Prediction system for static pressure method prestressed concrete pile bearing capacity and implementation method thereof

Abstract

The invention discloses a system for predicting the bearing capacity of a prestressed concrete pile through a static pressure method. The system comprises a static pressure pile machine construction monitoring module, a digital static sounding module, a data receiving module, an interactive touch screen module, a bearing capacity calculation module and a cloud data center. The invention further discloses an implementation method of the static pressure method prestressed concrete pile bearing capacity prediction system, the feed-forward neural network model is used for correlation analysis of bearing capacity prediction data, input original data can be efficiently analyzed, after each time of model iteration, the updated model is transmitted to the on-site algorithm analysis unit through the 5G module, and the prediction result of the bearing capacity of the static pressure method prestressed concrete pile is obtained. Dynamic updating and optimization of the field model are achieved, the real-time performance and accuracy of prediction of the ultimate bearing capacity of the prestressed concrete piles are effectively improved through the mechanism, the limitation that a traditional acceptance method only depends on static load tests of a few piles is overcome, the bearing capacity of all the piles can be comprehensively evaluated, and potential safety hazards of construction are reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of monitoring and control of pile foundation machinery and equipment, and particularly relates to a prediction system for the bearing capacity of prestressed concrete piles by the static pressure method and an implementation method thereof. Background Art

[0002] Prestressed concrete pipe piles (PHC piles) are made of high-strength concrete and prestressed steel bars through a centrifugal forming process, and have excellent material properties and construction convenience. They are widely used in infrastructure construction, especially in projects such as urban infrastructure, bridges, high-rise buildings, and transportation hubs. Their concrete strength can reach above 80 MPa, and their bearing capacity far exceeds that of traditional bored cast-in-place piles, making them the first choice for high-bearing-capacity foundation projects. PHC piles constructed by the static pressure method have the advantages of low noise, no vibration, no oil fume, and no mud, and are suitable for construction in noise-sensitive areas and under complex geological conditions. With the growth of infrastructure demand in China, PHC piles have been more and more widely used due to their high efficiency, environmental protection, and economy, and China has become the country with the largest number of static pressure piles used globally.

[0003] With the expansion of the application range of PHC pipes by the static pressure method, the deficiencies in design, construction, and related standards and specifications have gradually emerged. Currently, there is no unified standard for the design and construction of PHC piles, resulting in the failure to fully exert the bearing capacity and affecting the safety and stability of the pile foundation.

[0004] Traditional design methods usually use the maximum pile pressing force of the static pile press as the basis for bearing capacity, but this has the risk of misjudgment; when predicting the bearing capacity using the data of the static cone penetration test (CPT), variables during the construction process may also affect the pile forming quality. In addition, although the static load test can directly verify the bearing capacity of the pile foundation, due to its high cost and long cycle, and the inability to comprehensively evaluate the bearing capacity of all piles, its wide application is limited. The results of the static pressure test usually show that the actual bearing capacity is higher than the design value, resulting in the failure to fully exert the bearing potential of PHC pipes, increasing the project cost and wasting resources. At the same time, when accepting the static pressure pile foundation, only the results of the static load tests of a few piles are relied on to evaluate the bearing capacity of the pile foundation, and the ultimate bearing capacity of all piles cannot be comprehensively evaluated, which may pose potential safety hazards.

[0005] Therefore, how to scientifically and accurately predict the ultimate bearing capacity of PHC pipes is a key problem that needs to be solved urgently at present. Through an effective bearing capacity prediction method, not only can the project quality and safety be improved, but also the design scheme can be adjusted in time during the construction process, the design accuracy can be improved, the resource allocation can be optimized, the test quantity and cost can be reduced, and the construction period can be shortened. Bearing capacity prediction helps to cope with complex geological and construction conditions, promotes the application of PHC pipes in a wider range of fields, further improves the construction efficiency, and realizes an economic and efficient construction plan.

[0006] However, existing research is still insufficient in improving the prediction accuracy of the vertical ultimate bearing capacity of single static-pressure PHC piles, and the relationships among the physical parameters of the pile body, the key parameters during the construction process, and the soil layer geological information are not fully considered.

[0007] Therefore, exploring an implementation method for a simple and reliable prediction system can provide a theoretical basis for pile foundation bearing capacity prediction and has important practical significance. Summary of the Invention

[0008] The purpose of the present invention is to provide a prediction system for the bearing capacity of static-pressure prestressed concrete piles to solve the problems raised in the above background technology. The prediction system for the bearing capacity of static-pressure prestressed concrete piles provided by the present invention has the characteristics of being able to improve project quality and safety, optimizing the design accuracy by timely adjusting the design plan, optimizing resource allocation, reducing the number of static load tests, reducing costs, and shortening the construction period.

[0009] Another purpose of the present invention is to provide an implementation method for a prediction system for the bearing capacity of static-pressure prestressed concrete piles.

[0010] To achieve the above purpose, the present invention provides the following technical solution: A prediction system for the bearing capacity of static-pressure prestressed concrete piles includes a construction monitoring module of a static-pressure pile driver, a digital static cone penetration testing module, a data receiving module, an interactive touch screen module, a bearing capacity calculation module, and a cloud data center. Among them, the construction monitoring module of the static-pressure pile driver is used to collect the construction data of the static-pressure pile driver in real time; the digital static cone penetration testing module is used to collect the physical and mechanical parameters of the soil layer around the points to be constructed of the prestressed concrete pile; the data receiving module is used to receive the construction data collected in real time by the construction monitoring module of the static-pressure pile driver, receive the physical and mechanical parameters of the soil layer obtained by the digital static cone penetration testing module through a wireless transmission method, and realize data intercommunication and storage with the cloud data center through a 5G communication network; the interactive touch screen module is used to receive and store the design information of the prestressed concrete pile; the bearing capacity calculation module adopts a feedforward neural network model, takes the physical and mechanical parameters of the soil layer, construction data, and soil layer characteristics of the pile foundation as input variables, and optimizes the parameters of the model based on historical training data to accurately capture the non-linear characteristics of the pile foundation bearing capacity; the cloud data center is used to receive and process the real-time construction data, the physical and mechanical parameters of the soil layer at each construction point, and the bearing capacity prediction information uploaded to the cloud platform server by the data receiving module through a 5G communication network.

