A method and system for adaptive control of a printing concrete mixing process

Abstract

The application provides a printing concrete mixing process adaptive control method and system; the method comprises the following steps: acquiring a material surface visual image sequence in a mixing bin in real time, monitoring material thermal spectrum evolution, and acquiring a main shaft instantaneous torque; obtaining visual uniformity, temperature change rate, and a linear mapping relationship between the main shaft instantaneous torque and offline yield stress from three original signals; S3, according to the visual uniformity, the temperature change rate, and the main shaft instantaneous torque, adaptively controlling the printing concrete mixing process; the system comprises a monitoring module, a calculation center, and a control unit; the application eliminates the blind spot of single perception by using the physical compensation effect between multi-modal signals, realizes the paradigm leap from static experience direction to dynamic material state response, and can accurately ensure the rheological consistency and batch stability in the concrete structure construction process.

Description

Technical Field

[0001] This application relates to the fields of civil engineering and automated construction technology, and in particular to an adaptive control method and system for the printing concrete mixing process. Background Technology

[0002] 3D concrete printing technology, as a novel additive manufacturing technology, has shown great potential in the field of automated construction, such as tunnel lining and building components. Unlike traditional casting processes, 3D printed concrete requires materials to have extremely high rheological consistency after mixing and discharge to meet stringent requirements for pumpability, extrudability, and stacking stability. This necessitates precise control over the fresh mix properties of 3D printed concrete, as even slight fluctuations in homogeneity or rheology can lead to pump blockage, structural collapse, or poor interlayer bonding.

[0003] Currently, industry relies primarily on empirical timed stirring or single-modal sensing and monitoring, such as machine vision, torque, and temperature monitoring. Machine vision technology is widely used to assess the geometric uniformity of material surfaces, but due to the wavelength limitation of visible light, it cannot penetrate the material surface to perceive the internal volumetric rheological evolution and hydration hardening degree. While monitoring schemes based on spindle power or dynamic torque can reflect the macroscopic resistance of the material, they are easily affected by localized material accumulation or mechanical load fluctuations within the mixing chamber, and it is difficult to distinguish whether the temperature rise is caused by physical friction or condensation caused by chemical exothermics. Temperature can monitor temperature changes inside or on the surface of the material system during stirring, but single-mode temperature monitoring only reflects the total heat effect of the system and cannot distinguish between hydration exothermics, mechanical frictional heat, and ambient heat conduction. Furthermore, temperature data is mostly single-point data, making it difficult to accurately reflect the three-dimensional temperature field distribution within the mixing chamber. These single-modal monitoring methods have perceptual "blind spots" between surface uniformity and deep rheological properties, preventing current technologies from accurately capturing the transition of materials from "physical mixing" to "chemical reaction" and the critical states of staged chemical reactions. Especially when there are fluctuations in ambient temperature or minor adjustments to the composition of raw materials, static formulation execution often leads to batch-to-batch performance deviations in printing materials, which in turn can cause construction accidents such as pipe blockage or collapse.

[0004] Therefore, how to establish a multimodal fusion monitoring system with physical interpretability, utilize complementary physical signals from different dimensions to eliminate perception blind spots, and achieve adaptive discrimination and dynamic control of the entire mixing process has become a key technical challenge for realizing unmanned concrete construction and high-quality closed-loop production. Summary of the Invention

[0005] The purpose of this application is to provide an adaptive control method and system for the printing concrete mixing process. It aims to eliminate the blind spots of single perception by utilizing the physical compensation effect of multi-source sensing modes of vision, thermodynamics and mechanics. By establishing a physical-driven "physical-chemical" heat generation decoupling arbitration mechanism, it can achieve accurate identification and rheological closed-loop control of the entire mixing process evolution stage, and ensure that the output material has high rheological consistency and batch stability.

