[0027] Among them, θ represents the model parameters, including weights and biases; η represents the learning rate, Represents the gradient of the loss function with respect to the model parameters.
[0028] Through the deep cement mixing pile construction method in the present application, the soil layer with the largest proportion in different pile depth sections can be taken as the target soil layer, and the stirring resistance that the stirring paddle part can receive when the soil layer sample in the target soil layer is evenly mixed with the spraying cement slurry can be used as the benchmark reference, and training is performed to predict the stirring speed, spraying volume and movement speed of the spraying drill bit in different pile depth sections, thereby providing a basis for guidance for the subsequent secondary pile driving construction process, and finally achieving uniform mixing of the spraying cement slurry and the soil layer samples in different pile depth sections to ensure the quality of the pile.
[0029] The deep cement mixing pile construction method of the present invention can be applied to different geological conditions and has broad universal guidance. During the initial construction of the drill rod, data can be collected and prediction training can be performed on the basis of clear target parameters, so that the prediction training model can output data on control parameters that are relatively instructive, providing practical protection for secondary sinking drilling construction.
[0030] Other features and advantages of the present application will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.
[0032] Figure 1 It is a schematic diagram of the process of deep cement mixing pile construction method in this application. DETAILED DESCRIPTION
[0033] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.
[0034] In the description of this application, it should be noted that the terms "inside", "outside", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the product of the application is usually placed when in use, which is only for the convenience of describing this application and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on this application. In addition, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.
[0035] In the description of this application, it should also be noted that, unless otherwise clearly specified and limited, the terms "disposed" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.
[0036] The deep cement mixing pile construction method of the present invention mainly predicts three different variable parameters of mixing speed, shotcrete volume and shotcrete drill bit movement speed by determining the form of target parameters and inputting the data collected during the pile test into the prediction training model, and outputs the prediction results, which are then used as guiding data for the operating parameters in the secondary drilling process, and can ensure the mixing effect of the underground soil layer and the spraying cement slurry to the greatest extent.
[0037] By comparing the corresponding relationship between the stirring resistance of the drill pipe stirring blade in the target stratum and the reference resistance, the target parameters in the actual pile formation process when the soil layer sample and the spraying cement slurry are evenly mixed can be truly restored. This can then be used as a reference to accurately predict the relationship between the composition of different soil layers in different pile depths and the stirring speed, spraying volume and the moving speed of the spraying drill bit. In the process of secondary drilling, the operating parameters of the stirring speed, spraying volume and the moving speed of the spraying drill bit can be controlled and adjusted in real time according to the correlation results, so that the spraying cement slurry and the different soil layer compositions in different pile depths can be fully and effectively mixed, thereby improving the quality of pile formation.
[0038] See also Figure 1 The deep cement mixing pile construction method provided by the present invention comprises the following steps:
[0039] Determine the mix ratio of the sprinkler cement slurry components, mix the target soil layer sample with the determined sprinkler cement slurry sample, and at the same time, measure the stirring resistance that the stirring paddle can receive when the soil layer sample and the sprinkler cement slurry are evenly mixed. The detected stirring resistance is used as the benchmark resistance under the uniformly mixed state.
[0040] Specifically, the mix ratio of the sprinkler cement slurry components is determined during the pile investigation and testing stage. During the pile investigation and testing stage, soil layer samples at different pile depths are collected and corresponding soil layer data are obtained, and then the target soil layer is determined.
[0041] The soil layer data obtained in the pile investigation and testing stage is used to determine the soil layer composition with the largest proportion at different pile depths, and the soil layer composition with the largest proportion is calibrated as the target soil layer sample. Due to the construction of deep cement mixing piles, the soil layer composition at different pile depths cannot be determined, but the soil layer with the largest proportion at different pile depths can be determined. By determining the soil layer with the largest proportion as the target soil layer, the main soil layer composition can be used as a reference during the deep cement mixing pile construction process to better ensure the stability of the entire pile. At the same time, it can also facilitate the determination of the mix ratio of the sprinkler cement slurry components based on the composition of the target soil layer sample.
[0042] After determining the mix ratio of the sprinkler cement slurry components, a stirring test is carried out in the preliminary experimental process. Under experimental conditions, the stirring resistance that the stirring paddle can receive when the soil layer sample and the sprinkler cement slurry are evenly mixed is measured, so that the fluid state in the evenly mixed state can be used as a basis for subsequent testing.
[0043] When the slurry spraying pile machine and the mortar pump are turned on at the same time, the stirring speed, spraying volume, spraying drill bit movement speed and the stirring resistance that the drill rod stirring blade can withstand during different pile depths are collected in real time, and the corresponding relationship between the collected stirring resistance of the drill rod stirring blade and the benchmark resistance is established.
