A two-dimensional consolidation coefficient determination device and method based on physical information neural network

Abstract

This invention provides a two-dimensional consolidation coefficient determination device and method based on a physical information neural network, solving the problems of current methods that cannot simultaneously detect consolidation coefficients in both vertical and horizontal two-dimensional directions, and whose detection process is cumbersome and requires improvement in accuracy. The device includes an outer frame, within which a measuring cylinder is fixed. The inner cavity of the measuring cylinder serves as a pressing station. Both the upper and lower openings of the measuring cylinder are sealed with permeable stone covers. Several pore water pressure gauges are installed on the side walls of the measuring cylinder. Hydraulic pumps are pressed against the outer sides of both the upper and lower permeable stone covers. The hydraulic pumps are electrically connected to a hydraulic control system. A control panel is mounted on the outer wall of the outer frame, and the hydraulic control system is connected to the control panel via a circuit. This invention selects two-dimensional consolidation theory as the basis for determining the consolidation coefficient, avoiding the use of empirical formulas. While ensuring calculation accuracy, it simplifies the steps for obtaining the soil consolidation coefficient to the greatest extent possible, and also has good universality.

Description

Technical Field

[0001] This invention belongs to the field of measurement technology and relates to a soil property detection method, particularly a two-dimensional consolidation coefficient determination device and method based on a physical information neural network. Background Technology

[0002] Currently, there are four main methods for obtaining the consolidation coefficient: the logarithmic time method, the square root time method, the spiral plate load test method, and the pore pressure static cone penetration test method. The logarithmic time method and the square root time method are both laboratory testing methods. Their disadvantages lie in the significant impact of initial compression in the early stages of the test and secondary consolidation in the later stages on the test results, as well as the considerable influence of human factors. The spiral plate load test method and the pore pressure static cone penetration test method are both field testing methods. Both methods have significant limitations in terms of monitoring cycle and cost.

[0003] Current methods for determining the consolidation coefficient are mostly limited to measuring the vertical consolidation coefficient, neglecting the anisotropic characteristics of clay strata. Since most clay minerals have a platy structure, clay particles tend to align horizontally during deposition and consolidation. Therefore, clay strata encountered in engineering often exhibit anisotropic consolidation characteristics. Consequently, in practical engineering, the horizontal and vertical consolidation coefficients of soil are not the same, necessitating methods and instruments capable of simultaneously measuring both horizontal and vertical consolidation coefficients.

[0004] In addition, a Chinese patent discloses a method for determining the horizontal consolidation coefficient of saturated soft clay using a flat shovel lateral expansion C-value dissipation test, patent application number: CN201710487874.8. This invention discloses a method for determining the horizontal consolidation coefficient of saturated soft clay using a flat shovel lateral expansion C-value dissipation test, including the following steps: 1) conducting a flat shovel lateral expansion C-value dissipation test and collecting several flat shovel lateral expansion Ct values ​​within the test time, where the subscript t represents time; 2) plotting the dissipation curve; 3) determining the initial state value C0 and the stable state value C100; 4) determining the time t50-C when the flat shovel lateral expansion C-value dissipates by 50%; 5) calculating the horizontal consolidation coefficient Ch. This invention solves the problem of directly testing the horizontal consolidation coefficient of saturated soft clay using a flat shovel lateral expansion test without increasing testing techniques, expands the application function of the flat shovel lateral expansion test, saves exploration costs, improves efficiency, and has significant effects.

[0005] The above technical solution is only used to detect the horizontal consolidation coefficient. It cannot achieve the joint detection of the consolidation coefficient in both the vertical and horizontal two-dimensional directions. Moreover, the detection process is relatively cumbersome, and its measurement accuracy still needs to be improved. Summary of the Invention

[0006] The purpose of this invention is to address the aforementioned problems in existing technologies by proposing a device and method for determining the two-dimensional consolidation coefficient based on a physical information neural network. This method involves automatically adjusting the hydraulic pressure to gradually change the pressure at both ends of a saturated soil sample, continuously monitoring the changes in pore water pressure, and expressing the physical residual term in a loss function using physical information. This enables accurate prediction based on small sample data.

