Deep Learning Method, Device, Equipment and Medium for Inverting Vibration Piling Parameters

Abstract

The present invention relates to a deep learning method, device, equipment and medium for dynamic inversion of dynamic parameters of vibro - pile driving, and the method includes: determining the known parameters and unknown parameters of the vibro - pile driving model, defining the known parameters as constant parameters in the deep learning framework, and defining the unknown parameters as variable parameters in the deep learning framework; constructing the dynamic partial differential equation of the vibro - pile driving foundation, and defining the dynamic partial differential equation as physical knowledge in the deep learning framework; obtaining the real - time measured vibro - pile driving data, and using the vibro - pile driving data as the driving data for parameter inversion; based on the constructed deep learning framework including physical knowledge embedding and the driving data, performing the deep learning process of dynamic inversion of dynamic parameters of vibro - pile driving, and outputting the parameter values obtained through deep learning inversion. The present invention improves the credibility of the parameter inversion of the vibro - pile driving dynamic model by embedding the dynamic partial differential equation as physical knowledge into the deep learning framework.

Description

Technical Field

[0001] The present invention relates to the technical field of vibration pile driving parameter inversion, and particularly relates to a deep learning method, device, equipment and medium for dynamically inverting the dynamic parameters of vibration pile driving. Background Technique

[0002] With the gradual penetration of the concept of green environmental protection and the continuous improvement of the national requirements for noise pollution prevention and control, the vibration pile driving process with low noise pollution has been popularized and used, and the vibration pile driving process is widely applied to the pile driving operations of pipe piles and solid piles. The basic principle of vibration pile driving is to liquefy the soil around the pile foundation through the periodic external force of the vibration hammer, so as to reduce the frictional resistance between the pile and the soil, and thus achieve the purpose of pile driving.

[0003] The interaction between the pile and the soil is an extremely complex dynamic problem, that is, the dynamic characteristics of the soil itself and the dynamic contact behavior between the pile and the soil are extremely complex. During the vibration pile driving construction, the soil conditions around the pile foundation and the frictional resistance between the pile and the soil will have obvious non-linear dynamic changes with the change of the construction process. At present, many analysis studies on the dynamic parameters of pile driving obtained based on engineering sites and indoor tests have given the general change laws and ranges of the resistance between the pile and the soil. However, due to the non-uniformity, anisotropy of the soil and the static and random nature of the empirical formula, the credibility of the existing inversion of the dynamic model parameters of vibration pile driving is greatly reduced.

[0004] Therefore, it is an urgent problem to be solved what method to choose to invert the parameters of the vibration pile driving dynamic model to improve the credibility of the inversion of the vibration pile driving dynamic model parameters. Summary of the Invention

[0005] In view of this, it is necessary to provide a deep learning method, device, equipment and medium for dynamically inverting the dynamic parameters of vibration pile driving to improve the credibility of the inversion of the dynamic model parameters of vibration pile driving.

[0006] In a first aspect, the present invention provides a deep learning method for dynamically inverting the dynamic parameters of vibration pile driving, including:

[0007] Determine the known parameters and unknown parameters of the vibration pile driving model, define the known parameters as constant parameters in the deep learning framework, and define the unknown parameters as variable parameters in the deep learning framework;

[0008] Construct the dynamic partial differential equation of the vibration pile driving pile foundation, and define the dynamic partial differential equation as physical knowledge in the deep learning framework;

[0009] Obtain the vibration pile driving data measured in real time, and use the vibration pile driving data as the driving data for parameter inversion;

[0010] Based on the constructed deep learning framework including physical knowledge embedding and the driving data, perform a deep learning process for dynamic inversion of the dynamic parameters of vibro - pile sinking, and output the parameter values obtained through deep learning inversion.

[0011] Furthermore, the known parameters of the vibro - pile sinking model include the geometric parameters of the pile foundation, the material parameters of the pile foundation, the soil parameters, and the vibro - hammer parameters;

[0012] Among them, the geometric parameters of the pile foundation include the length, radius, and thickness of the pile foundation;

[0013] The material parameters of the pile foundation include the elastic modulus, Poisson's ratio, and density of the pile foundation;

[0014] The soil parameters include the density parameter of the soil;

[0015] The vibro - hammer parameters include the exciting force, penetration force, penetration frequency, theoretical power, and amplitude of the vibro - hammer.