[0006] To achieve the above objectives, this application provides the following technical solution:

[0007] This application provides an adaptive control method for the concrete mixing process in printing, including the following steps:

[0008] S1. Real-time acquisition of visual image sequences of material surface in the mixing chamber, monitoring of material thermal spectrum evolution, and acquisition of instantaneous torque of the main shaft;

[0009] S2. Feature extraction and physical mapping are performed on the three original signals collected in S1 to obtain the linear mapping equation between visual uniformity, temperature change rate, and instantaneous torque of the spindle and offline yield stress.

[0010] S3. Based on the visual uniformity, temperature change rate, and instantaneous spindle torque, adaptive control is performed on the concrete mixing process: using the visual uniformity threshold as a basic constraint and the temperature change rate as a chemical arbitration signal, the current mixing process stage is determined autonomously; if the visual uniformity reaches 95% and the temperature change rate is in the high-level frictional exothermic range, it is determined that the dry mixing stage is completed, and an instruction to add water with pre-mixed water-reducing agent is issued; if the visual uniformity reaches 95% or more and the temperature change rate is in the low-level plateau range, it is determined that the slow hydration state is reached, and an instruction to inject accelerator is issued; if the visual uniformity reaches 95% or more, the temperature change rate enters the stable hydration range, and the instantaneous spindle torque reaches the target calibration range, it is determined that the target rheological window has been reached, and a discharge instruction is issued; otherwise, return to step S1.

[0011] Furthermore, in step S1, the sampling frequency is 1Hz, and the three data sampling frequencies are strictly synchronized.

[0012] Furthermore, in step S2, visual uniformity is obtained by extracting the ROI from the video frames captured by the camera and removing interference areas; converting the color image to a grayscale image and calculating the standard deviation of the pixel grayscale histogram.

[0013] Furthermore, the visual uniformity The formula is as follows:

[0014] ;

[0015] In the formula, This represents the maximum value of the visual texture index within the test interval. This represents the final stable value of the visual texture index. This refers to the real-time visual texture index; among which, , It was obtained by fitting the test results of the real-time visual texture index.

[0016] Furthermore, the temperature change rate is dynamically extracted from the highest temperature on the material surface as a representative value, and the temperature data within a preset sliding window is fitted using a linear least squares regression algorithm to calculate its slope.

[0017] Furthermore, the rate of temperature change The formula is:

[0018] ;

[0019] In the formula, This represents the temperature increment, indicating the difference between the real-time measured temperature and the ambient temperature, where n is the number of data points in the window. It is all within the window The average value of the data points The average value of the time series. In order to be in Temperature increment over time.

[0020] Furthermore, in step S3, the system evaluates the rate of temperature change. If the visual uniformity reaches 95% or higher and the temperature change rate is greater than 0.01 K / s, the dry mixing stage is considered complete, and an instruction to add water premixed with the water-reducing agent is issued. If the visual uniformity reaches 95% or higher and the temperature change rate is less than 0.001 K / s, it indicates the low-heat activity characteristic of the induction period, and is considered a slow hydration state, issuing an instruction to inject the quick-setting agent in one go. If the visual uniformity reaches 95% or higher and 0.001 K / s ≤ If the instantaneous torque of the spindle reaches ≤0.01K / s, it is determined to be a rapid hydration state. If the instantaneous torque of the spindle reaches the target calibration range, it is determined to have reached the target rheological window and the unloading signal is activated; otherwise, return to step S1.

[0021] Furthermore, the steps for determining the linear mapping equation between the instantaneous torque of the spindle and the offline yield stress, and the target calibration interval, include:

[0022] Acquire the instantaneous torque of the spindle at multiple different times during the concrete mixing process, as well as the offline yield stress data obtained by synchronous sampling.

[0023] A linear regression algorithm was used to establish a linear mapping equation between the instantaneous torque of the spindle and the offline yield stress.

[0024] Based on the static yield stress threshold requirement for the bottom material to prevent collapse in the 3D printing process, the theoretical minimum static yield stress threshold required for the bottom material to prevent collapse is substituted into the linear mapping equation to back-calculate the critical value of the target spindle torque, and combined with the preset safety redundancy range, the target calibration interval is determined.

