A device for long-term viscosity prediction of cement-based grouting material

Abstract

This invention pertains to the field of grouting material performance research technology and provides a device for predicting the long-term viscosity of cement-based grouting materials. The technical solution includes a viscosity testing device and a viscosity prediction device connected to it. The viscosity testing device comprises an admixture dispensing device, a grouting material preparation device, a cement single-liquid slurry storage device, a slurry holding device, and a viscosity meter. The admixture dispensing device is located at the upper end of the grouting material preparation device, and its two sides are connected to the cement single-liquid slurry storage device and the slurry holding device, respectively, which have openable and closable channels. The viscosity meter is located above the slurry holding device. Based on liquid viscosity data obtained under different admixture addition schemes and a constructed liquid viscosity prediction model, the target time viscosity prediction value under a specific admixture addition scheme is obtained. Based on the acquired reliable data, the long-term viscosity of the material is predicted.

Description

Technical Field

[0001] This invention belongs to the technical field of grouting material performance research, and in particular relates to a device for predicting the long-term viscosity of cement-based grouting materials. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] In the field of tunnel construction, sudden water inrush is one of the most severe challenges. Cement-based grouting materials, as commonly used for controlling tunnel water inrushes, directly affect the safety, progress, and cost control of tunnel construction. Viscosity is a key indicator of the fluidity of cement-based grouting materials, directly affecting the diffusion distance and rate of the grout in the injected medium during grouting operations. When sealing tunnel water inrush channels, appropriate viscosity ensures that cement-based grouting materials can effectively penetrate, fill, and consolidate in complex geological environments, such as different rock pore structures and groundwater pressure conditions, thereby forming a reliable water-stopping curtain and preventing further groundwater inrush. Ensuring that the viscosity of cement-based grouting materials is within a suitable range is crucial for the smooth implementation of grouting operations.

[0004] To achieve precise viscosity control, it is often necessary to optimize the formulation of cement-based grouting materials according to the specific needs of the engineering site. However, traditional viscosity tests of cement-based grouting materials under the influence of admixtures rely on manual operation by laboratory personnel, which is cumbersome and inefficient. The test process involves a large amount of manual reading, data recording, and preliminary data processing, and the inefficient manual measurement greatly limits the speed and scale of data acquisition. At the same time, because cement slurry is prone to precipitation and segregation, traditional viscosity testing equipment is difficult to obtain accurate long-term viscosity of cement slurry. Even when manual mixing is used during testing, obtaining long-term viscosity still consumes a lot of time, affecting the efficiency of grouting material research and development and property studies. Summary of the Invention

[0005] To address at least one of the technical problems mentioned in the background, this invention provides a long-term viscosity testing device and analysis method for cement-based grouting materials. This method solves the difficulty in obtaining accurate long-term viscosity values ​​during testing, providing an efficient and convenient intelligent method for studying the influence of admixtures on the viscosity properties of grouting materials. It achieves integrated operation of viscosity testing and analysis optimization for cement-based grouting materials, significantly improving the speed and accuracy of data acquisition, reducing human error, and enhancing the reliability of test results. Furthermore, based on a large amount of experimental data, an intelligent algorithm predicts the long-term viscosity of grouting materials under the influence of admixtures, providing scientific guidance for grouting operations and promoting the efficient research and practical application of cement-based grouting materials.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] The present invention provides a long-term viscosity prediction device for cement-based grouting materials, including a viscosity testing device and a long-term viscosity prediction device connected thereto.

[0008] The viscosity testing device includes an admixture dispensing device, a grouting material preparation device, a cement single-liquid slurry storage device, a slurry holding device, and a viscosity meter; the admixture dispensing device is installed at the upper end of the grouting material preparation device, and the two sides of the grouting material preparation device are respectively connected to the cement single-liquid slurry storage device and the slurry holding device with openable and closable channels; the viscosity meter is installed above the slurry holding device.

[0009] The viscosity tester is used to obtain the viscosity of the liquid in the slurry holding device obtained by adding different admixtures using the admixture dosing device;

[0010] The long-term viscosity prediction device is configured to predict the target time viscosity value under the admixture addition scheme based on the grouting material viscosity data obtained under different admixture addition schemes and the trained long-term viscosity prediction model.