[0016] Furthermore, the unknown parameter of the vibro - pile sinking model includes the dynamic stress - strain curve relationship.

[0017] Furthermore, the construction of the dynamic partial differential equation of the vibro - pile sinking foundation includes: constructing the dynamic partial differential equation of the vibro - pile sinking foundation based on the exciting force of the vibro - hammer, the gravity of the pile foundation, the frictional force on the side of the pile foundation by the soil, and the resistance at the end.

[0018] Furthermore, the vibro - pile sinking data includes the data measured for the partial vibration of the pile foundation above the soil;

[0019] The vibro - pile sinking data includes any one of the displacement, velocity, and acceleration in the axial direction of the pile foundation.

[0020] Furthermore, the constructed deep learning framework including physical knowledge embedding is formulated based on the computing conditions and performance available in engineering practical applications;

[0021] The parameters that need to be formulated for the neural network corresponding to the deep learning framework include:

[0022] Input and output quantities, number of neurons, number of neuron layers, activation function, initial state, optimizer, and learning rate.

[0023] Furthermore, the deep learning process for dynamic inversion of the dynamic parameters of vibro - pile sinking based on the constructed deep learning framework including physical knowledge embedding and the driving data, and outputting the parameter values obtained through deep learning inversion, includes:

[0024] Using the axial coordinates of the pile foundation as the input of the neural network and the axial displacement of the pile foundation as the output of the neural network, through continuous iterative calculations, the loss function including the dynamic partial differential equation and the measured vibration pile driving data is minimized, and the vibration pile driving dynamics results and inversion parameters are output.

[0025] In a second aspect, the present invention also provides a deep learning device for dynamically inverting the dynamic parameters of vibration pile driving:

[0026] A parameter definition module, configured to determine the known parameters and unknown parameters of the vibration pile driving model, define the known parameters as constant parameters in the deep learning framework, and define the unknown parameters as variable parameters in the deep learning framework;

[0027] A physical knowledge definition module, configured to construct a dynamic partial differential equation of the pile foundation of vibration pile driving, and define the dynamic partial differential equation as physical knowledge in the deep learning framework;

[0028] A data acquisition module, configured to acquire the vibration pile driving data measured in real time, and use the vibration pile driving data as the driving data for parameter inversion;

[0029] A learning module, configured to perform a deep learning process for dynamically inverting the dynamic parameters of vibration pile driving based on the constructed deep learning framework including physical knowledge embedding and the driving data, and output the parameter values obtained by deep learning inversion.

[0030] In a third aspect, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned deep learning method for dynamically inverting the dynamic parameters of vibration pile driving are implemented.

[0031] In a fourth aspect, the present invention also provides a computer storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned deep learning method for dynamically inverting the dynamic parameters of vibration pile driving are implemented.

[0032] The beneficial effects of adopting the above embodiments are as follows:

[0033] By embedding the dynamic partial differential equation as physical knowledge into the deep learning framework, the present invention can provide scientific and definite physical information for the deep learning neural network. Then, based on the basic physical knowledge, the measured data in engineering construction or experimental testing is used as the driving data of the vibration pile driving dynamics model for guidance and correction. Finally, the proposed deep learning method for inverting the dynamic parameters of vibration pile driving is dynamic, which can fully reflect the change process of soil characteristics and the interaction between pile and soil, and improve the credibility of the inversion of the dynamic parameters of the vibration pile driving model. Description of the Drawings

[0034] Figure 1 It is a schematic flowchart of an embodiment of a deep learning method for dynamic inversion of vibration pile driving dynamics parameters provided by the present invention;

[0035] Figure 2 It is a schematic structural diagram of a vibration pile driving provided by an embodiment of the present invention;