[0025] This application also proposes an adaptive control system for the printing concrete mixing process, comprising:

[0026] Monitoring module: Used to collect multimodal raw signals of vision, thermal sensing and instantaneous torque of the main shaft in real time during the stirring process;

[0027] The computing center is used to perform feature mapping on the original signal, run a phased logic gate fusion algorithm to identify the process stage, and generate decision instructions for feeding or discharging materials based on the real-time status of the materials.

[0028] Control unit: Used to receive the decision instructions, drive the feeding execution mechanism or the discharging mechanism to operate, and realize closed-loop control of the mixing process.

[0029] Furthermore, the monitoring module includes a 4K industrial camera installed at the center of the top of the mixing chamber, an infrared thermal sensor deployed inside the mixing drum, and a dynamic torque sensor installed on the mixing spindle; a sunshade is provided on the top of the mixing chamber to shield the top of the 4K industrial camera.

[0030] The technical solution of this application has the following beneficial effects:

[0031] The visual texture index, temperature change rate, and torque proxy index extracted by this invention have clear physical meanings, making the control logic conform to the basic laws of concrete materials science and improving the robustness of the system under complex working conditions. By using the temperature change rate as a "chemical arbitration signal", it is possible to accurately distinguish between physical frictional heat and chemical exothermic heat, so that even under fluctuations in raw materials and environment, it can still accurately identify the optimal feeding point and discharge time, ensuring rheological consistency and construction quality in the 3D printing process.

[0032] This invention utilizes complementary physical signals from three dimensions—visual, torque, and temperature—to monitor both the geometric uniformity of the surface and capture deep-seated volumetric rheological evolution and chemical kinetic information, effectively overcoming the limitations of single-mode monitoring. It also enables a shift from experience-based timed stirring to material state-driven monitoring, reducing human intervention and mitigating the risk of pipe blockage or collapse due to fluctuations in material properties. Attached Figure Description

[0033] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. Wherein:

[0034] Figure 1 This is a schematic diagram of the logic gate control of the method of the present invention.

[0035] Figure 2 This is a flowchart illustrating the texture information extraction process according to an embodiment of the present invention.

[0036] Figure 3 The graph shows the evolution of the visual texture index over time during the dry mixing process of concrete raw materials; where (a), (b), (c), and (d) are the evolution graphs for the dry mixing process, adding water to the dry mixed material, adding water and PCS to the dry mixed material, and adding an accelerator after the mixture with added PCS reaches a stable state.

[0037] Figure 4 To print the temperature increment over time during the dry mixing process of concrete raw materials; where (a), (b), (c), and (d) are the evolution diagrams of the dry mixing process, adding water to the dry mixed material, adding water and PCS to the dry mixed material, and adding accelerator after the mixture with added PCS reaches a stable state.

[0038] Figure 5 The graph shows the torque evolution over time during the dry mixing process of concrete raw materials. (a), (b), (c), and (d) are graphs showing the evolution of torque over time during the dry mixing process, the addition of water to the dry mixing material, the addition of water and PCS to the dry mixing material, and the addition of accelerator after the mixture with PCS reaches a stable state.

[0039] Figure 6 The graphs show the changes in static yield stress, dynamic yield stress, and plastic viscosity over time for verification examples of the present invention; wherein (a), (b), and (c) are graphs showing the changes when water is added to the dry mixture (Group B1), water and PCS are added to the dry mixture (Group B2), and an accelerator is added after the mixture with added PCS reaches a stable state (Group C), respectively.

[0040] Figure 7 A mapping relationship between static yield stress, dynamic yield stress and instantaneous torque of the spindle was established for verification purposes. Detailed Implementation

[0041] The present application will now be described in detail with reference to the accompanying drawings and embodiments. Various examples are provided by way of explanation and not by way of limitation. In fact, those skilled in the art will recognize that modifications and variations can be made to the present application without departing from the scope or spirit thereof. For example, a feature shown or described as part of one embodiment may be used in another embodiment to produce yet another embodiment. Therefore, it is desirable that the present application encompass such modifications and variations that fall within the scope of the appended claims and their equivalents.