[0011] Further in the present invention, the interactive touch screen module is used to input the physical parameters and construction design parameters of the prestressed concrete pile, or connect to the cloud data center through 5G communication, and automatically match and obtain the corresponding parameters according to the pile number.

[0012] Further in the present invention, the construction monitoring module of the static pile press includes a depth sensor, an inclination sensor, a hydraulic pressure sensor, a pile position coordinate sensor, and a joint welding quality detection system for piles; the depth sensor is used to accurately collect the penetration depth of the pile; the inclination sensor is used to accurately measure the real-time verticality of the pile after it penetrates into the soil; the hydraulic pressure sensor monitors the hydraulic pressure of the pile pressing platform of the pile press in real time and converts it into the pile pressing force applied to the top of the pile; the pile position coordinate sensor collects the construction point coordinates through high-precision positioning; the joint welding quality detection system for piles uses a deep learning model to evaluate the joint welding quality of piles.

[0013] Further in the present invention, the digital static cone penetration testing module includes a static cone penetration testing vehicle, a static cone penetration testing probe, a sounding point positioning unit, a data recording and processing unit, and a data transmission unit; the static cone penetration testing probe is used to collect the tip resistance, side friction resistance, pore water pressure of different soil layers, and the real-time azimuth angle of the probe in real time; the sounding point positioning unit is used to accurately collect the coordinate information of the static cone penetration testing points to ensure the spatial positioning accuracy of the data; the data recording and processing unit is responsible for receiving the real-time data from the static cone penetration testing probe, performing data correction based on the relationship between depth and azimuth angle, calculating and determining the penetration depth and the soil parameters of each soil layer. The calculation formula is:

[0014]

[0015] wherein, L is the total corrected penetration depth, and data is collected every 0.05 m; α i is the angle between the i-th drill pipe and the plumb line; the data transmission unit transmits the data report to the data receiving module in real time through a local wireless network, and uses 5G communication technology to transmit the data to the cloud data center efficiently and securely to achieve remote data storage and sharing.

[0016] Further in the present invention, the data receiving module is also used to preprocess the received data, including correcting the construction depth and penetration verticality of the prestressed pile to accurately calculate the actual penetration depth of the pile bottom; based on the spatial relationship between the pile position coordinates and the surrounding static cone penetration testing points, using the interpolation method to accurately determine the corresponding relationship between different depths and soil parameters at the construction point to be constructed; generating a comprehensive data set including depth, soil parameters, and construction parameters as the key input for the subsequent bearing capacity prediction model.

[0017] Further in the present invention, the implementation method of the static pressure prestressed concrete pile bearing capacity prediction system includes the following steps:

[0018] S1: The digital static cone penetration testing module conducts soil layer exploration near the construction site to be constructed, collects soil layer parameters, which include but are not limited to tip resistance, side friction resistance, and pore water pressure, and corrects the data through the real-time azimuth angle of the probe. The generated soil layer parameter data report is transmitted to the data receiving module in real time through the local wireless network and uploaded to the cloud data center using 5G communication technology;

[0019] S2: Before the construction of prestressed concrete piles, input the physical parameters of prestressed concrete piles into the interactive touch screen module. The data receiving module receives the real-time data of the static pile press construction monitoring module and the digital static cone penetration testing module, or obtains the uploaded soil layer parameter data report from the cloud data center, and accurately calculates the soil body parameter data of each soil layer at the construction site to be constructed by using the interpolation method based on the spatial relationship between the pile position coordinates and the surrounding static cone penetration points;

[0020] S3: The feedforward neural network model in the bearing capacity calculation module receives the physical parameters of the pile foundation, construction data, and soil layer characteristics as input variables, and outputs the bearing capacity prediction value in real time during the pile driving construction process. If the bearing capacity prediction value does not meet the design requirements after the construction reaches the design depth, it is judged whether to continue the pile splicing construction according to the predicted value; when the bearing capacity prediction value reaches the design standard, but there are still unspliced pile segments, the system will end the construction in advance after reporting to the construction management department;

[0021] S4: Transmit the original data collected by the static pile press construction monitoring module and the bearing capacity prediction data output by the bearing capacity calculation module to the cloud data center in real time through 5G wireless communication technology.

[0022] Further in the present invention, in step S3, the steps of the feedforward neural network model predicting the bearing capacity include:

[0023] S31: Input data processing, perform min-max standardization on the physical parameters of prestressed concrete piles, key parameters during the construction process, and soil layer geological information data included in the pile foundation bearing capacity prediction, and scale the values of the input and output parameters to the interval [0,1]. The formula is:

[0024]

[0025] where x' is the normalized value, x max and x min are the maximum and minimum values of the input data;

[0026] S32: Neural network modeling, adopt a feedforward neural network model with a single hidden layer. The input layer contains 8 neurons, representing 8 input parameters; the hidden layer contains 5 to 17 neurons, and the output layer contains one neuron, representing the predicted ultimate axial bearing capacity of the pile foundation. The output of the feedforward neural network model is calculated through the following mathematical formula:

[0027]

[0028] Among them, w ij is the weight between the i-th neuron of the input layer and the j-th neuron of the hidden layer; x i is the input data; b j is the bias of the j-th neuron in the hidden layer; θ j is the weight connecting the j-th neuron in the hidden layer and the neuron in the output layer; f(·) is the activation function;

[0029] S33: Activation function selection. The hyperbolic tangent Sigmoid function is used as the activation function in the hidden layer to effectively capture the non-linear relationship between input features and improve the fitting ability of the network; the linear activation function is used in the output layer to ensure the continuity and unconstrainedness of the output value and more accurately predict the ultimate bearing capacity of the pile foundation;

[0030] S34: Model training and optimization. The backpropagation algorithm is used to train the feedforward neural network model, combined with the Adam optimizer and the Bayesian regularization algorithm to accelerate the learning process. The early stopping mechanism is used to monitor the validation error during the training process and automatically stop the training when the validation error has not decreased for several consecutive rounds;

[0031] S35: Verification and evaluation. The Cohen's d effect size is used to measure the difference between the training set and the test set to ensure the statistical consistency of the data set, and the cross-validation technique is used to evaluate the performance of the model; the prediction accuracy and stability of the model are verified by calculating the mean square error and the coefficient of determination;

[0032] S36: Model output and application. The trained feedforward neural network model is used to predict the ultimate axial bearing capacity of the pile foundation. By inputting the physical parameters of the pile foundation, the key construction parameters and the soil layer geological information, the model outputs the predicted value of the bearing capacity of the pile foundation, which is applicable to the pile foundation design and optimization in engineering practice.