[0007] The objective of this invention can be achieved through the following technical solution: a two-dimensional consolidation coefficient measuring device based on a physical information neural network, comprising an outer frame, a measuring cylinder fixed inside the outer frame, the inner cavity of the measuring cylinder being a pressing station, permeable stone covers sealing the upper and lower openings of the measuring cylinder, several pore water pressure gauges being installed on the side wall of the measuring cylinder, hydraulic pumps being pressed on the outer sides of the upper and lower permeable stone covers, the hydraulic pumps being electrically connected to a hydraulic control system, a control panel being installed on the outer wall of the outer frame, and the hydraulic control system being connected to the control panel via a circuit.

[0008] In the aforementioned two-dimensional consolidation coefficient measuring device based on physical information neural network, the hydraulic control system is connected to the hydraulic cylinder via a high-pressure oil pipe, and the hydraulic cylinder is connected to the hydraulic pump via a hydraulic pipe.

[0009] In the above-mentioned two-dimensional consolidation coefficient measuring device based on physical information neural network, a drain outlet is opened at the bottom of the outer frame, and the drain outlet is located below several of the pore water pressure gauges.

[0010] In the above-mentioned two-dimensional consolidation coefficient measuring device based on physical information neural network, the two hydraulic pumps are arranged symmetrically at the top and bottom, and a pressure plate is fixed at the outer end of the telescopic rod of the hydraulic pump, and the pressure plate presses against the outer wall surface of the permeable stone cover.

[0011] In the above-mentioned two-dimensional consolidation coefficient measuring device based on physical information neural network, the hydraulic pump is fixed on the top or bottom wall of the outer frame by a support frame, with the hydraulic pump on the top wall pressing downward and the hydraulic pump on the bottom wall pressing upward.

[0012] In the above-mentioned two-dimensional consolidation coefficient measuring device based on physical information neural network, the measuring cylinder includes two vertically arranged steel plates and two permeable stone slabs, which are alternately connected and spliced ​​to form a square cylinder structure.

[0013] In the above-mentioned two-dimensional consolidation coefficient measuring device based on physical information neural network, a number of support rods are fixed on the inner wall of the outer frame, and the inner circumference of the support rods is fixed to the steel plate; a number of pore water pressure gauges are arranged on the outer wall of at least one steel plate.

[0014] The method for measuring the two-dimensional consolidation coefficient using a physical information neural network-based device includes the following steps:

[0015] 1) After taking soil samples at the construction site, place them in the compaction station of the measuring cylinder and apply constant pressure using different levels of pressure to obtain N. m Pore ​​water pressure u varies with time and space i m ;

[0016] 2) The total loss function of the designed physical information neural network is as follows:

[0017]

[0018] In the above formula, u(x) i y i , t i ) represents the predicted value of the neural network; N c The number of coordination sites in the neural network, (x i y i , t i ) represents the coordinates of the selected coordination site; θ is the weight value of the neural network; c x c is the consolidation coefficient in the horizontal direction; y is the consolidation coefficient in the vertical direction; f(x, y, t) represents the physical constraints associated with the two-dimensional consolidation equation;

[0019] 3) Introduce the above physical equations into the loss function of the physical information neural network. After setting the boundary conditions, use the observed pore water pressure that changes with time as training data to train the physical information neural network until the set loss function reaches its minimum value. Finally, obtain the consolidation coefficients in the horizontal and vertical directions through inversion.

[0020] In the above-described method for measuring the two-dimensional consolidation coefficient using a physical information neural network-based device, the characteristic feature is that, in step 2), N c The value can be any natural number.

[0021] In the above-mentioned method for measuring the two-dimensional consolidation coefficient using a physical information neural network-based device, the characteristic feature is that, in step 2), the equation for f(x, y, t) is as follows:

[0022]

[0023] Compared with existing technologies, the two-dimensional consolidation coefficient determination device and method based on physical information neural networks have the following advantages:

[0024] The advantage of this invention lies in simplifying the traditional steps for determining the consolidation coefficient. It eliminates the need for repeated tests using a consolidation apparatus; continuous pressure application to the sample can be achieved simply by setting the pressure mode of the pressurization device, thus reducing the influence of human factors to some extent. For determining the consolidation coefficient, based on fundamental theory, two-dimensional consolidation theory is selected as the basis for the inversion of the consolidation coefficient. Empirical formulas are not used for solution, thus simplifying the steps for obtaining the soil consolidation coefficient to the greatest extent while ensuring calculation accuracy, and still maintaining good universality. It combines the advantages of traditional indoor tests—low cost, high precision, and short cycle—with the realism and convenience of outdoor tests. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the structure of a two-dimensional consolidation coefficient measuring device based on a physical information neural network.