[0042] An adaptive control method for a concrete mixing process includes the following steps:

[0043] S1. Using a sampling frequency of 1Hz, acquire the visual image sequence of the material surface in the mixing chamber in real time, monitor the evolution of the material thermal spectrum, and acquire the instantaneous torque of the main shaft;

[0044] S2. Feature extraction and physical mapping are performed on the three raw signals acquired in S1, including:

[0045] S21. From visual texture to visual uniformity: Extract the ROI (region of interest) from the video frames captured by the camera and remove interfering areas such as stirring blades; convert the color image to grayscale image and calculate the standard deviation of the pixel grayscale histogram as the visual texture index, which is used to quantify the entropy value of the spatial distribution of material particles, i.e., visual uniformity.

[0046] S22. From material thermal spectrum to temperature change rate: In order to accurately reflect the overall temperature and reduce thermal interference from the low-temperature mixing chamber, mixing blades, etc., the highest surface temperature of the material is dynamically extracted as a representative value. The temperature data within the preset sliding window is fitted using a linear least squares regression algorithm, and its slope is calculated as the temperature change rate. The temperature change rate serves as a "chemical arbitration signal" to distinguish between heat generated by physical friction and heat released by chemical reaction, and is used to arbitrate whether the material is in the dry mixing stage, the slow hydration stage, or the rapid hydration stage.

[0047] S23. Torque and Material Yield Stress: Torque is directly correlated with key rheological parameters, including apparent viscosity and yield stress; the calculation center establishes a linear mapping equation between the instantaneous torque of the main shaft and the offline yield stress, and uses the instantaneous torque value as a real-time online proxy index for the evolution of material yield stress; in the control logic of this application, a specific torque is the target.

[0048] S3. Input visual uniformity, temperature change rate, and instantaneous spindle torque into the calculation center and run the multimodal fusion algorithm. Using the visual uniformity threshold as the basic constraint and the temperature change rate as the chemical arbitration signal, autonomously determine the current stirring process stage. If the visual uniformity reaches 95% and the temperature change rate is in the high-level frictional exothermic range, it is determined that the dry mixing stage is completed, and an instruction to add water with pre-mixed water-reducing agent is issued. If the visual uniformity reaches 95% or more and the temperature change rate is in the low-level plateau range, it is determined that the slow hydration state is reached, and an instruction to inject accelerator is issued. If the visual uniformity reaches 95% or more, the temperature change rate enters the stable hydration range, and the instantaneous spindle torque reaches the target calibration range, it is determined that the target rheological window has been reached, and a discharge instruction is issued. If not (either visual uniformity or torque fails to meet the standard), return to step S1 to continue stirring.

[0049] Furthermore, in step S21, during the processing of the original visual signal, consecutive video frames are first adjusted to specified pixels, and then processed using a Region of Interest (ROI) extraction algorithm. Afterward, the image cropped by the ROI is converted to a grayscale image to eliminate color noise while preserving topological texture information. The degree of material dispersion within the mixing chamber is determined using a visual texture index. To quantify this, the visual texture index is the standard deviation of the gray-level histogram excluding the ROI:

[0050] (1);

[0051] In the formula, This represents a grayscale value from 1 to 255. This represents the average grayscale value of the effective ROI. Gray values ​​appearing in the unremoved area normalized frequency, This is a calibration factor used to standardize the formula based on baseline lighting conditions, enhancing the system's robustness to ambient light fluctuations while ensuring efficient computation for low-latency monitoring; dynamic calibration factor. The specific value is: during the dry mixing stage The value is preset to 1; for subsequent stirring process stages (such as the liquid phase dispersion stage after water addition), its The value is taken as: at the end of the previous process stage (i.e., visual uniformity) The ratio of the final stable value of the visual texture index (at 95%) to the initial visual texture index measurement value at the current process stage.