[0033] Further in the present invention, in step S34, the training and optimization steps of the feedforward neural network model include:

[0034] S341: Preliminary training. By collecting the pile foundation construction historical data in actual engineering projects, including the physical characteristics of the pile foundation, the real-time construction monitoring data and the soil layer parameters, and combining the bearing capacity measurement values of the static load test, the data is reasonably divided into a training set and a validation set; the feedforward neural network model is used to optimize the parameters through the training set, and the network weights and biases are adjusted; the hyperbolic tangent function is used as the activation function in the hidden layer to capture the complex non-linear relationship between input features and improve the fitting accuracy; the linear activation function is used in the output layer to ensure the continuity and unconstrainedness of the results and accurately predict the ultimate bearing capacity of the pile foundation;

[0035] S342: Cloud training. The cloud data center receives the pile foundation construction data and static load test data transmitted by the on-site system in real time through 5G communication technology, and uses incremental learning and online learning methods to regularly update the training set, integrate new data into the existing data set, and achieve continuous optimization of the model. The cloud data center accelerates the training using a high-performance parallel computing platform, and fine-tunes the trained model using transfer learning to make it more adaptable to new data and different construction scenarios.

[0036] S343: Edge deployment. Through 5G communication technology, the updated model is deployed to the algorithm analysis unit on-site to achieve local edge computing. With the low-latency advantage of edge computing, the on-site module can quickly process real-time data and output the pile foundation bearing capacity prediction results in real time, reducing the dependence on the cloud data center.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0038] 1. The present invention uses the static pile press construction monitoring module to collect key parameters during the construction of prestressed concrete piles in real time, such as the verticality of pile penetration, the depth of pile penetration, the speed of pile penetration, and the hydraulic pressure, etc., and obtains the physical and mechanical parameters of the soil layer around the pile foundation through the digital static cone penetration module, such as the tip resistance, the side friction resistance, and the pore water pressure, etc., which can provide necessary soil layer parameters for the analysis of the pile foundation bearing capacity.

[0039] 2. The bearing capacity calculation module of the present invention uses a feedforward neural network model, combines the physical parameters of the pile foundation, construction data, and soil layer characteristics, and optimizes the model through historical data, and can accurately predict the pile foundation bearing capacity.

[0040] 3. The feedforward neural network model of the present invention has a powerful non-linear modeling ability and a high-precision prediction ability, can effectively process complex input-output relationships, provide accurate bearing capacity prediction results, can reduce the risk of overfitting through appropriate regularization means and training strategies, and can optimize the mapping relationship through data driving to ensure the high efficiency and reliability of the system.

[0041] 4. The present invention significantly improves the prediction accuracy of the vertical ultimate bearing capacity of single piles of static PHC pipes, avoids the limitation of relying only on the results of static load tests of a few piles for pile foundation acceptance, comprehensively evaluates the bearing capacity of all piles, and thus reduces potential safety hazards.

[0042] 5. The cloud data center of the present invention is used to receive and process the real-time construction data, the soil physical and mechanical parameters of each construction point, and the bearing capacity prediction information uploaded by the data receiving module to the cloud platform server through the 5G communication network. It is responsible for storing, managing, and visually displaying the received data, and performing iterative training on the model based on the data feedback to continuously optimize the bearing capacity prediction algorithm and improve the accuracy;

[0043] 6. The interactive touch screen module of the present invention effectively integrates on-site data and cloud information, ensures the accuracy and real-time update of parameters, and provides data support for subsequent bearing capacity prediction and construction optimization;

[0044] 7. All the construction data collected by the construction monitoring module of the static pile driver of the present invention will be transmitted to the data receiving module in real time by wire and integrated, providing a key basis for the bearing capacity prediction of prestressed concrete piles;

[0045] 8. The data receiving module of the present invention combines the efficient integration and correction of multi-source data, improves the prediction accuracy, and optimizes the data application in the pile foundation design and construction process;

[0046] 9. The feedforward neural network model in the present invention is used for the correlation analysis of bearing capacity prediction data. It can not only efficiently analyze the input original data, but also has the function of receiving cloud data. Through the 5G communication module, the local system can upload the collected data to the cloud in real time, and regularly sample and label the data to enrich the training data set, thereby continuously optimizing the performance of the deep learning model. After each model iteration, the updated model is transmitted to the algorithm analysis unit on-site through the 5G module to realize the dynamic update and optimization of the on-site model. This mechanism effectively improves the real-time performance and accuracy of the ultimate bearing capacity prediction of prestressed concrete piles, ensuring that the system can continuously provide accurate prediction results in a changing construction environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is the system block diagram of the present invention;

[0048] Figure 2 is the method flow chart of the present invention;

[0049] Figure 3 is the schematic diagram of the feedforward neural network model of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0050] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0051] Embodiment 1

[0052] Please refer to Figures 1 - 3 , the present invention provides the following technical solutions: A prediction system for the bearing capacity of static pressure prestressed concrete piles, including a construction monitoring module for static pressure pile drivers, a digital static cone penetration module, a data receiving module, an interactive touch screen module, a bearing capacity calculation module, and a cloud data center. Among them, the construction monitoring module for static pressure pile drivers is used to collect the construction data of the static pressure pile driver in real time. The construction data includes but is not limited to the verticality of the prestressed concrete pile entering the soil, the penetration depth, the penetration speed, and the hydraulic pressure of the pile pressing platform of the static pressure pile driver.