[0026] Figure 2 This is a schematic diagram illustrating the principle of physical-driven deep learning in the measurement method of this two-dimensional consolidation coefficient measuring device based on physical information neural network.

[0027] In the diagram, 1. Outer frame; 1a. Drainage outlet; 2. Support rod; 3. Steel plate; 4. Permeable stone slab; 5. Permeable stone cover; 6. Pore water pressure gauge; 7. Hydraulic pump; 8. Pressure plate; 9. Support frame; 10. Hydraulic cylinder; 11. High-pressure oil pipe; 12. Hydraulic control system; 13. Control panel. Detailed Implementation

[0028] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0029] like Figure 1 As shown, this two-dimensional consolidation coefficient measuring device based on physical information neural network includes an outer frame 1, a measuring cylinder fixed inside the outer frame 1, the inner cavity of the measuring cylinder is a pressing station, the upper and lower openings of the measuring cylinder are sealed with permeable stone covers 5, several pore water pressure gauges 6 are installed on the side wall of the measuring cylinder, and hydraulic pumps 7 are pressed on the outer side of the upper and lower permeable stone covers 5. The hydraulic pumps 7 are electrically connected by a hydraulic control system 12. A control panel 13 is installed on the outer wall of the outer frame 1, and the hydraulic control system 12 is connected to the control panel 13 through a circuit.

[0030] The hydraulic control system 12 is connected to the hydraulic cylinder 10 via a high-pressure oil pipe 11, and the hydraulic cylinder 10 is connected to the hydraulic pump 7 via a hydraulic pipe.

[0031] A drain outlet 1a is provided at the bottom of the outer frame 1, and the drain outlet 1a is located below several pore water pressure gauges 6.

[0032] Two hydraulic pumps 7 are arranged symmetrically, one above the other. A pressure plate 8 is fixed to the outer end of the telescopic rod of the hydraulic pump 7, and the pressure plate 8 presses against the outer wall of the permeable stone cover 5.

[0033] The hydraulic pump 7 is fixed to the top or bottom wall of the outer frame 1 by the support frame 9. The hydraulic pump 7 on the top wall presses downward and the hydraulic pump 7 on the bottom wall presses upward.

[0034] The measuring cylinder includes two vertically arranged steel plates 3 and two permeable stone slabs 4, which are alternately connected and spliced ​​to form a square cylinder structure.

[0035] Several support rods 2 are fixed to the inner wall of the outer frame 1, and steel plates 3 are fixed to the inner circumference of the support rods 2; several pore water pressure gauges 6 are set on the outer wall of at least one steel plate 3.

[0036] In summary, this measuring device mainly consists of four parts: a frame system, a hydraulic system, a detection system, and a control system. The specific structural composition of each part is as follows:

[0037] 1. Framework System

[0038] The frame system primarily serves to protect and support the entire internal structure. It consists of an outer frame 1, an internal support structure, and steel plates 3. The support structure includes two types: support frames 9 and support rods 2. Support frames 9 mainly maintain the stability of the hydraulic pump 7 during hydraulic system operation. Support rods 2 maintain the stability of the entire soil mass during sample pressurization, ensuring stable and reliable pore water pressure data. Steel plates 3 are only installed on the left and right sides; they are not installed on the top, bottom, front, or back sides.

[0039] 2. Hydraulic system

[0040] The hydraulic system mainly consists of two hydraulic pumps 7 (set at the top and bottom), a hydraulic cylinder 10, a high-pressure oil pipe 11, and a hydraulic control system 12. The hydraulic pressure can be directly controlled from the control panel 13, or it can be automatically pressurized according to a preset pressure to apply continuous pressure to the upper and lower surfaces of the saturated soil sample.

