Under the background of the rapid development of highway bridge and bridge construction, the load of the superstructure of the bridge is increasing, which brings great challenges to the construction of foundation engineering. Necessary for research of foundation engineering. Bored piles. When the bearing capacity of bored piles is analyzed, the BP neural network can simulate any complex sample feature mapping, and has nonlinear optimization function. BP grid has strong learning and imitation ability in many fields, strong generalization and high robustness. In this paper, the BP neural network is used for the prediction of the bearing capacity of the pile foundation in the drilling construction of highway bridges, which reduces the errors caused by some semi-empirical formulas obtained by the traditional statistical regression analysis, and the prediction effect is good
The focus of bridge construction is the foundation, and a good foundation is an important guarantee for the success of bridge construction. Due to the needs of foundation construction, hole-forming machinery has been vigorously developed. At present, the diameter of the bored pile construction can reach more than 4 meters, and the length of the pile can even reach more than 100 meters. The design and construction of the bridge superstructure and the pier and abutment directly determine the capacity. Throughout the current research situation, the quality of bored piles is closely related to foundation construction. In the actual construction process, how to choose the correct and reasonable construction plan according to the different hydrogeological conditions and the construction characteristics of the project, and how to prevent and deal with the quality problems that may occur in the construction, so that the selected plan can guarantee the whole Construction quality of the project. At the same time, the construction unit often conducts construction according to previous construction experience in the construction process, but lacks the ability to judge the construction quality. It is necessary to find out the key points of quality control to ensure the construction quality of bored piles.
Uncertainty factors must be quantitatively studied in the bearing capacity analysis of bored piles. For example, soil parameter uncertainty, geometric parameter uncertainty and computational model uncertainty. Back Propagation (BP) neural networks [2, 3] can simulate arbitrarily complex mappings with nonlinear optimization capabilities. Engineering application based on BP neural network is a hot research topic in recent years. Applying neural network to predict the bearing capacity of pile foundation can overcome the difficulties caused by the incompatibility of the problem to a certain extent. It avoids the errors caused by some semi-empirical formulas obtained by traditional statistical regression analysis, and has a good development prospect.
What is the key of Design Method of Bored Piles for Highway Bridges?
When choosing the pile type [4,5], factors such as the load on the superstructure, engineering geological conditions, environmental conditions around the construction site, construction feasibility, and construction cost and construction period should be considered. Each pile type has its own characteristics and Applicable conditions, when selecting the pile type, according to the actual situation of the construction site, a comprehensive study is carried out on its rationality, economy, construction feasibility, etc., and finally a safe, applicable, economical and reasonable pile type is selected.
When selecting a bearing layer, first consider whether the upper load and settlement deformation meet the requirements. If the requirements are met, then consider the requirements of rationality, construction feasibility and economy. Under normal circumstances, the bearing layer should be selected on a hard soil layer, which can ensure the pile have a larger bearing capacity. If there is no hard soil layer when selecting the bearing layer, try to choose a soil layer with high strength and low compressibility to minimize the settlement of the pile foundation.
When choosing the length of the pile, the factors such as soil properties, load size, geological conditions, the interaction between the pile and soil, the buckling instability, the length-diameter ratio of the pile, and the economic conditions should be considered comprehensively. The determination of pile length and pile diameter are mutually restricted. On the premise of meeting the requirements, the factors of construction cost should be considered to maximize the bearing capacity of the pile foundation. Generally, according to the geological conditions of the investigation and the feasibility of the construction, the appropriate bearing layer at the pile end is selected to preliminarily determine the length of the pile. When selecting the pile diameter, it should be determined in combination with the construction characteristics of the pile, construction experience and commonly used dimensions.
After the appropriate pile type is determined, a specific calculation method for the bearing capacity of a single pile is selected according to the actual construction situation of the project.
When checking the vertical force at the top of the pile, when the standard combination of load effects is under the action of the vertical force of the axis.
the average vertical force of the foundation pile under the action of the vertical force of the standard combined axis of the load effect. is the characteristic value of vertical bearing capacity of foundation pile or composite foundation pile.
Before drilling construction of highway bridges, the following preparations need to be completed. First, after measuring and releasing the pile site, the position of the steel casing is determined by the cross-hair method, and then the pile site is checked and the steel casing is buried. Then carry out the positioning of the drilling rig, check whether the bottom of the drilling rig foundation is firm, correct the verticality of the drill bit, and start drilling after meeting the requirements
During drilling construction, electronic control and manual observation are usually used to ensure the verticality of the hole formed by the pile. The drilling speed is determined according to the geological conditions of the construction site. If drilling from a hard stratum to a soft stratum, the drilling speed can be appropriately increased. If drilling from soft stratum to hard stratum, slow down the speed appropriately. When encountering a sand layer, the specific gravity and viscosity of the mud should be appropriately increased, and the construction should be carried out by means of slow speed and slow drilling. Before drilling, it is necessary to review the performance indicators of the mud to ensure that the quality of the mud meets the requirements. If there is a problem with the mud indicator, find out the reason and adjust it in time.