[0044] The corresponding relationship between the collected stirring resistance of the drill pipe stirring blade and the reference resistance is constructed during the pile test stage. Specifically, the deep cement mixing pile construction in the present invention mainly includes pile survey, pile test and secondary drilling, and finally the whole pile construction is completed after the drill is lifted.
[0045] During the pile survey phase, the target soil layer samples and the benchmark resistance are determined.
[0046] During the pile test phase, the slurry spraying pile machine is used to spray the soil from top to bottom, and the mortar pump is turned on to spray the mortar with the determined mix ratio of the cement slurry components into the stratum while the drill bit is running, rotating and spraying it until it reaches the bottom of the pile. At the same time, during the drilling process, the mixing speed, spraying volume, movement speed of the spraying drill bit and the stirring resistance that the drill pipe stirring blade can withstand during different pile depths are collected in real time. When the drill bit reaches the pile depth section with the same composition as the soil layer with the largest proportion, the stirring resistance that the drill pipe stirring blade can withstand is obtained, and the corresponding relationship between the collected stirring resistance of the drill pipe stirring blade and the benchmark resistance is constructed.
[0047] During the collection process, when the drill bit reaches different pile depth sections with the same soil layer composition as the largest proportion, the stirring resistance that can be received by multiple groups of identical or similar drill pipe stirring blades can be collected, and based on the stirring resistance, a corresponding relationship with the benchmark resistance is established.
[0048] It should be pointed out that the model input data for restoring each variable parameter is based on the correlation between the experimental stage and the actual operation. Specifically, the stirring resistance that the drill pipe stirring blade can receive is the data collected under the actual operation state, and the corresponding relationship between the actual operation state data and the reference resistance is constructed in order to effectively and accurately restore and amplify. That is to say, by performing a correlation analysis between the collected stirring resistance of the drill pipe stirring blade and the reference resistance under the mixed uniform state, the correlation analysis coefficient is obtained, and the input data of each variable parameter to the prediction training model is restored according to the correlation analysis coefficient, which can fully consider various factors in the actual operation and ensure the accuracy of the model input data. The correlation factors include the amplification ratio between the amount of cement slurry sprayed in the actual operation and the amount of experimental materials, the difference between the agitator and the drill pipe stirring blade, and the formation pressure.
[0049] The correlation analysis of the variable factors of stirring speed, shotcrete volume and shotcrete drill head moving speed can restore the model input data through the determined independent correlation analysis coefficients due to their different characteristics.
[0050] After the drilling process in the pile test phase, the slurry spraying pile machine is lifted at the same speed as the drilling process to collect real-time data on the mixing speed, spraying volume, movement speed of the spraying drill bit, and mixing resistance that the drilling rod mixing blade can withstand during different pile depths. The data is reviewed after the drilling is completed. The drilling process can increase the amount of data collected, ensure the adequacy of the data, and ensure the stability and reliability of the output results of the prediction training model.
[0051] A prediction training model is established, and the data before and after the review are cleaned. The data are standardized or normalized before being input into the prediction training model. The cleaning process includes eliminating the data with large differences before and after the review process, and restoring the model input data of each variable parameter according to the corresponding relationship between the stirring resistance and the reference resistance of the drilling rod stirring blade in the pile depth section with the same soil layer composition as the largest proportion. The model input data is input into the prediction training model to obtain the output result, and the correlation between the different soil layer compositions in different pile depth sections and the stirring speed, shotcrete volume and the moving speed of the shotcrete drill bit is determined according to the output result.
[0052] During the prediction training process, the stirring resistance that the stirring paddle can receive when the soil layer sample and the sprinkler cement slurry are evenly mixed under the soil layer composition with the largest proportion is taken as the target variable parameter, and model training with different variable parameters is carried out.
[0053] The output results of the prediction training model include the mixing speed, shotcrete volume and shotcrete drill bit movement speed corresponding to different soil layer compositions in different pile depth sections. Based on this, the operating parameters of the mixing speed, shotcrete volume and shotcrete drill bit movement speed are adjusted during the secondary repeated sinking process.
[0054] It should be noted that the mechanism of correlation analysis is to truly restore the accuracy of the data and ensure the accuracy of the data in the input model. In the process of outputting data, since the training model gives an angle that is biased towards actual operation, the model output itself actually represents information such as the actual parameters of the variables. Therefore, there is no need for reverse restoration. The model output can provide actual engineering guidance.