[0053] Specifically, the digital static cone penetration module is used to collect the physical and mechanical parameters of the soil layer around the point where the prestressed concrete pile is to be constructed. The physical and mechanical parameters of the soil layer include but are not limited to the tip resistance (q c ), the side friction resistance (f s ), and the pore water pressure (u), providing necessary soil layer parameters for the analysis of the bearing capacity of the pile foundation.

[0054] Specifically, the data receiving module is used to receive the construction data collected in real time by the construction monitoring module for static pressure pile drivers, receive the physical and mechanical parameters of the soil layer obtained by the digital static cone penetration module through a wireless transmission method, and realize data intercommunication and storage with the cloud data center through a 5G communication network.

[0055] Specifically, the interactive touch screen module is used to receive and store the design information of the prestressed concrete pile; the user can directly input the design information through this module, or download the design document matching the pile foundation number from the cloud data center through the data receiving module. In addition, the interactive touch screen module can also display the construction data, soil layer information, and the prediction result of the pile foundation bearing capacity in real time.

[0056] Specifically, the bearing capacity calculation module adopts a feedforward neural network model. By receiving the physical and mechanical parameters of the soil layer of the pile foundation, construction data, and soil layer characteristics as input variables, and optimizing the parameters of the model based on historical training data, it can accurately capture the nonlinear characteristics of the pile foundation bearing capacity. The feedforward neural network model has powerful nonlinear modeling capabilities and high-precision prediction capabilities, can effectively handle complex input-output relationships, and provide accurate bearing capacity prediction results. In addition, through appropriate regularization methods and training strategies, this model can reduce the risk of overfitting and optimize the mapping relationship between input and output in a data-driven manner, ensuring the efficiency and reliability of the system.

[0057] Specifically, the cloud data center is used to receive and process the real-time construction data, physical and mechanical parameters of the soil layer at each construction point, and bearing capacity prediction information uploaded by the data receiving module to the cloud platform server through the 5G communication network. It is responsible for storing, managing, and visually displaying the received data, and performing iterative training of the model based on the data feedback to continuously optimize the bearing capacity prediction algorithm and improve the accuracy.

[0058] Specifically, the interactive touch screen module is used to input the physical parameters and construction design parameters of the prestressed concrete pile, or connect to the cloud data center through 5G communication, and automatically match and obtain the corresponding parameters according to the pile number. The physical parameters of the prestressed concrete pile include, but are not limited to, pile diameter, pile section length, total number of pile sections, concrete compressive strength, pile cross-section shape, and pile tip shape at the bottom of the pile. The construction design parameters include, but are not limited to, the total designed pile length, number of pile joints, designed bearing capacity, pile top elevation, and pile bottom elevation. This module effectively integrates on-site data and cloud information, ensures the accuracy and real-time update of the parameters, and provides data support for subsequent bearing capacity prediction and construction optimization.

[0059] Specifically, the static pile press construction monitoring module includes a depth sensor, an inclination sensor, a hydraulic pressure sensor, a pile position coordinate sensor, and a pile joint welding quality detection system. The depth sensor is used to accurately collect the penetration depth of the pile body. The inclination sensor is used to accurately measure the real-time verticality of the pile body after penetration. The hydraulic pressure sensor monitors the hydraulic pressure of the pile pressing platform of the pile press in real time and converts it into the pile pressing force applied to the top of the pile. The pile position coordinate sensor collects the construction point coordinates through high-precision positioning. The pile joint welding quality detection system uses a deep learning model to evaluate the pile joint welding quality. All the collected construction data will be transmitted to the data receiving module in real time through a wired method and integrated, providing a key basis for the bearing capacity prediction of prestressed concrete piles.

[0060] Specifically, the digital static cone penetration testing module includes a static cone penetration testing vehicle, a static cone penetration testing probe, a sounding point positioning unit, a data recording and processing unit, and a data transmission unit. The static cone penetration testing probe is used to collect the tip resistance (q c ) of different soil layers in real time, and the side friction resistance (fs ), pore water pressure (u), and the real-time azimuth angle (α) of the probe; the probe positioning unit is used to accurately collect the coordinate information of the static cone penetration test points to ensure the spatial positioning accuracy of the data; the data recording and processing unit is responsible for receiving the real-time data from the static cone penetration probe, and based on the relationship between depth and azimuth angle, performing data correction, calculating and determining the penetration depth and the soil parameters of each soil layer. The calculation formula is:

[0061]

[0062] Among them, L is the total corrected penetration depth, and data is collected every 0.05 m; α i is the angle between the i-th drill pipe and the plumb line; the data transmission unit transmits the data report to the data receiving module in real time through the local wireless network, and uses 5G communication technology to transmit the data to the cloud data center efficiently and securely to achieve remote data storage and sharing;

[0063] After the drilling depth is corrected, the depth of each soil layer should be redefined, and according to the corrected depth, the corresponding cone tip resistance (q c ), side friction resistance (f s ) and pore water pressure (u) are recalculated.

[0064] Specifically, the data receiving module is also used to preprocess the received data, including correcting the construction depth and the verticality of the prestressed pile to accurately calculate the actual penetration depth of the pile bottom; based on the spatial relationship between the pile position coordinates and the surrounding static cone penetration points, using the interpolation method to accurately determine the corresponding relationship between different depths and soil parameters at the points to be constructed; generating a comprehensive data set including depth, soil parameters and construction parameters as the key input for the subsequent bearing capacity prediction model. This data processing method combines the efficient integration and correction of multi-source data, improves the prediction accuracy, and optimizes the data application in the pile foundation design and construction process.

[0065] Example 2

[0066] Specifically, taking a factory building as the engineering background, PHC pipe piles are selected as the pile foundation structure, the pile type is PHC400, and it is equipped with an A-type cross pile tip. The pile length is 32 meters, and the main pile section lengths are 11 meters, 11 meters and 10 meters respectively. The pile tip bearing layer is silty clay interbedded with silty soil of layer 4. According to the design requirements, the designed vertical bearing capacity of the PHC pile is 1520 kN, and a PHC static pile driver is used for construction. The end-bearing pile form is adopted in the design, and it is required that the depth of the pile body penetrating into the bearing layer is 0.8 to 1.2 meters to ensure that the stability and bearing capacity of the pile foundation meet the engineering requirements.