[0011] Furthermore, the admixture dispensing device includes multiple admixture dispensing cylinders, each admixture dispensing cylinder including a feeding turntable and multiple storage tubes. The feeding turntable has a groove on its surface that fits into the bottom of the storage tube. An opening is provided at the bottom of the groove. The storage tube has an opening at the bottom and a screw cap is provided at the top.

[0012] Furthermore, the grouting material preparation device includes a grouting material preparation tank and a first slurry stirring device. The first slurry stirring device includes a first stirring rotating shaft, a first stirring blade, a first motor, and a speed-reducing transmission gear set. The first stirring rotating shaft and the first stirring blade are arranged inside the grouting material preparation tank. The first stirring rotating shaft is arranged at the bottom of the grouting material preparation tank, and multiple layers of first stirring blades of different sizes are distributed on it. One end of the speed-reducing transmission gear set is connected to the first stirring rotating shaft, and the other end is connected to the first motor.

[0013] Furthermore, the grouting material preparation device also includes a first mounting platform, and a porous cleaning nozzle is provided on the top of the grouting material preparation tank. The porous cleaning nozzle and the water pipe connected to it pass through the center of the first mounting platform.

[0014] Furthermore, the admixture dispensing cylinder comprises three cylinders arranged in an equilateral triangle on the first mounting platform. The three admixture dispensing cylinders are respectively filled with three types of water-reducing agents, three types of retarders, and three types of early-strength agents.

[0015] Furthermore, the cement single-liquid slurry storage device includes a cement single-liquid slurry storage tank, a second motor, and a screw cap. The cement single-liquid slurry storage tank is mounted on the second motor, and a screw cap is provided on the top of the cement single-liquid slurry storage tank. A second rotating stirring shaft and a second stirring blade are provided at the bottom of the cement single-liquid slurry storage tank. The second stirring blade is evenly distributed from bottom to top on the second rotating stirring shaft, and the second rotating stirring shaft is connected to the second motor.

[0016] Furthermore, the viscosity testing module includes a second mounting platform, a slurry holding device, a third motor, and a viscosity testing instrument. The slurry holding device is mounted on the second mounting platform, and the viscosity testing instrument includes a rotational viscometer, a connecting rod, and a rotating cylinder. The rotational viscometer is positioned directly above the slurry holding device, and the rotational viscometer and the rotating cylinder are connected by the connecting rod.

[0017] Furthermore, a first opening is provided on the side of the cement single-liquid slurry storage tank, and a second opening is provided on the side of the grouting material preparation container. The first opening and the second opening are connected by a pipe.

[0018] The cement single-liquid slurry storage tank also has a third opening on its side, and the slurry holding device has a fourth opening on its side. The third opening and the fourth opening are connected by a pipe.

[0019] Furthermore, the device includes a first electric ball valve and a second electric ball valve, wherein the first electric ball valve is disposed on a pipe connecting a first opening and a second opening, and the second electric ball valve is disposed on a pipe connecting a third opening and a fourth opening.

[0020] Furthermore, the training process for the long-term viscosity prediction model includes:

[0021] When the admixture addition scheme is less than or equal to n, a set of viscosity tests is performed for each scheme. The viscosity tests are grouped according to the effective time of the viscosity change data, and the short-term viscosity data of each group of tests is obtained.

[0022] When there are more than n admixture addition schemes, only the short-term viscosity data corresponding to the current m-th scheme is obtained. The short-term viscosity result corresponding to the m-th scheme and the historical viscosity prediction results corresponding to the previous m-1 admixture addition schemes are input into the long-term viscosity prediction model to obtain the long-term viscosity prediction value of the m-th scheme, where m is greater than n.

[0023] Compared with the prior art, the beneficial effects of the present invention are:

[0024] 1. The present invention proposes a long-term viscosity prediction device for cement-based grouting materials. Through highly automated operation, it acquires liquid viscosity data and, combined with different admixture addition schemes, can determine the target time viscosity prediction value under a specific admixture addition scheme. This solves the problem of obtaining accurate long-term viscosity values ​​during experimental testing and provides an efficient and convenient intelligent method for studying the influence of admixtures on the viscosity performance of grouting materials.

[0025] 2. This invention proposes a viscosity testing device for cement-based grouting materials, significantly improving the testing efficiency and reducing manual operation steps. This avoids tedious measurement and data processing work, greatly accelerating data acquisition and analysis. It effectively eliminates human error, ensuring the accuracy and high repeatability of each test. Automated control and precise data processing improve the reliability of measurement results and enhance their credibility.