[0036] Figure 3 It is a deep learning framework diagram for dynamic inversion of vibration pile driving dynamics parameters provided by an embodiment of the present invention;

[0037] Figure 4 It is a schematic structural diagram of an embodiment of a deep learning device for dynamic inversion of vibration pile driving dynamics parameters provided by the present invention;

[0038] Figure 5 It is a schematic structural diagram of an electronic device provided by the present invention. Detailed Embodiments

[0039] The following will specifically describe the preferred embodiments of the present invention in conjunction with the drawings. The drawings form a part of this application and are used together with the embodiments of the present invention to explain the principles of the present invention, rather than to limit the scope of the present invention.

[0040] In the description of the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the meaning of "a plurality" is two or more, unless otherwise specifically defined. Referring to "embodiment" herein means that a specific feature, structure, or characteristic described in conjunction with the embodiment may be included in at least one embodiment of the present invention. The appearance of this phrase at various positions in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein may be combined with other embodiments.

[0041] The present invention provides a deep learning method, device, equipment and medium for dynamic inversion of vibration pile driving dynamics parameters. By combining the deep learning method with the vibration pile driving dynamics model, considering the objectivity of the physical model and the timeliness of real-time measurement data, a deep learning method for dynamic inversion of vibration pile driving dynamics parameters is established based on data-driven and physical embedding to achieve the training, verification and inversion of the dynamics parameters in the vibration pile driving process. The following will separately elaborate on the specific embodiments in detail:

[0042] Please refer toFigure 1 , Figure 1 This is a schematic flowchart of an embodiment of a deep learning method for dynamic inversion of vibration pile driving dynamic parameters provided by the present invention. A specific embodiment of the present invention discloses a deep learning method for dynamic inversion of vibration pile driving dynamic parameters, including:

[0043] Step S101: Determine the known parameters and unknown parameters of the vibration pile driving model, define the known parameters as constant parameters in the deep learning framework, and define the unknown parameters as variable parameters in the deep learning framework;

[0044] Step S102: Construct the dynamic partial differential equation of the vibration pile driving foundation, and define the dynamic partial differential equation as physical knowledge in the deep learning framework;

[0045] Step S103: Obtain the vibration pile driving data measured in real time, and use the vibration pile driving data as the driving data for parameter inversion;

[0046] Step S104: Based on the constructed deep learning framework including physical knowledge embedding and the driving data, perform the deep learning process of dynamic inversion of vibration pile driving dynamic parameters, and output the parameter values after deep learning inversion.

[0047] By embedding the dynamic partial differential equation as physical knowledge into the deep learning framework, the present invention can provide scientific and definite physical information for the deep learning neural network. Then, based on the basic physical knowledge, the measured data in engineering construction or experimental tests is used as the driving data of the vibration pile driving dynamic model for guidance and correction. Finally, the proposed deep learning method for the inversion of vibration pile driving dynamic parameters is dynamic, which can fully reflect the change process of soil characteristics and the interaction between pile and soil, and improve the credibility of the inversion of vibration pile driving dynamic model parameters.

[0048] In an embodiment of the present invention, the known parameters of the vibration pile driving model include the geometric parameters of the pile foundation, the material parameters of the pile foundation, the soil parameters, and the vibration hammer parameters;

[0049] Among them, the geometric parameters of the pile foundation include the length, radius, and thickness of the pile foundation;

[0050] The material parameters of the pile foundation include the elastic modulus, Poisson's ratio, and density of the pile foundation;

[0051] The soil parameters include the density parameter of the soil;

[0052] The vibration hammer parameters include the exciting force, penetration force, penetration frequency, theoretical power, and amplitude of the vibration hammer.

[0053] It can be understood that the known parameters of the vibro - pile driving model generally include the determinable parameters among the basic parameters such as the geometric and material parameters of the pile foundation, soil parameters, and vibratory hammer parameters, etc., which can be input into the deep - learning framework as constant parameters;

[0054] Specifically, the known parameters in this embodiment include the geometric parameters and material parameters of the pile foundation. Please refer to Figure 2 , Figure 2 which is a schematic structural diagram of a vibro - pile driving provided by an embodiment of the present invention. Combining Figure 2 , the known parameters in this embodiment include that the length of the pile foundation in the soil is L, the radius is R, the density of the pile foundation is ρ, and the one - dimensional elastic longitudinal wave velocity of the pile foundation is C. The above parameters are defined as constant parameters in the deep - learning framework.