[0052] In heterogeneous materials, the varying color contrast of different particle components leads to a random distribution of pixel intensity. As the material is stirred and mixed, the disordered particles form a uniform surface, resulting in a higher visual texture index. The asymptotes show gradual convergence and decay to a stable asymptote, indicating that the material system within the mixing chamber has achieved optimal homogeneity; homogeneity is measured by visual uniformity. Quantitative representation, the target visual uniformity is set to ≥95%, used to determine the endpoint of the dry mixing and wet mixing processes; the visual uniformity The formula is as follows:

[0053] (2);

[0054] In the formula, This represents the maximum value of the visual texture index within the test interval. This represents the final stable value of the visual texture index. This refers to the real-time visual texture index; among which, , It was obtained by fitting the test results of the real-time visual texture index.

[0055] Furthermore, the rate of temperature change in step S22 The formula is:

[0056] (3);

[0057] In the formula, This represents the temperature increment, indicating the difference between the real-time measured temperature and the ambient temperature, where n is the number of data points in the window. It is all within the window The average value of the data points The average value of the time series. In order to be in Temperature increment over time;

[0058] In addition, the piecewise linear regression method is used to determine the starting point of the steady temperature rise. By monitoring the temperature change rate, the system can distinguish different mixing stages. A higher temperature change rate indicates that the material is in the dry mixing stage or the rapid hydration stage due to the addition of a quick-setting agent, at which point there is high friction between the material particles. A lower temperature change rate indicates that the material is in the slippery wet mixing stage.

[0059] Furthermore, the mixing strategy for printed concrete is consistent with conventional methods, employing a segmented mixing strategy: an initial dry mixing stage, followed by the introduction of liquid admixtures, and finally the injection of accelerators. For example... Figure 1 As shown in Figure S3, the multimodal fusion algorithm adopts a hierarchical decision-making framework adapted to the three stages. The transition between each stage is controlled by three sequential logic gates, and each logic gate needs to simultaneously satisfy visual uniformity. ≥95% physical conditions and a specific stage check. First, the system assesses the rate of temperature change. If the visual uniformity reaches 95% or higher and the temperature change rate is greater than 0.01 K / s, the dry mixing stage is considered complete, and an instruction to add water premixed with the water-reducing agent is issued. If the visual uniformity reaches 95% or higher and the temperature change rate is less than 0.001 K / s, it indicates the low-heat activity characteristic of the induction period, and is considered a slow hydration state, issuing an instruction to inject the quick-setting agent in one go. If the visual uniformity reaches 95% or higher and 0.001 K / s ≤ If the temperature change rate is ≤0.01K / s, it is determined to be in a rapid hydration state. If the instantaneous torque of the main shaft reaches the target calibration range at this time, it is determined to have reached the target rheological window, and the unloading signal is activated; otherwise, return to step S1. The above temperature change rate thresholds (0.01K / s, 0.001K / s) are not arbitrarily set, but are empirical critical values ​​based on a large number of actual engineering application tests and statistical summaries. Specifically, this set of measured thresholds was obtained through multiple batches of on-site measurements and data summarization using an industrial-grade vertical shaft single-shaft five-jaw construction site mixer with a diameter of 2m and a height of 1m. This value truly reflects the significant difference across orders of magnitude between the physical frictional heat generation, liquid phase lubrication drag reduction, and chemical exothermic reaction of materials under actual industrial mixing scale, thus providing solid and reliable experimental data support for the multimodal phased chemical arbitration of this invention.

[0060] This application also establishes an adaptive control system for the concrete mixing process (hereinafter referred to as the adaptive control system), which aims to optimize the timing of additive addition and unloading during standardized mixing based on real-time multimodal monitoring; the adaptive control system includes:

[0061] Monitoring module: Used to collect multimodal raw signals of vision, thermal sensing and instantaneous torque of the main shaft in real time during the stirring process;

[0062] The computing center is used to perform feature mapping on the original signal, run a phased logic gate fusion algorithm to identify the process stage, and generate decision instructions for feeding or discharging materials based on the real-time status of the materials.

[0063] Control unit: Used to receive the decision instructions, drive the feeding execution mechanism or the discharging mechanism to operate, and realize closed-loop control of the mixing process.