[0026] 3. This invention combines a large amount of test data and experimental design schemes, and can predict the long-term viscosity of cement-based grouting materials by combining a small number of short-term viscosity test results. It establishes a dual-input channel LSTM model to simultaneously extract and process features from two different types of data, solving the problem of obtaining accurate long-term viscosity values ​​during experimental testing. This provides an efficient and convenient intelligent method for studying the influence of admixtures on the viscosity properties of grouting materials.

[0027] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0028] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0029] Figure 1 This is a schematic diagram of the viscosity testing device for cement-based grouting materials provided in an embodiment of the present invention;

[0030] Figure 2 This is a schematic diagram of the internal structure of the admixture dispensing device provided in an embodiment of the present invention;

[0031] Figure 3 This is a schematic diagram of the interior of the grouting material preparation tank and the speed reduction gear provided in an embodiment of the present invention;

[0032] Figure 4 This is a schematic diagram of the interior of a cement single-liquid slurry storage tank provided in an embodiment of the present invention;

[0033] Figure 5This is a schematic diagram of the viscosity testing module provided in an embodiment of the present invention;

[0034] Figure 6 This is a flowchart of the analysis and prediction module provided in an embodiment of the present invention;

[0035] The components include: 1. First mounting platform; 2. Admixture dispensing device; 201. Feeding turntable; 202. Storage pipe; 203. First cap; 3. Grouting material preparation device; 301. Grouting material preparation tank; 302. First slurry stirring device; 3021. First stirring rotating shaft; 3022. First stirring blade; 3023. First motor; 3024. Speed ​​reduction transmission gear set; 3025. Multi-hole cleaning nozzle; 4. Cement single-liquid slurry storage device; 401. Cement single-liquid slurry storage tank; 402. Second motor; 403. Second cap; 404. Second rotating stirring shaft; 405. Second stirring blade; 5. Slurry holding device; 6. Cleaning blade; 7. Second mounting platform; 8. Viscosity testing instrument; 801. Rotational viscometer; 802. Connecting rod; 803. Rotating cylinder; 9. Third motor; 10. First electric ball valve; 11. Second electric ball valve. Detailed Implementation

[0036] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0037] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0038] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0039] Traditional viscosity tests of cement-based grouting materials under the influence of admixtures rely on manual operation by laboratory personnel, resulting in a cumbersome and inefficient process. The tests involve extensive manual reading, data recording, and preliminary data processing, with inefficient manual measurements significantly limiting the speed and scale of data acquisition. Furthermore, due to the tendency of cement slurry to precipitate and segregate, traditional viscosity testing equipment struggles to obtain accurate long-term viscosity data. Even with manual mixing during testing, obtaining long-term viscosity still consumes considerable time, impacting the efficiency of grouting material research and development and property studies.

[0040] To address the aforementioned issues, this invention provides a long-term viscosity testing device and analysis method for cement-based grouting materials, comprising a material preparation module, a viscosity testing module, and an analysis and prediction module. This integrates viscosity testing and analysis optimization of cement-based grouting materials, significantly improving data acquisition speed and accuracy, reducing human error, and enhancing the reliability of test results. Furthermore, based on extensive experimental data, the analysis and prediction module uses intelligent algorithms to predict the long-term viscosity of the material, providing scientific guidance for grouting operations and promoting the efficient research and application of cement-based grouting materials.

[0041] Figure 1 This is a schematic diagram of a viscosity testing device for cement-based grouting materials, as shown below. Figure 1 As shown, this embodiment provides a long-term viscosity testing device for cement-based grouting materials, including an admixture dispensing device, a grouting material preparation device, a cement single-liquid slurry storage device, a slurry holding device, and a viscosity tester;

[0042] The material preparation module includes a first mounting platform 1, an admixture dispensing device, a grouting material preparation device 3, a cement single-liquid slurry storage device 4, and a slurry holding device 5. The admixture dispensing device 2 is set above the first mounting platform 1, and the grouting material preparation device 3 is set below it. The grouting material preparation device 3 is connected to the cement single-liquid slurry storage device 4 and the slurry holding device 5, respectively. The slurry holding device 5 is connected to a viscosity testing module.