[0055] In an embodiment of the present invention, the unknown parameters of the vibro - pile driving model include the dynamic stress - strain curve relationship.

[0056] It can be understood that the unknown parameters of the vibro - pile driving model are generally parameters such as the dynamic stress - strain curve relationship or the relationship curve between parameters. In this embodiment, they are mainly soil parameters.

[0057] Specifically, combining Figure 2 , the standard Voigt model is adopted, that is, an independent spring is in series with a Voigt body, and the pile - end resistance is replaced by a spring and a damper. That is, the unknown parameters in this embodiment are the spring constant and viscosity coefficient of the soil, and the spring constant and viscosity coefficient of the soil are defined as variable parameters in the deep - learning framework.

[0058] In an embodiment of the present invention, a dynamic partial differential equation of the vibro - pile driving pile foundation is constructed, including:

[0059] Based on the exciting force of the vibratory hammer, the gravity of the pile foundation, the frictional force on the side of the pile foundation by the soil, and the resistance at the end, a dynamic partial differential equation of the vibro - pile driving pile foundation is constructed.

[0060] It can be understood that the dynamic model of vibro - pile driving generally uses a longitudinal vibration partial differential equation. During the vibro - pile driving process, considering the combined action of the exciting force of the vibratory hammer, the gravity of the pile foundation, the frictional force on the side of the pile foundation by the soil, and the resistance at the end, the pile foundation is gradually sunk into the soil. Therefore, the above - mentioned various acting forces are all included in the longitudinal vibration equation of vibro - pile driving.

[0061] Specifically, in this embodiment, some assumptions are made to simplify the model. First, the stiffness of the pile foundation is much greater than the stiffness of the soil, so it is assumed that the pile foundation is an absolute rigid body. Second, the soil is regarded as an elastoplastic body, and only the viscous damping in which the resistance is linearly related to the moving speed is considered. Finally, the situation where the soil vibrates with the pile foundation and causes the mass of the entire vibration system to increase is not considered.

[0062] Based on the above assumptions, the dynamic model of vibro - pile driving adopts the partial differential equation of longitudinal vibration as follows:

[0063]

[0064] This embodiment mainly considers the action of the frictional force on the side and the resistance at the end of the pile foundation by the soil, which is represented by f(x, t) in the above formula. The specific derivation process is as follows:

[0065] Combined with Figure 2 , assume that the displacement of the k - th section of the pile body at a certain moment is u, and the acting force is f k (x, t). Then the acting force of the soil on the pile side is simulated by an independent spring and a Voigt body. u k1 and u k2 represent the displacements of the independent spring and the Voigt body respectively. k1, k2 and η2 represent the spring constant per unit area and the viscosity coefficient of the independent spring and the Voigt body respectively, and satisfy the following relationship:

[0066] u k (x, t) = u k1 (x, t)+u k2 (x, t),

[0067]

[0068] Therefore, the frictional resistance suffered by the k - th section of the pile foundation can be obtained as: F k (x, t)=2πRh k f(x, t).

[0069] In an embodiment of the present invention, the vibro - pile driving data includes the data measured for the partial vibration of the pile foundation above the soil;

[0070] The vibro - pile driving data includes any one of the displacement, velocity and acceleration in the axial direction of the pile foundation.

[0071] It can be understood that during the vibro - pile driving process, part of the pile foundation is inserted into the soil and part is still above the soil. For the convenience of measurement, generally, the dynamic response of the pile foundation above the soil is selected for measurement. A small number of measuring points are selected along the axial direction of the pile foundation to measure the dynamic response of the pile foundation, including but not limited to physical quantities such as displacement, velocity and acceleration. After obtaining the real - time measured vibro - pile driving dynamic response data, it is imported into the deep - learning framework as the driving data for the deep - learning parameter inversion.