[0064] The adaptive control system proposed in this application integrates three different sensing methods to ensure comprehensive process control: visual texture is used to quantify spatial uniformity, the instantaneous torque of the master shaft (hereinafter referred to as torque) is used to characterize the overall rheological parameters, and thermal analysis acts as a chemical arbitrator. The integration of these three methods is crucial for overcoming the inherent sensing blind spots of single-modal methods, achieving automated and precise control of the concrete mixing process. This adaptive control system eliminates variability caused by human intervention, accurately determining the optimal timing for material addition based on the state of the materials in the mixing chamber, ensuring consistency in the preparation of printed concrete across different batches, and guaranteeing the homogeneity and yield stress evolution required for high-quality 3D printing.

[0065] Example

[0066] The raw material composition of the printed concrete is as follows: 1 part cement, 2 parts fine aggregate, 1.5 parts coarse aggregate, water-cement ratio 0.38, and sand ratio 0.57. The contents of hydroxypropyl methylcellulose (HPMC), polycarboxylate superplasticizer (PCS), and aluminum sulfate-based alkali-free quick-setting agent are 0.1%, 1.00%, and 2.50% of the cement mass, respectively. The cement has a density of 3.1 g / cm³. 3 P.O42.5 silicate cement; coarse aggregate particle size 5mm-20mm, fine aggregate particle size 0-5mm, density modulus 3.24, dry density of both aggregates 2.65g / cm³. 3 Hydroxypropyl methylcellulose (HPMC) viscosity 200,000 mPa·s. The 28-day compressive strength of this formula, tested according to the "Standard for Test Methods of Physical and Mechanical Properties of Concrete" GB / T50081-2019, is 42 MPa.

[0067] Mixing platform and adaptive control system: The mixing platform adopts a JW2000B single-shaft vertical mixer with a mixing chamber capacity of 3 cubic meters, a drive motor of 18.5KW, and five mixing blades. The mixing speed is set to a constant 22 rpm. Hydroxypropyl methylcellulose (HPMC) is added by a 1.1KW pump.

[0068] The monitoring module consists of a 4K industrial camera installed at the top center of the single-shaft vertical mixer, a HIKMICROK09 infrared thermal sensor deployed inside the mixing tank, and a Lanmec TR-10 dynamic torque sensor installed on the mixing spindle. These are used to acquire real-time visual image sequences of the material surface, monitor the thermal spectrum evolution of the mixture, and acquire the instantaneous torque of the spindle, respectively. To achieve stable visual acquisition, a sunshade is installed on the top of the mixing tank to shield the top of the 4K industrial camera.

[0069] The plan is to print a single layer of concrete with a height of 80 mm on a horizontal ground. The printed product will be rectangular in the vertical plane. According to preliminary tests, the printable static yield stress is greater than 1.17 kPa, and the target calibration range is [1.580 kN·m, 1.650 kN·m].

[0070] An adaptive control method for a concrete mixing process includes the following steps:

[0071] S1. Using a sampling frequency of 1Hz, acquire the visual image sequence of the material surface in the mixing chamber in real time, monitor the evolution of the material thermal spectrum, and acquire the instantaneous torque of the main shaft, while ensuring alignment with time.

[0072] S2. Feature extraction and physical mapping are performed on the three raw signals acquired in S1, including:

[0073] S21. First, ImageJ software was used to crop each frame of the image to extract the Region of Interest (ROI) containing the concrete inside the mixing chamber. These ROIs were then uniformly adjusted to a standard resolution of 1080×1080 pixels to ensure consistency in subsequent analysis. Since the background inside the mixing chamber exhibits a distinct red characteristic, MATLAB software was used to extract the red channel grayscale values ​​from the cropped RGB image. By setting a threshold of 0.5, the largest connected regions with red channel intensity exceeding this threshold were identified and retained, thus generating a binary mask. This mask is used to exclude non-concrete areas, such as mixing blades and chamber walls, in subsequent analysis. The area covered by the binary mask is then converted into a grayscale image as a visual texture index for later analysis. Input data, see Formula (1), and further use its quantitative indicator visual uniformity. The representation is shown in formula (2); the flowchart for texture information extraction is shown in [link to flowchart]. Figure 2 ;

[0074] S22, The rate of temperature change is calculated by formula (3);

[0075] S23. While monitoring the instantaneous torque of the spindle online, an offline rheometer TR-CRI is used as the calibration benchmark. The key parameters measured include static yield stress, dynamic yield stress, and plastic viscosity. The torque at different stirring moments is paired with the yield stress (static yield stress and dynamic yield stress) using the offline rheometer to establish a linear mapping equation between the instantaneous torque of the spindle and the offline yield stress. The target torque is calculated based on the target yield stress required by the 3D printing process using this linear mapping equation.