[0043] like Figure 2 As shown, the admixture dispensing device includes multiple admixture dispensing cylinders 2, which are placed on the first mounting platform 1;

[0044] It is understood that the admixture dispensing cylinder includes three cylinders, and the three admixture dispensing cylinders 2 are arranged in an equilateral triangle on the first mounting platform 1. Other embodiments may also be configured according to actual conditions.

[0045] Each additive dispensing cylinder 2 includes a dispensing turntable 201 and multiple storage tubes 202. The storage tubes 202 have openings at the bottom and caps 203 at the top. Reagents can be added into the storage tubes 202 by opening the caps 203. The surface of the dispensing turntable has grooves. Preferably, the dispensing turntable 201 is a disc shape with an annular groove on its surface, which fits into the bottom of the storage tubes 202. The bottom of the groove has an opening for releasing the reagents in the storage tubes. The dispensing turntable 201 passes through the first mounting platform 1 and is fixed and connected to the first mounting platform 1 by bearings.

[0046] In this embodiment, the storage tubes of the three admixture dosing cylinders are respectively filled with three types of water-reducing agents, three types of retarders, and three types of early-strength agents.

[0047] like Figure 3As shown, the grouting material preparation device 3 includes a grouting material preparation tank 301 and a first slurry stirring device 302. The first slurry stirring device 302 includes a first stirring rotating shaft 3021, a first stirring blade 3022, a first motor 3023, and a speed reduction transmission gear set 3024.

[0048] The first stirring shaft 3021 and the first stirring blade 3022 are disposed inside the grouting material preparation tank 301. The first stirring shaft 3021 is disposed at the bottom of the grouting material preparation tank 301, and multiple layers of first stirring blades 302 of different sizes are distributed on it, for example, three layers. The first motor 3023 and the speed reduction transmission gear set 3024 are disposed below the grouting material preparation tank 301. One end of the speed reduction transmission gear set 3024 is connected to the first stirring shaft 3021, and the other end is connected to the first motor 3023. The first motor 3023 is connected to a controller and is used to drive the stirring shaft 3021.

[0049] The top of the grouting material preparation bucket 301 is provided with a porous cleaning nozzle 3025. The porous cleaning nozzle 3025 and the water pipe connected thereto are fixed below the first mounting platform 1 through the center position of the first mounting platform 1.

[0050] A liquid turbidity test sensor is installed inside the grouting material preparation tank 301 to detect whether the cleaning is clean.

[0051] like Figure 4 As shown, the cement single-liquid slurry storage device 4 includes a cement single-liquid slurry storage tank 401, a second motor 402, and a screw cap 403. The cement single-liquid slurry storage tank 401 is mounted on the second motor 402, and the top of the cement single-liquid slurry storage tank 401 is provided with a screw cap 403. Cement single-liquid slurry can be added into the storage tank by opening the screw cap 403.

[0052] like Figure 4 As shown, the bottom of the cement single-liquid slurry storage tank 401 is provided with a second rotating stirring shaft 404 and a second stirring blade 405. The second stirring blade 405 is evenly distributed from bottom to top on the second rotating stirring shaft 404. The second motor 402 is connected to the second rotating stirring shaft 404 and is used to drive the second rotating stirring shaft 404 and the second stirring blade 405 to ensure the uniformity of the cement single-liquid slurry and prevent it from settling and solidifying.

[0053] The cement single-liquid slurry storage tank 401 is provided with a first opening on its side, and the grouting material preparation device 3 is provided with a second opening on its side. The first opening and the second opening are connected by a pipe.

[0054] A liquid level monitoring sensor is installed on the inner wall of the cement single-liquid slurry storage tank 401 at the same height as the connected pipeline to ensure that enough cement single-liquid slurry is stored. When the liquid level of the cement single-liquid slurry drops to this point, the liquid level sensor will send the information that the liquid level is insufficient to the controller, reminding the user to add more.

[0055] The viscosity testing module includes a slurry holding device 5, a third motor 9, and a viscosity testing instrument 8. The slurry holding device 5 is set on the second mounting platform 7 and is used to hold the grouting material prepared by the material preparation module. It is equipped with a liquid turbidity testing sensor to detect whether the cleaning is clean.

[0056] The cement single-liquid slurry storage tank 401 also has a third opening on its side, and the slurry holding device 5 has a fourth opening on its side. The third opening and the fourth opening are connected by a pipe.

[0057] The device includes a first electric ball valve 10 and a second electric ball valve 11. The first electric ball valve 10 is disposed on a pipe connecting a first opening and a second opening, and the second electric ball valve 11 is disposed on a pipe connecting a third opening and a fourth opening.