[0055] During the secondary repeated sinking process in the secondary drilling stage, the operating parameters of the mixing speed, shotcrete volume and shotcrete drill bit moving speed are controlled and adjusted in real time according to the correlation results between the different soil layer compositions in different pile depth sections and the mixing speed, shotcrete volume and shotcrete drill bit moving speed.
[0056] The slurry spraying pile machine is lifted at the same speed as the secondary repeated sinking to carry out the drilling and complete the construction of the mixing pile.
[0057] In the present invention, by taking the soil layer composition with the largest proportion as the target soil layer, the main soil layer composition of the stratum where the pile is located can be restored, and by determining the target variable parameters based on the components of the target soil layer, the correlation between the stirring speed, shotcrete volume and shotcrete drill bit movement speed corresponding to different soil layer compositions in different pile depths can be determined, thereby improving the authenticity and reliability of the prediction training model application, and being suitable for the construction needs of deep cement mixing piles in different geological and strata.
[0058] In the present invention, the prediction training of variable parameters is performed through a multi-layer perceptron training model, which mainly includes forward propagation, loss calculation, back propagation, weight update and iterative training.
[0059] The multilayer perceptron can map input data to a high-dimensional space through multi-layer nonlinear transformation, thereby better representing the internal structure of the data. During the training process, the model has better generalization ability by learning the common features of the data. Therefore, the multilayer perceptron in the present invention can be applied to the training prediction between the stirring speed, the shotcrete amount and the moving speed of the shotcrete drill bit, and give a more accurate prediction result.
[0060] The training of multilayer perceptrons is usually done by backpropagation, which is based on a gradient descent optimization strategy to adjust the weights and biases of the network. By choosing the appropriate loss function and optimization algorithm, the multilayer perceptron can gradually optimize the model during the training process to minimize the error between the predicted output and the true label.
[0061] Specifically, in the multilayer perceptron training model of the present application, during the forward propagation process, the input signal generates output through the hidden nodes of each network layer, and the forward propagation formula of the multilayer perceptron includes:
[0062] a (l+1) =g l (W l a l +b l )
[0063] Among them, a l represents the activation value or output value of the lth layer; W l represents the weight matrix of the lth layer; b l represents the bias vector of the lth layer; g l represents the activation function of the lth layer; a (l+1) It represents the input of the l+1th layer, or the output of the lth layer after being processed by the activation function.
[0064] Back propagation is used to calculate the gradient of the loss function with respect to the hidden layer and update the weights by gradient descent. The back propagation formula includes:
[0065]
[0066] Among them, z 2 is the weighted input of the hidden layer, that is, W 2 a 1 +b 2 , is the derivative of the activation function of the hidden layer, is the weight matrix W 2 The bias of as well as Respectively represent the gradient of the loss function L between adjacent layers;
[0067] Weight updates include:
[0068]
[0069] Among them, θ represents the model parameters, including weights and biases; η represents the learning rate, Represents the gradient of the loss function with respect to the model parameters.
[0070] The multi-layer perceptron prediction training model in the present invention can provide a relatively true and reliable operating basis for the secondary drilling process, thereby ensuring the mixing effect of the underground soil layer and the sprinkler cement slurry to the greatest extent and ensuring the pile quality of the entire pile.
[0071] It should be noted that, in the absence of conflict, the features in the embodiments of this application may be combined with each other.
[0072] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims
1. A deep cement mixing pile construction method, characterized in that: The following steps are involved: Determine the mix ratio of the sprinkler cement slurry components, mix the target soil layer sample with the determined sprinkler cement slurry sample, and at the same time, measure the stirring resistance that the stirring paddle can receive when the soil layer sample and the sprinkler cement slurry are evenly mixed under the stirring state, and use the detected stirring resistance as the reference resistance under the evenly mixed state; When the slurry spraying pile machine and the mortar pump are turned on at the same time, the stirring speed, spraying volume, movement speed of the spraying drill bit and the stirring resistance that the drill pipe stirring blade can withstand in different pile depth sections are collected in real time, and the corresponding relationship between the collected stirring resistance of the drill pipe stirring blade and the reference resistance is established; The grouting pile machine is lifted at the same speed as the drilling process to collect the mixing speed, grouting volume, grouting drill bit movement speed and the mixing resistance that the drilling rod mixing blade can withstand in real time during different pile depths. After the drilling is lifted, the data is reviewed. A prediction training model is established, and data with large differences before and after the review process are eliminated. According to the corresponding relationship between the stirring resistance and the reference resistance of the mixing blade of the drill rod in the pile depth section with the same soil layer composition as the largest proportion, the model input data of each variable parameter is restored, and the model input data is input into the prediction training model to obtain the output results. Based on the output results, the correlation between different soil layer compositions in different pile depth sections and the stirring speed, shotcrete volume, and the moving speed of the shotcrete drill bit is determined; Repeat the sinking twice, and adjust the operating parameters of the mixing speed, shotcrete volume and shotcrete drill bit moving speed in real time according to the correlation results between the different soil layer compositions in different pile depth sections and the mixing speed, shotcrete volume and shotcrete drill bit moving speed; The slurry spraying pile machine is lifted at the same speed as the secondary repeated sinking to carry out the drilling and complete the construction of the mixing pile.