[0067] Before the construction starts, the static cone penetration module will conduct soil property exploration in the area to be constructed, collect the key parameters of each soil layer at different exploration points, including the tip resistance (q c ), the side friction resistance (f s ), the pore water pressure (u), and the real-time azimuth angle (α) of the probe; meanwhile, record the plane coordinate information of each point. The data recording and processing unit is responsible for receiving the real-time data from the static cone penetration probe, and based on the relationship between depth and azimuth angle, conduct data correction, calculate and determine the penetration depth and the soil parameters of each soil layer. The data transmission unit transmits the data report in real-time to the system data receiving module installed at the static pile press through the local wireless network, and uses 5G communication technology to achieve efficient and secure data transmission to the cloud data center, so as to realize remote data storage and sharing.

[0068] In the interactive touch screen module, the user can input or select the saved construction design parameters. The user needs to input a pile length of 32 meters, 2 jointing times, a designed pile top elevation of -2 meters. The lengths of the pile sections are 11 meters, 11 meters, and 10 meters respectively, the cross-section type is circular, the pile tip type is closed type, the pile pressing speed is set at 1 meter / minute, the pile diameter is 400 mm, and the plane coordinates are (x, y). In addition, the concrete strength grade is C80. After all the parameters are input and the system receives the design information, it will initialize the data receiving module and the bearing capacity calculation module to ensure that each module is in a standby state and ready for construction operations at any time.

[0069] The data receiving module is used to receive the real-time data from the construction monitoring module of the static pile press and the digital static cone penetration module, or obtain the soil layer parameter data report uploaded by the digital static cone penetration module from the cloud platform. The data receiving module queries the exploration information of the nearby static cone penetration points according to the plane coordinates of the construction points, and based on the spatial relationship between the pile position coordinates and the surrounding static cone penetration points, uses the interpolation method to accurately determine the soil parameters at different depths at the construction points to be constructed. After data processing, generate the soil physical information of this construction point, providing accurate soil layer data support for subsequent construction and bearing capacity analysis.

[0070] After the construction starts, the construction monitoring module of the static pile press begins to collect various construction parameters in real time. The depth sensor is used to monitor the penetration depth of the pile; the inclination sensor accurately measures the real-time verticality of the pile after it enters the soil; the hydraulic pressure sensor monitors the hydraulic pressure of the pile pressing platform of the pile press in real time and converts it into the pile pressing force applied to the top of the pile. In addition, the joint welding quality detection system uses a deep learning model to evaluate the joint welding quality. All the collected construction data is transmitted to the data receiving module for integration in real time through a wired transmission method, and the construction depth and the verticality of the prestressed pile entering the soil are corrected to accurately calculate the actual penetration depth of the pile bottom. Finally, a comprehensive data set containing depth, soil parameters, and construction parameters is generated as the key input for the subsequent bearing capacity prediction model.

[0071] During the construction process, the feedforward neural network model (FFNN) in the bearing capacity calculation module receives multiple input variables such as the physical parameters of the pile foundation, construction data, and soil layer characteristics. These data are processed in real time during the construction of the static pile press and the predicted bearing capacity value of the pile foundation is output. This predicted value will be displayed in real time on the interactive touch screen module for the reference of construction personnel and management personnel. The management personnel judge whether to continue the construction based on the comparison between the displayed predicted bearing capacity value and the designed bearing capacity.

[0072] Specifically, when the pile construction reaches the designed depth, if the predicted bearing capacity value does not meet the design requirements, the system will judge whether to continue the joint pile construction according to the predicted value. If the predicted bearing capacity value does not meet the design requirements, the construction personnel can decide whether to continue the pile foundation construction according to the real-time feedback of the system to ensure that the pile foundation can meet the design bearing capacity standard.

[0073] On the contrary, when the predicted bearing capacity value reaches the design standard, but there are still unjointed sections during the construction process, the system will automatically end the construction in advance according to the predicted data after reporting to the construction management department. This intelligent dynamic adjustment method not only improves the construction efficiency, but also avoids over-construction or unnecessary resource waste, which helps to ensure the project quality and construction safety. Through this closed-loop control, the system can optimize the construction process in real time, improve the scientificity and timeliness of construction decisions, and reduce the possibility of human intervention.

[0074] After the construction is completed, through the 5G wireless communication module, the original data collected by the construction monitoring module of the static pile press and the bearing capacity prediction data output by the bearing capacity calculation module are transmitted to the background server in real time. This process realizes the efficient storage and real-time visualization display of data, and provides support for the iterative training of the deep learning model, thus promoting the continuous optimization and update of the quality monitoring algorithm.

[0075] Embodiment 3

[0076] Further, a method for implementing a bearing capacity prediction system for prestressed concrete piles by the static pressure method of the present invention includes the following steps:

[0077] S1: The digital static cone penetration testing module conducts soil layer exploration near the construction site to be constructed, collects soil layer parameters, including but not limited to tip resistance, side friction resistance, and pore water pressure, and corrects the data through the real-time azimuth angle of the probe. The generated soil layer parameter data report is transmitted to the data receiving module in real time through the local wireless network and uploaded to the cloud data center using 5G communication technology to achieve remote storage and sharing;

[0078] S2: Before the construction of prestressed concrete piles, input the physical parameters of prestressed concrete piles into the interactive touch screen module. The data receiving module receives the real-time data of the static pressure pile machine construction monitoring module and the digital static cone penetration testing module, or obtains the uploaded soil layer parameter data report from the cloud data center, and accurately calculates the soil body parameter data of each soil layer at the construction site to be constructed by using the interpolation method based on the spatial relationship between the pile position coordinates and the surrounding static cone penetration points;

[0079] S3: The feedforward neural network model in the bearing capacity calculation module receives the physical parameters of the pile foundation, construction data, and soil layer characteristics as input variables, and outputs the bearing capacity prediction value in real time during the construction of the pile machine. If the bearing capacity prediction value does not meet the design requirements after the construction reaches the design depth, it is judged whether to continue the pile splicing construction according to the predicted value; when the bearing capacity prediction value reaches the design standard, but there are still unspliced pile segments, the system will end the construction in advance after reporting to the construction management department;

[0080] S4: Transmit the original data collected by the static pressure pile machine construction monitoring module and the bearing capacity prediction data output by the bearing capacity calculation module to the cloud data center in real time through 5G wireless communication technology to achieve efficient data storage, real-time visualization display, and provide support for the iterative training of the model, thereby promoting the continuous optimization and update of the quality monitoring algorithm.