[0058] The bottom of the slurry holding device 5 is equipped with cleaning blades 6. When it is necessary to clean the slurry holding device 5, it can receive clean water from the grouting material preparation tank 301 and connect the cleaning blades 6 to the third motor 9. When the third motor 9 is working, the cleaning blades 6 drive the clean water to rotate at high speed to complete the cleaning of the slurry holding device 5.

[0059] like Figure 5 As shown, the viscosity testing instrument 8 includes a rotational viscometer 801, a connecting rod 802, and a rotating cylinder 803. The rotational viscometer 801 is positioned directly above the slurry holding device 5. The rotational viscometer 801 and the rotating cylinder 803 are connected via the connecting rod 802, and the rotating cylinder 803 can be replaced according to the type of slurry. The rotating cylinder 803 is completely immersed in the slurry during operation.

[0060] The rotational viscometer 801 is equipped with a torque sensor and a speed sensor to measure the torque and speed of the rotating cylinder 803. After signal processing, the current liquid viscosity and speed information can be transmitted to the viscosity analysis module for viscosity analysis.

[0061] The specific process for obtaining viscosity test data includes:

[0062] Step 1: Open the first electric ball valve 10 and the second cap 403 to allow the cement single-liquid slurry to enter the grouting material preparation tank 301. The volume of the cement single-liquid slurry is controlled by adjusting the opening time of the first electric ball valve 10. During the operation of the device, the second motor 402 remains on to ensure that the stirring shaft inside the cement single-liquid slurry storage tank 401 is in working condition.

[0063] Preferably, the cement single-liquid slurry stored in the cement single-liquid slurry storage tank 401 has a water-cement ratio of 1:1.

[0064] Step 2: Based on the experimental plan for studying the effect of the current additive dosage on long-term viscosity, i.e., the types of additives and their corresponding dosages, rotate the three independent feeding turntables 201 sequentially so that the circular holes of the feeding turntables 201 are aligned with the bottom openings of the storage tubes 202 corresponding to the preset additives. The dosage of the additives is controlled by controlling the opening time. At this time, both the first electric ball valve 10 and the second electric ball valve 11 are in the closed state.

[0065] Step 3: Turn on the first motor 3023, and drive the stirring shaft 3021 to rotate through the speed reduction transmission gear set 3024 to fully stir the cement-based grouting material in the grouting material preparation tank 301, and obtain the cement-based grouting material under the research and test scheme of the effect of the current amount of admixture on the long-term viscosity;

[0066] Preferably, the stirring time is set to 10 minutes.

[0067] Step 4: Close the second electric ball valve 11, open the first electric ball valve 12, so that the cement-based grouting material in the grouting material preparation bucket 301 enters the grout holding device 5 of the viscosity test module, and close the second electric ball valve 11 and the third motor 9;

[0068] Step 5: Turn on the viscosity testing instrument 8 to test the viscosity of the cement-based grouting material. To ensure that the cement-based grouting material does not segregate and precipitate during the test, a stirring process needs to be added during the test. Specifically, the test process is set as follows: First, enter the testing phase, with a duration of 10 minutes, and record the viscosity value every 5 seconds. After the testing phase, enter the stirring phase, turn on the cleaning blade 701, continuously stir the grout for 30 seconds, and then let it stand for 30 seconds.

[0069] Step 6: After the test is completed, the viscosity data of the grouting material is transmitted to the viscosity analysis module, and the second electric ball valve 11 is opened to discharge the cement-based grouting material.

[0070] Step 7: The equipment cleaning and drying process is as follows: close the second electric ball valve 11, inject clean water into the grouting material preparation tank 301 through the water pipe connected to the multi-hole cleaning nozzle 3025, turn on the first motor connected to the cleaning blade 6, and clean the grouting material preparation tank 301 and the grout holding device 5. The cleaning time is set to 10 minutes. After one round of cleaning is completed, the second electric ball valve 11 will discharge the sewage.

[0071] Step 8: Close the second electric ball valve 11, inject clean water again through the water pipe connected to the multi-hole cleaning nozzle 3025, and obtain the turbidity of the cleaning liquid at this time through the liquid turbidity test sensor inside the grouting material preparation tank 301 and the grout holding device 5. When the turbidity reaches the preset standard, the second electric ball valve 11 can be opened to discharge the sewage and end the cleaning process; when the turbidity exceeds the standard, the first motor connected to the cleaning blade 6 is turned on again to start the next round of cleaning until the turbidity reaches the standard.