[0041] 3. Detection System

[0042] The detection system mainly consists of steel plates 3 on the left and right sides, permeable stone covers 5 on the top and bottom sides, permeable stone slabs 4 on the front and back sides, and a pore water pressure gauge 6. Since plane strain problems are frequently encountered in actual engineering, seepage generally occurs in both vertical and horizontal directions. Therefore, only the top, bottom, front, and back sides are designed to accommodate permeable stones for drainage, while the left and right sides are fitted with non-draining steel plates 3 to simulate two-dimensional soil consolidation, which meets the needs of most engineering projects for determining the consolidation coefficient. This invention controls the seepage direction of the soil by adjusting the placement of the permeable stones. During the experiment, a saturated soil sample is placed in the pressing position of the measuring cylinder. Then, permeable stone covers 5 are placed on the top and bottom sides of the measuring cylinder, and the hydraulic system is activated for continuous pressurization. The pore water pressure gauge 6 records the change in pore pressure of the sample over time and inputs the data into a computer.

[0043] 4. Control System

[0044] The main functions of the control system are integrated into the operation panel, including the control of the hydraulic system and the monitoring and adjustment of the pore water pressure gauge 6, allowing the operator to easily set various parameters. The control system also has an output interface, which can input the measured pore water pressure data into a computer for subsequent consolidation coefficient calculation.

[0045] The working process and principle of this measuring device are as follows:

[0046] After the soil sample is placed in the chamber of the measuring cylinder, the sample is automatically pressurized, and the change of pore water pressure at different locations over time is recorded. Then, the physical prior knowledge of the two-dimensional consolidation partial differential equation is embedded into a deep learning neural network, which is coupled with the data on the change of pore water pressure at different locations over time measured by the experimental device. Finally, a physical information neural network architecture is designed to establish a method for analyzing and identifying the horizontal and vertical consolidation coefficients of the soil sample based on experimental data and physical knowledge.

[0047] The method for measuring the two-dimensional consolidation coefficient using a physical information neural network-based device includes the following steps:

[0048] 1) After taking soil samples at the construction site, place them in the compaction station of the measuring cylinder and apply constant pressure using different levels of pressure to obtain N. m Pore ​​water pressure u varies with time and space i m ;

[0049] 2) The total loss function of the designed physical information neural network is as follows:

[0050]

[0051] In the above formula, u(x) i y i, t i ) represents the predicted value of the neural network; N c The number of coordination sites in the neural network, (x i y i , t i ) represents the coordinates of the selected coordination site; θ is the weight value of the neural network; c x c is the consolidation coefficient in the horizontal direction; y is the consolidation coefficient in the vertical direction; f(x, y, t) represents the physical constraints associated with the two-dimensional consolidation equation;

[0052] In step 2), N c The value can be any natural number.

[0053] In step 2), the equation for f(x, y, t) is as follows:

[0054]

[0055] 3) Introduce the above physical equations into the loss function of the physical information neural network. After setting the boundary conditions, use the observed pore water pressure that changes with time as training data to train the physical information neural network until the set loss function reaches its minimum value. Finally, obtain the consolidation coefficients in the horizontal and vertical directions through inversion.

[0056] This measurement method utilizes the PINN algorithm to incorporate two-dimensional consolidation theory into the loss function, thereby constraining the training of the neural network.

[0057] Data-physics-driven deep learning methods couple the physical mechanisms required to solve a research problem into a deep neural network framework. Physical information is expressed through the physical residual term in the loss function, which penalizes solutions that do not meet the corresponding physical conditions. This constrains the neural network to be trained in a space that satisfies physical relationships, thereby achieving accurate predictions based on small sample data.

[0058] Compared with existing technologies, the two-dimensional consolidation coefficient determination device and method based on physical information neural networks have the following advantages:

[0059] The advantage of this invention lies in simplifying the traditional steps for determining the consolidation coefficient. It eliminates the need for repeated tests using a consolidation apparatus; continuous pressure application to the sample can be achieved simply by setting the pressure mode of the pressurization device, thus reducing the influence of human factors to some extent. For determining the consolidation coefficient, based on fundamental theory, two-dimensional consolidation theory is selected as the basis for the inversion of the consolidation coefficient. Empirical formulas are not used for solution, thus simplifying the steps for obtaining the soil consolidation coefficient to the greatest extent while ensuring calculation accuracy, and still maintaining good universality. It combines the advantages of traditional indoor tests—low cost, high precision, and short cycle—with the realism and convenience of outdoor tests.