[0081] Specifically, in step S3, the steps for the feedforward neural network model to predict the bearing capacity include:

[0082] S31: Input data processing, perform min-max standardization on the physical parameters of prestressed concrete piles, key parameters during the construction process, and soil layer geological information data included in the pile foundation bearing capacity prediction, and scale the values of the input and output parameters to the interval [0,1] to improve the network training efficiency and prediction accuracy. The formula is:

[0083]

[0084] where x' is the normalized value, x max and x min are the maximum and minimum values of the input data;

[0085] S32: Neural network modeling, using a feedforward neural network model with a single hidden layer. The input layer contains 8 neurons, representing 8 input parameters; the hidden layer contains 5 to 17 neurons, and the output layer contains one neuron, representing the predicted ultimate axial bearing capacity of the pile foundation. The output of the feedforward neural network model is calculated through the following mathematical formula:

[0086]

[0087] where, w ij is the weight between the i-th neuron in the input layer and the j-th neuron in the hidden layer; x i is the input data; b j is the bias of the j-th neuron in the hidden layer; θ j is the weight connecting the j-th neuron in the hidden layer and the neuron in the output layer; f(·) is the activation function;

[0088] S33: Activation function selection. The hyperbolic tangent sigmoid function (tanh) is used as the activation function in the hidden layer to effectively capture the non-linear relationship between input features and improve the fitting ability of the network. The linear activation function (purelin) is used in the output layer to ensure the continuity and unconstrainedness of the output value and more accurately predict the ultimate bearing capacity of the pile foundation. This combined activation function strategy can balance the non-linear fitting and accurate output of the model, meet the accuracy requirements of pile foundation bearing capacity prediction, and improve the generalization ability of the network, especially when facing complex construction parameters and soil layer conditions in practical engineering applications.

[0089] S34: Model training and optimization. The feedforward neural network model is trained using the backpropagation algorithm, combined with the Adam optimizer and the Bayesian regularization algorithm (BRB) to accelerate the learning process. The early stopping mechanism (EarlyStopping) is used to monitor the validation error during the training process of the model and automatically stop the training when the validation error does not decrease for several consecutive rounds;

[0090] S35: Verification and evaluation. The Cohen's d effect size is used to measure the difference between the training set and the test set to ensure the statistical consistency of the data set, and the cross-validation technique is used to evaluate the performance of the model. The mean squared error (MSE) and the coefficient of determination (R 2 ) are calculated to verify the prediction accuracy and stability of the model;

[0091] S36: Model output and application. The trained feedforward neural network model is used to predict the ultimate axial bearing capacity of the pile foundation. By inputting the physical parameters of the pile foundation, the key construction parameters, and the soil layer geological information, the model outputs the predicted value of the bearing capacity of the pile foundation, which is applicable to the pile foundation design and optimization in engineering practice.

[0092] Specifically, in step S34, the training and optimization steps of the feedforward neural network model include:

[0093] S341: Preliminary training. By collecting historical data of pile foundation construction in actual engineering projects, including pile foundation physical characteristics, real-time construction monitoring data, and soil layer parameters, and combining with the bearing capacity measurement values of static load tests, the data is reasonably divided into a training set and a validation set; using a feedforward neural network model, parameter optimization is carried out through the training set to adjust the network weights and biases; the hyperbolic tangent function (tanh) is used as the activation function for the hidden layer to capture the complex nonlinear relationships between input features and improve the fitting accuracy; the linear activation function (purelin) is used for the output layer to ensure the continuity and unconstrainedness of the results and accurately predict the ultimate bearing capacity of the pile foundation; this combination of activation functions effectively balances the nonlinear fitting ability and output accuracy, improves the generalization ability of the model, and ensures the application accuracy in complex soil layers and construction environments.

[0094] S342: Cloud training. The cloud data center receives the pile foundation construction data and static load test data transmitted by the on-site system in real time through 5G communication technology, and uses incremental learning and online learning methods to regularly update the training set, integrating new data into the existing data set to achieve continuous optimization of the model; the cloud data center uses a high-performance parallel computing platform to accelerate training and combines transfer learning to fine-tune the trained model to make it more adaptable to new data and different construction scenarios; transfer learning can efficiently utilize existing knowledge, quickly adjust model parameters, improve the adaptability and generalization ability of the model under different geological conditions, and ensure its prediction accuracy and reliability in a changing environment.

[0095] S343: Edge deployment. Through 5G communication technology, the updated model is deployed to the algorithm analysis unit on-site to achieve local edge computing; relying on the low-latency advantage of edge computing, the on-site module can quickly process real-time data and output the pile foundation bearing capacity prediction results in real time, reducing the dependence on the cloud data center; this local processing method improves the timeliness and response speed of data processing, while reducing communication costs and latency, ensuring stability in complex construction environments, and enhancing the timeliness and accuracy of decision-making.