[0072] Step 9: Connect the air dryer to the pipeline and air dry the grouting material preparation tank 301 and the grout container 5. Set the air drying time to 5 minutes.

[0073] The viscosity analysis module includes a data receiving module, a data preprocessing module, and a long-term viscosity prediction module.

[0074] The data receiving module is configured to acquire grouting material viscosity data as a function of time, showing the changes in viscosity data of different admixture addition schemes during viscosity testing.

[0075] In this embodiment, while the equipment is being cleaned and dried, a viscosity analysis database is established based on the viscosity data of the grouting material and the current admixture addition scheme data;

[0076] Specifically, for historical viscosity data, a sliding window approach is used to establish a time-series data sample set;

[0077] First, since the viscometer is set to record every 5 seconds, the historical viscosity data has the following dimensions: Furthermore, by setting a sliding window with a fixed size of 120, a sliding step of 1, and a target output size of 10, the sliding window can generate... The time series data sample set consists of several samples.

[0078] The data preprocessing module is configured to perform data cleaning, data standardization, and data partitioning.

[0079] Data cleaning involves removing abnormal data that deviates from the normal range due to instrument malfunction or data transmission errors; optional data standardization methods include: Min-Max standardization, Z-Score standardization, exponential transformation, etc.

[0080] Data partitioning should include randomly dividing the initial viscosity data into training and test sets, with the training set accounting for 70% of the initial data and the test set accounting for 30%.

[0081] The long-term viscosity prediction module is configured to obtain the viscosity prediction value at the target time based on the external additive scheme to be analyzed and the trained long-term viscosity prediction model.

[0082] The long-term viscosity prediction model adopts a dual-input channel LSTM model, which includes a static input branch, a temporal input branch, a fusion layer, and an output layer.

[0083] Since the amount of additives added does not change over time, it is a static characteristic, while viscosity is a time-series data.

[0084] Therefore, the external additive data is input into the static input branch to obtain static features, and the liquid viscosity data is input into the time-series input branch to obtain time-series features. The liquid viscosity prediction features are then fused by the fusion layer, and the final long-term viscosity prediction results are generated by the fully connected layer.

[0085] Specifically, the static input branch is configured with two fully connected layers and one regularization layer: fully connected layer 1 has 64 neurons and includes one ReLU activation function; fully connected layer 2 has 128 neurons and includes one ReLU activation function. The regularization layer is a Dropout layer with a dropout ratio of 0.2.

[0086] The temporal input branch is configured with two stacked LSTM layers and one regularization layer: LSTM1 has 64 LSTM units and LSTM2 has 128 LSTM units. Stacking LSTM layers increases network depth, enabling the capture of more layers of temporal features and effectively improving the model's expressive power. The regularization layer is a Dropout layer with a dropout ratio of 0.2.

[0087] Specifically, the fusion layer includes a splicing layer and a fully connected layer. The splicing layer splices the outputs of the two branches; the fully connected layer has 128 neurons and includes a ReLU activation function. The output layer is set as a fully connected layer with 216 neurons and includes a linear activation function. The training process of the long-term viscosity prediction model adopts the idea of ​​stepwise recursive prediction, specifically including:

[0088] When the number of experiments exceeds the set number, such as 3 times, the short-term viscosity data obtained by the material viscosity test module is input into the stepwise recursive prediction module to perform stepwise recursive prediction of the long-term viscosity of the cement-based grouting material under the current admixture addition scheme. The dataset used for prediction will slide forward step by step after each training to ensure that the model can use the latest time series information for prediction.

[0089] Specifically, it includes:

[0090] First, after performing data preprocessing on the short-term viscosity data, the short-term viscosity data is input into a trained dual-input channel LSTM model to predict the time-series viscosity data at a set time step in the future.

[0091] Next, the predicted time-series viscosity data at the next set time step will replace the first N data points in the short-term viscosity data, where N is the same as the number of set time steps, and will be added to the end of the short-term viscosity data. Then, it will be input into the dual-input channel LSTM model for prediction.

[0092] Repeat the above recursive prediction process until the prediction time step is updated to the preset target time step. At this point, the model output is the viscosity prediction value at the target time in the study experiment on the effect of additive dosage on long-term viscosity.