[0072] Specifically, combined with Figure 2, in this embodiment, considering the feasibility and convenience of measurement, the vibration of the part of the pile foundation above the soil mass is measured, and 4 measuring points are arranged along the axial direction of the pile foundation. The physical quantities measured for the vibration data can be the displacement, velocity or acceleration in the axial direction of the pile foundation. Those skilled in the art can easily understand the relationship among the displacement, velocity and acceleration of the pile foundation, that is, by taking the first and second derivatives of the displacement signal with respect to time, the vibration velocity and acceleration can be obtained respectively. Conversely, by performing the first and second integrations on the vibration acceleration signal, the velocity and displacement signals can be obtained respectively. The data measured in this embodiment is the axial velocity of the pile foundation. After obtaining the time-domain velocity time-domain response data of the pile foundation measured in real time, a single integration is performed to obtain the axial displacement, which is then imported into the deep learning framework as the driving data for the inversion of deep learning parameters.

[0073] In one embodiment of the present invention, the constructed deep learning framework including the embedding of physical knowledge is formulated based on the computing conditions and performance possessed by the actual engineering applications;

[0074] The parameters that need to be formulated for the neural network corresponding to the deep learning framework include:

[0075] Input and output quantities, number of neurons, number of neuron layers, activation function, initial state, optimizer and learning rate.

[0076] Based on the constructed deep learning framework including the embedding of physical knowledge and the driving data, a deep learning process for dynamically inverting the dynamic parameters of the vibrating pile driving is carried out, and the parameter values obtained through deep learning inversion are output, including:

[0077] Based on the axial coordinates of the pile foundation as the input of the neural network and the axial displacement of the pile foundation as the output of the neural network, by continuously iterating and calculating, the loss function including the dynamic partial differential equation and the measured vibrating pile driving data is minimized, and the dynamic results and inversion parameters of the vibrating pile driving are output.

[0078] It can be understood that on the basis of steps S101 to S103, a deep learning neural network is built, including input and output quantities, number of neurons, number of neuron layers, activation function, initial state, etc. At the same time, hyperparameters such as the optimizer and learning rate to be used should be determined.

[0079] Specifically, in actual engineering applications, the neural network architecture that meets the computing requirements should be formulated according to the available computing conditions and performance. Generally speaking, with dozens to hundreds of neurons per layer, three to five neuron layers, activation functions of sine function or tangent function, Adam optimizer and a learning rate of 0.001, relatively satisfactory inversion results can generally be obtained. In this embodiment, please refer to Figure 3 , Figure 3It is a diagram of a deep learning framework for dynamic inversion of vibration pile driving dynamic parameters provided by an embodiment of the present invention. Among them, the input of the neural network is the axial coordinate of the pile foundation, and the output is the axial displacement of the pile foundation; the neural network adopts a fully connected neural network, and the neural network has a three-layer network structure with 350 neurons in each layer; based on the characteristics of vibration pile driving, a sine function is selected as the activation function to achieve non-linearity; the Adam optimizer is used for stochastic gradient descent, and the learning rate is set to 0.001.

[0080] After establishing the neural network, the loss function including partial differential equations and measured data is minimized through continuous iterative calculations, so as to obtain accurate vibration pile driving dynamic results and inversion parameters.

[0081] Furthermore, by setting the period of obtaining measured data and inputting it into the deep learning framework, dynamic solution of the inversion parameters can be obtained.