2. The deep cement mixing pile construction method according to claim 1 is characterized in that: The mix ratio of the sprinkler cement slurry components is determined during the pile investigation and testing stage. During the pile investigation and testing stage, soil samples at different pile depths are collected and corresponding soil layer data are obtained. The soil layer data obtained in the pile investigation and testing stage are used to determine the soil layer composition with the largest proportion at different pile depths, and the soil layer composition with the largest proportion is calibrated as the target soil layer sample. The mix ratio of the sprinkler cement slurry components is determined based on the composition of the target soil layer sample.
3. The deep cement mixing pile construction method according to claim 2 is characterized in that: The corresponding relationship between the collected stirring resistance of the drill pipe stirring blade and the benchmark resistance is established during the pile test stage. During the pile test stage, the slurry spraying pile machine is used to spray the soil from top to bottom, and the mortar pump is turned on to spray the mortar with the determined mix ratio of the spraying cement slurry components into the formation until the bottom of the pile while rotating along with the operation of the drill bit. At the same time, during the drilling process, the stirring speed, spraying volume, spraying drill bit movement speed and stirring resistance that the drill pipe stirring blade can withstand in different pile depth sections are collected in real time. When the drill bit reaches the pile depth section with the same soil layer composition as the largest proportion, the stirring resistance that the drill pipe stirring blade can withstand is obtained, and the corresponding relationship between the collected stirring resistance of the drill pipe stirring blade and the benchmark resistance is established.
4. The deep cement mixing pile construction method according to claim 2 is characterized in that: The model input data for restoring each variable parameter is based on the correlation analysis coefficient between the stirring resistance of the drill pipe stirring blade and the benchmark resistance under the mixed uniform state when the shotcrete drill bit reaches the pile depth section with the same composition as the soil layer with the largest proportion. The input data of the prediction training model for each variable parameter is restored based on the correlation analysis coefficient.
5. The deep cement mixing pile construction method according to claim 2 is characterized in that: During the training process, the reviewed data are cleaned before and after, and the data are standardized or normalized before being input into the prediction training model. The stirring resistance that the stirring paddle can receive when the soil layer sample and the sprinkler cement slurry are evenly mixed under the soil layer composition with the largest proportion is used as the target variable parameter, and model training with different variable parameters is carried out.
6. The deep cement mixing pile construction method according to claim 2, characterized in that: The output results of the prediction training model include the mixing speed, shotcrete volume and shotcrete drill bit movement speed corresponding to different soil layer compositions in different pile depth sections. Based on this, the operating parameters of the mixing speed, shotcrete volume and shotcrete drill bit movement speed are adjusted during the secondary repeated sinking process.
7. The deep cement mixing pile construction method according to claim 2, characterized in that: The multi-layer perceptron training model is used to perform predictive training on variable parameters, which mainly includes forward propagation, loss calculation, back propagation, weight update and iterative training.
8. The deep cement mixing pile construction method according to claim 7, characterized in that: In the forward propagation process, the input signal generates output through the hidden nodes of each network layer. The forward propagation formula of the multilayer perceptron includes: a (l+1) =g l (W l a l +b l ) Among them, a l represents the activation value or output value of the lth layer; W l represents the weight matrix of the lth layer; b l represents the bias vector of the lth layer; g l represents the activation function of the lth layer; a (l+1) It represents the input of the l+1th layer, or the output of the lth layer after being processed by the activation function.
9. The deep cement mixing pile construction method according to claim 8, characterized in that: Back propagation is used to calculate the gradient of the loss function with respect to the hidden layer and update the weights by gradient descent. The back propagation formula includes: Among them, z2 is the weighted input of the hidden layer, that is, W2a1+b2, is the derivative of the activation function of the hidden layer, is the bias of the weight matrix W2; Weight updates include: Among them, θ represents the model parameters, including weights and biases; η represents the learning rate, Represents the gradient of the loss function with respect to the model parameters.