[0096] In summary, the present invention collects key parameters during the construction of prestressed concrete piles in real time through the construction monitoring module of the static pile driver, such as the verticality of pile penetration, the depth of pile penetration, the speed of pile penetration, and the hydraulic pressure, etc. The digital static cone penetration module is used to obtain the physical and mechanical parameters of the soil layer around the location of the pile foundation, such as the tip resistance, the side friction resistance, and the pore water pressure, etc., which can provide necessary soil layer parameters for the analysis of pile foundation bearing capacity. The bearing capacity calculation module of the present invention adopts a feedforward neural network model, combines the physical parameters of the pile foundation, construction data, and soil layer characteristics, and optimizes the model through historical data, and can accurately predict the bearing capacity of the pile foundation. The feedforward neural network model of the present invention has powerful non-linear modeling ability and high-precision prediction ability, can effectively handle complex input-output relationships, and provide accurate bearing capacity prediction results. Through appropriate regularization means and training strategies, the risk of overfitting can be reduced, and the mapping relationship can be optimized through data-driven to ensure the efficiency and reliability of the system. The present invention significantly improves the prediction accuracy of the vertical ultimate bearing capacity of single static PHC pipe piles, avoids the limitation of relying only on the results of a few pile static load tests for pile foundation acceptance, comprehensively evaluates the bearing capacity of all piles, and thus reduces potential safety hazards. The cloud data center of the present invention is used to receive and process the real-time construction data, the physical and mechanical parameters of the soil layer at each construction point, and the bearing capacity prediction information uploaded by the data receiving module to the cloud platform server through the 5G communication network, is responsible for storing, managing, and visually displaying the received data, and performs iterative training of the model based on the data feedback to achieve continuous optimization and accuracy improvement of the bearing capacity prediction algorithm. The interactive touch screen module of the present invention effectively integrates the on-site data and cloud information, ensures the accuracy and real-time update of the parameters, and provides data support for subsequent bearing capacity prediction and construction optimization. All the construction data collected by the construction monitoring module of the static pile driver of the present invention will be transmitted to the data receiving module in real time by wire and integrated, providing a key basis for the prediction of the bearing capacity of prestressed concrete piles. The data receiving module of the present invention combines the efficient integration and correction of multi-source data, improves the prediction accuracy, and optimizes the data application in the pile foundation design and construction process. The feedforward neural network model in the present invention is used for the correlation analysis of bearing capacity prediction data, not only can efficiently analyze the input original data, but also has the function of receiving cloud data. Through the 5G communication module, the local system can upload the collected data to the cloud in real time, and regularly conduct data sampling and annotation to enrich the training data set, thereby continuously optimizing the performance of the deep learning model. After each model iteration, the updated model is transmitted to the algorithm analysis unit on-site through the 5G module to realize the dynamic update and optimization of the on-site model. This mechanism effectively improves the real-time and accuracy of the prediction of the ultimate bearing capacity of prestressed concrete piles, and ensures that the system can continuously provide accurate prediction results in a changing construction environment.

[0097] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A prediction system for the bearing capacity of prestressed concrete piles by the static pressure method, characterized in that: It includes a construction monitoring module for static pile drivers, a digital static cone penetration testing module, a data receiving module, an interactive touch screen module, a bearing capacity calculation module, and a cloud data center. Among them, The construction monitoring module for static pile drivers is used to collect the construction data of static pile drivers in real time; The digital static cone penetration testing module is used to collect the physical and mechanical parameters of the soil layer around the points where prestressed concrete piles are to be constructed; The data receiving module is used to receive the construction data collected in real time by the construction monitoring module for static pile drivers, receive the physical and mechanical parameters of the soil layer obtained by the digital static cone penetration testing module through wireless transmission, and realize data intercommunication and storage with the cloud data center through the 5G communication network; The interactive touch screen module is used to receive and store the design information of prestressed concrete piles; The bearing capacity calculation module adopts a feedforward neural network model. By receiving the physical and mechanical parameters of the soil layer of the pile foundation, construction data, and soil layer characteristics as input variables, and optimizing the parameters of the model based on historical training data, it accurately captures the nonlinear characteristics of the bearing capacity of the pile foundation; The cloud data center is used to receive and process the real-time construction data, the physical and mechanical parameters of the soil layer at each construction point, and the bearing capacity prediction information uploaded by the data receiving module to the cloud platform server through the 5G communication network.

2. The prediction system for the bearing capacity of a static pressure prestressed concrete pile according to claim 1, wherein: The interactive touch screen module is used to input the physical parameters and construction design parameters of prestressed concrete piles, or connect to the cloud data center through 5G communication, and automatically match and obtain the corresponding parameters according to the pile number.

3. A prediction system for the bearing capacity of static pressure prestressed concrete piles according to claim 1, characterized in that: The construction monitoring module for static pile drivers includes a depth sensor, an inclination sensor, a hydraulic pressure sensor, a pile position coordinate sensor, and a joint welding quality detection system for piles.

4. A prediction system for the bearing capacity of static pressure prestressed concrete piles according to claim 3, characterized in that: The depth sensor is used to accurately collect the penetration depth of the pile body; the inclination sensor is used to accurately measure the real-time verticality of the pile body after penetration; the hydraulic pressure sensor monitors the hydraulic pressure of the pile pressing platform of the pile driver in real time and converts it into the pile pressing force applied to the top of the pile; the pile position coordinate sensor collects the construction point coordinates through high-precision positioning; the joint welding quality detection system for piles uses a deep learning model to evaluate the joint welding quality of piles.

5. A prediction system for the bearing capacity of static pressure prestressed concrete piles according to claim 1, characterized in that: The digital static cone penetration testing module includes a static cone penetration testing vehicle, a static cone penetration testing probe, a sounding point positioning unit, a data recording and processing unit, and a data transmission unit.

6. The prediction system for the bearing capacity of static pressure prestressed concrete piles according to claim 5, characterized in that: The static cone penetration testing probe is used to collect the tip resistance, side friction resistance, pore water pressure of different soil layers, and the real-time azimuth angle of the probe in real time; the sounding point positioning unit is used to accurately collect the coordinate information of the static cone penetration testing points to ensure the spatial positioning accuracy of the data; the data recording and processing unit is responsible for receiving the real-time data from the static cone penetration testing probe, performing data correction based on the relationship between depth and azimuth angle, calculating and determining the penetration depth and the soil parameters of each soil layer. The calculation formula is: Among them, L is the total corrected penetration depth, and data is collected every 0.05 m; α i is the angle between the i-th drill pipe and the plumb line; the data transmission unit transmits the data report to the data receiving module in real time through the local wireless network, and uses 5G communication technology to transmit the data to the cloud data center efficiently and securely to achieve remote data storage and sharing.