[0093] Specifically, the training process includes the following steps:

[0094] In engineering applications, the viscosity change of cement-based grouting materials within 60 minutes usually has higher practical value, directly affecting the pumpability, stability and final effect of cement grout.

[0095] A set of viscosity effect tests is designed for each external additive scheme. Therefore, when there are ≤3 external additive schemes, a set of viscosity tests can be conducted for each external additive scheme. Since the prediction effect is not ideal when the viscosity change data exceeds 10 minutes, it is necessary to repeat the test 60min / 10min=6 times. Therefore, when there are ≤3 external additive schemes, the corresponding test data obtained for each external additive scheme includes 6 sets of short-term viscosity data.

[0096] However, in practical applications, there are more than three external additive schemes. If all data for each external additive scheme is tested, the efficiency is very low. Therefore, starting from the fourth external additive scheme, it is only necessary to obtain the short-term viscosity result corresponding to the current scheme. In this embodiment, this means obtaining the viscosity change data for 10 minutes, i.e., performing the above test and stirring operation once. Then, the short-term viscosity result of the fourth scheme and the historical viscosity prediction results corresponding to the first three external additive schemes are input together into the long-term viscosity prediction model to obtain the predicted value of the long-term viscosity, i.e., the viscosity change prediction of the fourth scheme within 60 minutes. Similarly, it can be deduced that by inputting the short-term viscosity result of the Nth scheme, i.e., the current scheme, and the historical viscosity prediction results corresponding to the first N-1 external additive schemes together into the long-term viscosity prediction model, the predicted value of the long-term viscosity can be obtained, i.e., the viscosity change of the cement-based grouting material within 60 minutes in this embodiment. This greatly improves the efficiency and accuracy of obtaining the long-term viscosity of the cement-based grouting material.

[0097] Furthermore, a loss function and optimizer are set, the loss is calculated based on the loss function, and the weights in the network are updated through the backpropagation algorithm and optimizer.

[0098] The static input branch calculates the gradient and updates the weights of the fully connected layer through backpropagation.

[0099] The LSTM network with the temporal input branch adjusts the weights of each LSTM layer through backpropagation in time (BPTT).

[0100] Specifically, the BPTT algorithm unfolds the LSTM network in the time dimension, treating it as a multi-layer feedforward neural network, and then applies the standard backpropagation algorithm to calculate the gradient.

[0101] Furthermore, repeat the training in multiple batches until the loss converges or the preset stopping condition is met.

[0102] Preferably, the loss function can be the mean squared error loss function, and the optimizer can be either the stochastic gradient descent algorithm or the Adam algorithm.

[0103] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A device for predicting the long-term viscosity of cement-based grouting materials, characterized in that, It includes a viscosity testing device and a long-term viscosity prediction device connected to it; The viscosity testing device includes an admixture dispensing device, a grouting material preparation device, a cement single-liquid slurry storage device, a slurry holding device, and a viscosity meter; the admixture dispensing device is installed at the upper end of the grouting material preparation device, and the two sides of the grouting material preparation device are respectively connected to the cement single-liquid slurry storage device and the slurry holding device with openable and closable channels; the viscosity meter is installed above the slurry holding device. The viscosity tester is used to obtain the viscosity of the grouting material in the grout container under different admixture addition schemes by the admixture dosing device. The long-term viscosity prediction device is configured to: predict the long-term viscosity of the grouting material under different admixture addition schemes based on the viscosity data of the grouting material obtained under different admixture addition schemes and the trained long-term viscosity prediction model; The long-term viscosity prediction model adopts a dual-input channel LSTM model, which includes a static input branch, a temporal input branch, a fusion layer, and an output layer. The admixture data is input into the static input branch to obtain static features, and the grouting material viscosity data is input into the time-series input branch to obtain time-series features. The static features and time-series features are fused through the fusion layer to obtain the fused grouting material viscosity prediction features, and the final long-term viscosity prediction result is generated through the output layer.

2. The long-term viscosity prediction device for cement-based grouting materials as described in claim 1, characterized in that, The admixture dispensing device includes multiple admixture dispensing cylinders (2), each admixture dispensing cylinder includes a feeding turntable (201) and multiple storage tubes (202). The feeding turntable (201) has a groove on its surface, which is fitted into the bottom of the storage tube (202). An opening is provided at the bottom of the groove. The storage tube (202) has an opening at the bottom and a screw cap is provided at the top.