[0082] In order to better implement the deep learning method for dynamic inversion of vibration pile driving dynamic parameters in the embodiments of the present invention, based on the deep learning method for dynamic inversion of vibration pile driving dynamic parameters, correspondingly, please refer to Figure 4 , Figure 4 It is a schematic structural diagram of an embodiment of a deep learning device for dynamic inversion of vibration pile driving dynamic parameters provided by the present invention. The embodiments of the present invention provide a deep learning device 400 for dynamic inversion of vibration pile driving dynamic parameters, including:

[0083] A parameter definition module 401, configured to determine known parameters and unknown parameters of the vibration pile driving model, define the known parameters as constant parameters in the deep learning framework, and define the unknown parameters as variable parameters in the deep learning framework;

[0084] A physical knowledge definition module 402, configured to construct a dynamic partial differential equation of the vibration pile driving pile foundation, and define the dynamic partial differential equation as physical knowledge in the deep learning framework;

[0085] A data acquisition module 403, configured to acquire vibration pile driving data measured in real time, and use the vibration pile driving data as driving data for parameter inversion;

[0086] A learning module 404, configured to perform a deep learning process for dynamic inversion of vibration pile driving dynamic parameters based on the constructed deep learning framework including physical knowledge embedding and the driving data, and output parameter values obtained through deep learning inversion.

[0087] It should be noted here that: the device 400 provided in the above embodiments can implement the technical solutions described in the above method embodiments. The specific implementation principles of the above modules or units can refer to the corresponding content in the above method embodiments, and will not be elaborated here.

[0088] Based on the deep learning method for dynamic inversion of the above vibration pile driving dynamic parameters, an embodiment of the present invention further provides an electronic device, including: a processor, a memory, and a computer program stored in the memory and executable on the processor; when the processor executes the computer program, it implements the steps in the deep learning method for dynamic inversion of the vibration pile driving dynamic parameters in the above respective embodiments.

[0089] Figure 5 The structural schematic diagram of the electronic device 500 suitable for implementing the embodiments of the present invention is shown. The electronic device in the embodiments of the present invention may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Tablet Computers), PMPs (Portable Multimedia Players), in-vehicle terminals (such as in-vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The shown electronic device is only an example and should not bring any limitation to the functions and usage scope of the embodiments of the present invention.

[0090] The electronic device includes: a memory and a processor, where the processor here may be referred to as the processing device 501 below, and the memory may include at least one of the read-only memory (ROM) 502, random access memory (RAM) 503, and storage device 508 below, as specifically shown below:

[0091] As Figure 5 shown, the electronic device 500 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 501, which may perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 502 or the program loaded from the storage device 508 into the random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the electronic device 500 are also stored. The processing device 501, ROM 502, and RAM 503 are connected to each other through a bus 504. The input / output (I / O) interface 505 is also connected to the bus 504.

[0092] Generally, the following devices may be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 508 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 509. The communication device 509 may allow the electronic device 500 to communicate with other devices wirelessly or wiredly to exchange data. Although Figure 5An electronic device 500 is shown with various devices, but it should be understood that it is not required to implement or have all the shown devices. Instead, more or fewer devices may be implemented or had.

[0093] In particular, according to an embodiment of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, an embodiment of the present invention includes a computer program product that includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program codes for performing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by a processing device 501, the above-mentioned functions defined in the methods of the embodiments of the present invention are executed.

[0094] Based on the deep learning method for dynamic inversion of the above-mentioned vibration pile driving dynamic parameters, an embodiment of the present invention also correspondingly provides a computer-readable storage medium that stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps in the deep learning method for dynamic inversion of the vibration pile driving dynamic parameters in the above various embodiments.

[0095] Those skilled in the art can understand that all or part of the processes for implementing the methods of the above embodiments can be completed by instructing relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. Among them, the computer-readable storage medium is a disk, an optical disc, a read-only memory or a random access memory, etc.

[0096] As described above, only the preferred specific embodiments of the present invention are given, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A deep learning method for dynamic inversion of dynamic parameters of vibro - pile driving, characterized in that, Including: Determine the known parameters and unknown parameters of the vibro - pile driving model, define the known parameters as constant parameters in the deep - learning framework, and define the unknown parameters as variable parameters in the deep - learning framework; Based on the exciting force of the vibratory hammer, the gravity of the pile foundation, the frictional force on the side of the pile foundation by the soil and the resistance at the end, construct the dynamic partial differential equation of the vibro - pile driving pile foundation, and define the dynamic partial differential equation as physical knowledge in the deep - learning framework; Obtain the real - time measured vibro - pile driving data, and use the vibro - pile driving data as the driving data for parameter inversion; Based on the constructed deep - learning framework including the embedding of physical knowledge and the driving data, perform the deep - learning process of dynamic inversion of the vibro - pile driving dynamic parameters, and output the parameter values obtained by deep - learning inversion.