The capacity[6,7] of a single pile depends on the material strength of the pile. If the supporting force of soil to the pile is greater than the material strength of the pile body, the bearing capacity of the pile should be calculated according to the maximum compressive strength of the material. If the supporting force of the soil on the pile is less than the material strength of the pile itself, the bearing capacity of the pile can only be determined by the strength and deformation of the soil, which is the case for most friction piles.
Since the ultimate bearing capacity of a single pile of most piles is formed by the combined action of the pile and soil, the factors affecting the ultimate bearing capacity of a single pile include the engineering properties of the soil around the pile, the pile forming process of the pile foundation and the length of the pile.
The pile forming process of the pile foundation has a certain influence on the bearing capacity of the pile. Even if the same type of piles are constructed under the same geological conditions, different pile forming techniques will lead to great changes in the bearing capacity of the piles. The influence of the pile forming process on the bearing capacity of the pile foundation is mainly manifested in that different pile forming processes have different degrees of disturbance to the soil, thereby affecting the bearing performance of the pile. The length of the pile has a great influence on the bearing capacity of the pile. The indoor model test and the field prototype test show that the pile tip resistance does not increase linearly with the depth of the pile, but varies with the depth of the soil, especially the depth of the bearing layer
During the construction of bored piles, due to the improper operation of the construction personnel, it is very likely to cause construction accidents. In order to avoid such accidents, during the construction of the cast-in-place piles, it is necessary to strengthen the quality of the holes and the quality of the piles. Control and management, put forward relevant quality control points to ensure the smooth operation of construction.
The verticality, hole depth, drilling speed and mud quality of the pile body are the most important factors that affect the pile hole formation. The verticality of the pile hole is the first control point in the hole-forming construction stage. If the requirements are not met, it will affect the subsequent construction process and cannot be carried out. In order to avoid the problem of pile hole inclination in the subsequent construction, the verticality of the drill bit should be reviewed, and the positioning of the drill and the foundation at the bottom of the drill should be checked to ensure that the verticality of the pile body meets the design and construction requirements. When the protective tube is buried, use GPS to check the elevation of the top surface of the protective tube to ensure that the allowable error of the center line does not exceed 5 cm. In order to ensure the quality of the hole, the depth of the hole must be controlled, mainly by measuring the length of the drill pipe or the steel measuring rope
Pile foundation inspection includes two aspects: integrity and bearing capacity. The low strain method is mainly used to test whether the integrity of the pile body meets the requirements. The basic principle of the low-strain detection method is to apply a force on the top of the pile foundation to make the top of the pile vibrate, and the vibration propagates from the top of the pile to the bottom of the pile in the form of stress waves. If there is a defect in the pile body, due to the change of the wave impedance, the stress wave will be emitted when it encounters the defect, and finally it will be reflected to the sensor on the top of the pile, and the returned stress wave will be transmitted back to the sensor according to the time and frequency. As well as the amplitude and other factors, the measured curve of the stress wave can be obtained, and the integrity of the pile body can be determined according to the curve.
The detection of the bearing capacity of the pile foundation is the most important part in the detection of the pile foundation. Among all the methods of pile foundation detection, the static load test method is the most intuitive and reliable method to determine the bearing capacity of a single pile. When doing a static load test, various factors affecting the use of the pile can be considered. Intuitiveness and reliability are the biggest advantages of the static load test method for pile foundation inspection. Each detection method has its own advantages. When choosing, we should choose the method suitable for engineering detection according to the actual situation of the construction site.
Application of BP Neural Network in Construction of Bored Piles of Highway Bridges
In the information age, with the improvement of information level and the breakthrough of computer algorithms, computer intelligent algorithms have completed the deep integration with many fields. Computer technology is widely used in 3D printing technology [8], disease prevention and diagnosis [9], pregnancy examination and evaluation [10], treatment of the COVID-19 epidemic [11], image segmentation [12] and many other fields. Neural networks are a typical representative of intelligent algorithms, and are gradually becoming mature in industries such as deep learning, image processing, smart wear, and engineering construction
Neurons are the most basic concept in deep learning [13], and the current mainstream neural networks are almost all composed of neurons in different ways. A complete neuron mainly consists of two parts, the linear function and the excitation function. The linear function expression is
The input is the data before neuron processing, x is not necessarily a number, it can be a matrix or other data. The output is the data processed by neurons, and the output data can also be data in various forms, which are determined by the input data and neurons. Weight processing The data multiplied when the input enters the neuron, which can be a number or a matrix. Generally, the initial weights are set randomly. After training, the computer will adjust the weights by itself, and assign larger weights to relatively important features, while Conversely, less important features are assigned smaller weights. Bias adds a linear component to the result of multiplying the input and the weight, which increases the linear range.