3. The long-term viscosity prediction device for cement-based grouting materials as described in claim 2, characterized in that, The grouting material preparation device includes a grouting material preparation tank (301) and a first slurry stirring device (302). The first slurry stirring device (302) includes a first stirring rotating shaft (3021), a first stirring blade (3022), a first motor (3023), and a speed-reducing transmission gear set (3024). The first stirring rotating shaft (3021) and the first stirring blade (3022) are arranged inside the grouting material preparation tank (301). The first stirring rotating shaft (3021) is arranged at the bottom of the grouting material preparation tank (301), and multiple layers of first stirring blades (3022) of different sizes are distributed on it. One end of the speed-reducing transmission gear set (3024) is connected to the first stirring rotating shaft (3021), and the other end is connected to the first motor (3023).

4. The long-term viscosity prediction device for cement-based grouting materials as described in claim 3, characterized in that, The grouting material preparation device also includes a first mounting platform (1), and a multi-hole cleaning nozzle (3025) is provided on the top of the grouting material preparation tank (301). The multi-hole cleaning nozzle (3025) and the water pipe connected thereto pass through the center of the first mounting platform (1).

5. The long-term viscosity prediction device for cement-based grouting materials as described in claim 4, characterized in that, The admixture dispensing cylinder (2) includes three cylinders, which are arranged in an equilateral triangle on the first mounting platform (1). The three admixture dispensing cylinders contain three types of water-reducing agents, three types of retarders and three types of early-strength agents, respectively.

6. The long-term viscosity prediction device for cement-based grouting materials as described in claim 1, characterized in that, The cement single-liquid slurry storage device (4) includes a cement single-liquid slurry storage tank (401), a second motor (402), and a screw cap (403). The cement single-liquid slurry storage tank (401) is mounted on the second motor (402). A screw cap (403) is provided on the top of the cement single-liquid slurry storage tank (401). A second rotating stirring shaft (404) and a second stirring blade (405) are provided at the bottom of the cement single-liquid slurry storage tank (401). The second stirring blade (405) is evenly distributed from bottom to top on the second rotating stirring shaft (404). The second rotating stirring shaft (404) is connected to the second motor (402).

7. The long-term viscosity prediction device for cement-based grouting materials as described in claim 1, characterized in that, The viscosity testing device includes a second mounting platform (7), a slurry holding device (5), a third motor (9), and a viscosity testing instrument (8). The slurry holding device (5) is mounted on the second mounting platform (7). The viscosity testing instrument (8) includes a rotational viscometer (801), a connecting rod (802), and a rotating cylinder (803). The rotational viscometer (801) is positioned directly above the slurry holding device (5), and the rotational viscometer (801) is connected to the rotating cylinder (803) via the connecting rod (802).

8. The long-term viscosity prediction device for cement-based grouting materials as described in claim 6, characterized in that, The cement single-liquid slurry storage tank (401) is provided with a first opening on its side, and the grouting material preparation device is provided with a second opening on its side. The first opening and the second opening are connected by a pipe. The cement single-liquid slurry storage tank (401) also has a third opening on its side, and the slurry holding device (5) has a fourth opening on its side. The third opening and the fourth opening are connected by a pipe.

9. The long-term viscosity prediction device for cement-based grouting materials as described in claim 8, characterized in that, The device includes a first electric ball valve (10) and a second electric ball valve (11). The first electric ball valve (10) is disposed on a pipe connected to a first opening and a second opening, and the second electric ball valve (11) is disposed on a pipe connected to a third opening and a fourth opening.

10. The long-term viscosity prediction device for cement-based grouting materials as described in claim 1, characterized in that, The training process for the long-term viscosity prediction model includes: When the admixture addition scheme is less than or equal to n, a set of viscosity tests is performed for each scheme. The viscosity tests are grouped according to the effective time of the viscosity change data, and the short-term viscosity data of each group of tests is obtained. When there are more than n admixture addition schemes, only the short-term viscosity data corresponding to the current m-th scheme is obtained. The short-term viscosity result corresponding to the m-th scheme and the historical viscosity prediction results corresponding to the previous m-1 admixture addition schemes are input into the long-term viscosity prediction model to obtain the long-term viscosity prediction value of the m-th scheme, where m is greater than n.

Images