7. A prediction system for the bearing capacity of static pressure prestressed concrete piles according to claim 1, characterized in that: The data receiving module is also used to preprocess the received data, including correcting the construction depth and penetration verticality of prestressed piles to accurately calculate the actual penetration depth of the pile bottom; based on the spatial relationship between the pile position coordinates and the surrounding static cone penetration testing points, using the interpolation method to accurately determine the corresponding relationship between different depths and soil parameters at the points to be constructed; generating a comprehensive dataset including depth, soil parameters, and construction parameters as the key input for the subsequent bearing capacity prediction model.

8. The implementation method of a static pressure method prestressed concrete pile bearing capacity prediction system according to any one of claims 1-7, characterized in that, It includes the following steps: S1: The digital static cone penetration testing module conducts soil layer exploration near the construction site to be constructed, collects soil layer parameters, which include but are not limited to tip resistance, side friction resistance, and pore water pressure, and corrects the data through the real-time azimuth angle of the probe. The generated soil layer parameter data report is transmitted to the data receiving module in real time through the local wireless network and uploaded to the cloud data center using 5G communication technology; S2: Before the construction of prestressed concrete piles, input the physical parameters of prestressed concrete piles into the interactive touch screen module. The data receiving module receives the real-time data of the static pile driver construction monitoring module and the digital static cone penetration testing module, or obtains the uploaded soil layer parameter data report from the cloud data center, and accurately calculates the soil body parameter data of each soil layer at the construction site to be constructed by using the interpolation method based on the spatial relationship between the pile position coordinates and the surrounding static cone penetration points; S3: The feedforward neural network model in the bearing capacity calculation module receives the physical parameters of the pile foundation, construction data, and soil layer characteristics as input variables, and outputs the bearing capacity prediction value in real time during the pile driver construction process. If the bearing capacity prediction value does not meet the design requirements after the construction reaches the design depth, it is judged whether to continue the pile splicing construction according to the predicted value; when the bearing capacity prediction value reaches the design standard, but there are still unspliced pile segments, the system will end the construction in advance after reporting to the construction management department; S4: Transmit the original data collected by the static pile driver construction monitoring module and the bearing capacity prediction data output by the bearing capacity calculation module to the cloud data center in real time through 5G wireless communication technology.

9. The implementation method of a static pressure method prestressed concrete pile bearing capacity prediction system according to claim 8, characterized in that: In the step S3, the steps for the feedforward neural network model to predict the bearing capacity include: S31: Input data processing, perform min-max standardization on the physical parameters of prestressed concrete piles, key parameters during the construction process, and soil layer geological information data included in the pile foundation bearing capacity prediction, and scale the values of the input and output parameters to the interval [0,1]. The formula is: where x is the normalized value, x max and x min are the maximum and minimum values of the input data; S32: Neural network modeling, adopt a feedforward neural network model with a single hidden layer. The input layer contains 8 neurons, representing 8 input parameters; the hidden layer contains 5 to 17 neurons, and the output layer contains one neuron, representing the predicted ultimate axial bearing capacity of the pile foundation. The output of the feedforward neural network model is calculated through the following mathematical formula: Among them, w ij is the weight between the i-th neuron in the input layer and the j-th neuron in the hidden layer; x i is the input data; b j is the bias of the j-th neuron in the hidden layer; θ j is the weight connecting the j-th neuron in the hidden layer to the neuron in the output layer; f(·) is the activation function; S33: Activation function selection, the hyperbolic tangent sigmoid function is used as the activation function for the hidden layer to effectively capture the non-linear relationship between input features and improve the fitting ability of the network; the linear activation function is used for the output layer to ensure the continuity and unconstraint of the output value and more accurately predict the ultimate bearing capacity of the pile foundation; S34: Model training and optimization, adopt the backpropagation algorithm to train the feedforward neural network model, combine the Adam optimizer and the Bayesian regularization algorithm to accelerate the learning process, adopt the early stopping mechanism to monitor the validation error of the model during the training process, and automatically stop the training when the validation error does not decrease for several consecutive rounds; S35: Verification and evaluation. Measure the differences between the training set and the test set through Cohen's d effect size to ensure the statistical consistency of the dataset. Use cross-validation techniques to evaluate the performance of the model. Verify the prediction accuracy and stability of the model by calculating the mean squared error and the coefficient of determination. S36: Model output and application. The trained feedforward neural network model is used to predict the ultimate axial bearing capacity of pile foundations. By inputting the physical parameters of pile foundations, key construction parameters, and soil layer geological information, the model outputs the predicted values of the bearing capacity of pile foundations, which is applicable to the design and optimization of pile foundations in engineering practice.

10. The implementation method of a static pressure method prestressed concrete pile bearing capacity prediction system according to claim 9, characterized in that: In the step S34 described above, the training and optimization steps of the feedforward neural network model include: S341: Preliminary training. Collect the historical data of pile foundation construction in actual engineering projects, including the physical characteristics of pile foundations, real-time construction monitoring data, and soil layer parameters, and combine with the bearing capacity measurement values of static load tests. Reasonably divide the data into a training set and a validation set. Use a feedforward neural network model to optimize the parameters through the training set, and adjust the network weights and biases. The hyperbolic tangent function is used as the activation function for the hidden layer to capture the complex nonlinear relationships between input features and improve the fitting accuracy. The linear activation function is used for the output layer to ensure the continuity and unconstrainedness of the results and accurately predict the ultimate bearing capacity of pile foundations. S342: Cloud training. The cloud data center receives the pile foundation construction data and static load test data transmitted by the on-site system in real time through 5G communication technology. Use incremental learning and online learning methods to regularly update the training set, integrate the new data into the existing dataset, and achieve continuous optimization of the model. The cloud data center uses a high-performance parallel computing platform to accelerate training and combines transfer learning to fine-tune the trained model to make it more adaptable to new data and different construction scenarios. S343: Edge deployment. Deploy the updated model to the on-site algorithm analysis unit through 5G communication technology to achieve local edge computing. With the low-latency advantage of edge computing, the on-site module can quickly process real-time data and output the predicted results of the pile foundation bearing capacity in real time, reducing the dependence on the cloud data center.

Images