[0060] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

[0061] Although this document frequently uses terms such as outer frame 1; drain outlet 1a; support rod 2; steel plate 3; permeable stone slab 4; permeable stone cover 5; pore water pressure gauge 6; hydraulic pump 7; pressure plate 8; support frame 9; hydraulic cylinder 10; high-pressure oil pipe 11; hydraulic control system 12; and control panel 13, the possibility of using other terms is not excluded. The use of these terms is merely for the convenience of describing and explaining the essence of the invention; interpreting them as any additional limitation would contradict the spirit of the invention.

[0062] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

Claims

1. A method for measuring a two-dimensional consolidation coefficient measurement device based on a physical information neural network, the two-dimensional consolidation coefficient measurement device based on a physical information neural network comprising an outer frame, a measuring cylinder is fixedly arranged in the outer frame, the inner cavity of the measuring cylinder is a material pressing station, the upper and lower openings of the measuring cylinder are blocked by water permeable stone covers, a plurality of pore water pressure gauges are arranged on the side wall of the measuring cylinder, hydraulic pumps are arranged on the outer sides of the upper and lower water permeable stone covers, the hydraulic pumps are electrically connected by a hydraulic control system, a control panel is arranged on the outer wall of the outer frame, and the hydraulic control system is connected to the control panel through a circuit; The determination method of the two-dimensional consolidation coefficient determination device based on the physical information neural network is characterized in that, The method comprises the following steps: 1), After taking good soil sample in engineering site, put into the pressure loading position of measuring cylinder, adopt different grade pressure to carry out constant pressure loading, obtain Pore water pressure changing with time and space ; 2.The total loss function of the designed physical information neural network is as follows: In the above formula, represents the predicted value of the neural network; is the number of coordination sites of the neural network, represents the selected coordination site coordinates; is the weight value of the neural network; is the consolidation coefficient in the horizontal direction; is the consolidation coefficient in the vertical direction; represents the physical constraints related to the two-dimensional consolidation equation; 3.The physical equation is introduced into the loss function of the physical information neural network, the observed pore water pressure changing with time is used as training data to train the physical information neural network after the boundary conditions are set, the loss function is minimized until the loss function is minimized, and finally the horizontal and vertical consolidation coefficients are obtained through inversion.

2. The measurement method of the two-dimensional consolidation coefficient measurement apparatus based on a physical information neural network according to claim 1, characterized by, The hydraulic control system is connected to the hydraulic cylinder through a high-pressure oil pipe, and the hydraulic cylinder is connected to the hydraulic pump through a hydraulic pipe.

3. The measurement method of the two-dimensional consolidation coefficient measurement apparatus based on a physical information neural network according to claim 1, characterized by, A drainage port is arranged at the bottom of the outer frame, and the drainage port is located below the plurality of pore water pressure gauges.

4. The measurement method of the two-dimensional consolidation coefficient measurement apparatus based on a physical information neural network according to claim 1, characterized by, The two hydraulic pumps are arranged symmetrically above and below, the outer end of the telescopic rod of the hydraulic pump is fixedly provided with a pressing plate, and the pressing plate presses the outer wall surface of the water permeable stone cover.

5. The measurement method of the two-dimensional consolidation coefficient measurement apparatus based on a physical information neural network according to claim 4, characterized by, The hydraulic pump is fixedly arranged on the top wall or the bottom wall of the outer frame through a support frame, the hydraulic pump on the top wall presses downward, and the hydraulic pump on the bottom wall presses upward.

6. The measurement method of the two-dimensional consolidation coefficient measurement apparatus based on a physical information neural network according to claim 1, wherein The measuring cylinder comprises two vertically arranged steel plates and two water permeable stone plates, and the two steel plates and the two water permeable stone plates are alternately connected and spliced to form a square cylinder structure.

7. The measurement method of the two-dimensional consolidation coefficient measurement apparatus based on a physical information neural network according to claim 6, characterized by, A plurality of support rods are fixedly arranged on the inner wall of the outer frame, the inner periphery of the plurality of support rods is connected to the steel plates, and the plurality of pore water pressure gauges are arranged on the outer wall of at least one steel plate.

8. The measurement method of the physical information neural network-based two-dimensional consolidation coefficient measurement apparatus according to claim 1, wherein, In step 2), The value of n is any natural number.

9. The measurement method of the two-dimensional consolidation coefficient measurement apparatus based on a physical information neural network according to claim 1, characterized by, In step 2), The equation is as follows: 。

Images