The role of the activation function is to add nonlinearity to the neuron. Because the content that linear functions can express has great limitations, many problems in reality are nonlinear. Therefore, adding nonlinear factors can increase the fitting effect of the model.
what is BP neural network?
The BP [16] learning algorithm is one of the most successful neural network learning algorithms at present. When neural networks are used in learning tasks, most of them are trained using the BP algorithm. The BP neural network performs error analysis according to the results obtained from the training and the expected results each time, and then modifies the weights and thresholds to obtain a model that can output consistent with the expected results step by step
The main factors affecting the bearing capacity of a single pile are the length of the pile, the diameter of the pile, the distribution of the soil layer around the pile and the physical and mechanical indicators, as well as the physical and mechanical indicators of the soil layer at the end of the pile. In the experiment, the static load test of 14 local road and bridge bored piles was collected as a learning sample, and a two-level network was established using python to train and predict the bearing capacity of a single pile. The activation function selects the Sigmoid function, and the relative error threshold of the two networks is set to 3%. The data of piles 1-10 were used for simulation experiments, and piles 11-14 were used for model verification.
The BP network inputs the pile length, pile diameter, bearing capacity of the pile end, soil indexes of the pile side and the end of the pile, and geometric dimensions of the pile, and the output is the ultimate bearing capacity. After pre-experiment, the network has 12 hidden layer units, and the number of iterations is 1236. The observed and predicted values are shown in the table below.
Table 1 Simulation and test results
| Number | Diameter (mm) | immersion depth (m) | Measured value (kN) | Predictive value (kN) | forecast relative error |
| 1 | 800 | 50 | 7200 | 6820 | -5.28% |
| 2 | 800 | 55.4 | 7500 | 7880 | 5.07% |
| 3 | 800 | 53.2 | 6430 | 6080 | -5.44% |
| 4 | 800 | 41.5 | 4760 | 4960 | 4.20% |
| 5 | 800 | 45.7 | 5320 | 5690 | 6.95% |
| 6 | 800 | 40.6 | 5180 | 5400 | 4.25% |
| 7 | 800 | 45.4 | 8540 | 8110 | -5.04% |
| 8 | 800 | 48.9 | 8900 | 9200 | 3.37% |
| 9 | 800 | 52.3 | 7890 | 8300 | 5.20% |
| 10 | 800 | 54.6 | 9200 | 8420 | -8.48% |
| 11 | 800 | 55.5 | 6750 | 6430 | -4.74% |
| 12 | 800 | 65.4 | 9500 | 9100 | -4.21% |
| 13 | 800 | 51.6 | 6700 | 6950 | 3.73% |
| 14 | 800 | 58.7 | 6500 | 6830 | 5.08% |
From Simulation and test results, BP predicts the prediction of the ultimate bearing capacity of the bored pile, and the relative error of the prediction is mostly within 10%. The relative prediction error of the No. 10 pile is the largest, the measured value is 9200kN, the predicted value of the BP neural network is 8420 kN, and the predicted value is 780kN lower than the actual bearing capacity. The relative prediction error of the No. 10 pile is the smallest, the measured value is 8900kN, the predicted value of the BP neural network is 9200 kN, and the predicted value is 300 kN higher than the actual bearing capacity.
In the simulation experiments of No. 1-10 piles, the predicted bearing capacity of 4 piles is lower than the measured bearing capacity of the piles, and the predicted bearing capacity of 6 piles is higher than the measured bearing capacity of the piles. In the inspection samples of No. 11-14 piles, the predicted bearing capacity of 2 piles is lower than the measured bearing capacity of the piles, and the predicted bearing capacity of 2 piles is higher than the measured bearing capacity of the piles. For the pile foundation engineering in the actual construction of bored piles of highway bridges, the error is reasonable and acceptable.
It can be seen from the above analysis and engineering examples that the BP neural network can better simulate the complex nonlinear mapping relationship in the data of highway bridge bored piles. The maximum error of simulation and prediction can meet the error requirements of pile foundation engineering, indicating that BP has a good development prospect in predicting the bearing capacity of single pile
example:
end:
BP neural network is a learning algorithm with tutors, which can adjust the weights and thresholds repeatedly by comparing the error between the actual output and the expected output of the sample after learning through the network, and gradually achieve the predetermined accuracy. Therefore, the prediction results do not have human factors and are objective. It is a new attempt to introduce BP neural network to predict the vertical ultimate bearing capacity of single pile. Prediction instance analysis shows that BP neural network prediction has better accuracy. Compared with the previous empirical formula, the prediction of ultimate bearing capacity has a great improvement in accuracy and can be used for reference in engineering.
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