2. The deep learning method for dynamic inversion of vibration pile driving dynamic parameters according to claim 1, characterized in that The known parameters of the vibro - pile driving model include the geometric parameters of the pile foundation, the material parameters of the pile foundation, the soil parameters and the vibratory hammer parameters; Among them, the geometric parameters of the pile foundation include the length, radius and thickness of the pile foundation; The material parameters of the pile foundation include the elastic modulus, Poisson's ratio and density of the pile foundation; The soil parameters include the density parameter of the soil; The vibratory hammer parameters include the exciting force, penetration force, penetration frequency, theoretical power and amplitude of the vibratory hammer.

3. The deep learning method for dynamic inversion of vibration pile driving dynamic parameters according to claim 1, characterized in that, The unknown parameter of the vibro - pile driving model includes the dynamic stress - strain curve relationship.

4. The deep learning method for dynamic inversion of vibration pile driving dynamic parameters according to claim 1, wherein, The vibro - pile driving data includes the data measured for the partial vibration of the pile foundation above the soil; The vibro - pile driving data includes any one of the axial displacement, velocity and acceleration of the pile foundation.

5. The deep - learning method for dynamic inversion of vibro - pile driving dynamic parameters according to claim 1, characterized in that: The constructed deep - learning framework including the embedding of physical knowledge is formulated based on the computing conditions and performance required for engineering practical applications; The parameters that need to be formulated for the neural network corresponding to the deep - learning framework include: Input quantity and output quantity, number of neurons, number of neuron layers, activation function, initial state, optimizer and learning rate.

6. The deep learning method for dynamic inversion of vibration pile driving dynamic parameters according to claim 5, characterized in that, The deep - learning process of dynamic inversion of vibro - pile driving dynamic parameters based on the constructed deep - learning framework including the embedding of physical knowledge and the driving data, and outputting the parameter values obtained by deep - learning inversion includes: Based on the axial coordinate of the pile foundation as the input of the neural network and the axial displacement of the pile foundation as the output of the neural network, through continuous iterative calculation, minimize the loss function including the dynamic partial differential equation and the measured vibro - pile driving data, and output the dynamic results of vibro - pile driving and the inverted parameters.

7. A deep learning device for dynamic inversion of dynamic parameters of vibro-sinking piles, characterized in that, Including: A parameter definition module, used to determine the known parameters and unknown parameters of the vibro - pile driving model, define the known parameters as constant parameters in the deep - learning framework, and define the unknown parameters as variable parameters in the deep - learning framework; A physical knowledge definition module, used to construct the dynamic partial differential equation of the vibro - pile driving pile foundation based on the exciting force of the vibratory hammer, the gravity of the pile foundation, the frictional force on the side of the pile foundation by the soil and the resistance at the end, and define the dynamic partial differential equation as physical knowledge in the deep - learning framework; A data acquisition module, configured to acquire vibration pile driving data measured in real time and use the vibration pile driving data as driving data for parameter inversion. A learning module, configured to perform a deep learning process for dynamic inversion of vibration pile driving dynamics parameters based on a constructed deep learning framework including physical knowledge embedding and the driving data, and output parameter values obtained through deep learning inversion.

8. An electronic device, characterized in that, It includes a memory and a processor. Among them, the memory is used to store programs; the processor is coupled to the memory and is configured to execute the programs stored in the memory to implement the steps in the deep learning method for dynamic inversion of vibration pile driving dynamics parameters described in any one of claims 1 to 6 above.

9. A computer-readable storage medium, characterized in that, It is used to store computer-readable programs or instructions, and when the programs or instructions are executed by a processor, the steps in the deep learning method for dynamic inversion of vibration pile driving dynamics parameters described in any one of claims 1 to 6 above can be implemented.

Images