[0076] S3. Input visual uniformity, temperature change rate, and instantaneous spindle torque into the calculation center and run the multimodal fusion algorithm. Using the visual uniformity threshold as the basic constraint and the temperature change rate as the chemical arbitration signal, autonomously determine the current stirring process stage. If the visual uniformity reaches 95% and the temperature change rate is in the high-level frictional exothermic range, it is determined that the dry mixing stage is completed and a water injection command is issued. If the visual uniformity reaches 95% or more and the temperature change rate is in the low-level plateau range, it is determined that the slow hydration state is reached and a accelerator injection command is issued. If the visual uniformity reaches 95% or more and the temperature change rate enters the stable hydration range and the instantaneous spindle torque reaches the target calibration range [1.580 kN·m, 1.650 kN·m], it is determined that the target rheological window has been reached and a discharge command is issued. If the target rheological window has not been reached (torque or visual uniformity has not met the standard), return to step S1.

[0077] Figure 3 To investigate the evolution of the visual texture index during the dry mixing process of printed concrete raw materials, which exhibits a monotonic decay, an exponential decay model was fitted to the experimental data. The fitted curve is:

[0078] ;

[0079] Correspondingly, visual uniformity The derivation result is as follows:

[0080] ;

[0081] In the formula, This represents the maximum value (starting value) of the visual texture index within the test interval. This represents the final stable value of the visual texture index. It is the characteristic time of the dry mixing and dispersion process. , , All of these are fitting coefficients obtained from fitting the above exponential decay model. The time when the visual texture index reaches its maximum value; It is a natural constant;

[0082] Figure 3 (a) Figure 4 Figure (a) shows the changes in visual texture index and temperature increment over time during the dry mixing process of printed concrete. Cement, fine aggregate, coarse aggregate, and hydroxypropyl methylcellulose (HPMC) were added to the mixing chamber for dry mixing. As the dry mixing of the printed concrete progressed, the visual uniformity... , The temperature converged to 95% of the target value at 217 s; after an initial stabilization period of 2 minutes, the temperature increment showed a continuous increase, with a temperature change rate of 1.53 × 10⁻⁶ from 140 s to 217 s. -2 K / s, greater than 0.01 K / s, is sufficient for the adaptive control system to determine that the dry mixing stage is complete. The adaptive control system then issues a water injection command.

[0083] Figure 3 Figures (b) and (c) show the changes in visual texture index over time for the dry-mixed material after adding water (control group) and after adding water and polycarboxylate superplasticizer (PCS), respectively, at 217 s. The figures show that after the liquid is added, both exhibit characteristic wetting peaks, with the visual texture index rising rapidly and then gradually decaying exponentially. Although the peak structures of both appear almost simultaneously, at 364 s and 362 s respectively, there are significant differences in the curves during the subsequent steady-state phase. The control group, with only water added, shows visual uniformity... , Converged to 95% of the target value at 648s; with water and PCS, visual uniformity... , The two converged to 95% of the target value at 512s; the different times it took for them to reach the target visual uniformity indicate that this method can distinguish and quantify the accelerated dispersion effect caused by additives. Figure 4 Figures (b) and (c) show the temperature changes over time when water was added to the dry-mixed material at 217 s (control group) and when both water and polycarboxylate superplasticizer (PCS) were added, respectively. It can be seen that after the addition of the liquid phase, the temperature curve exhibited a characteristic instantaneous fluctuation, followed by a rapid linear increase. Linear regression analysis of the stable segments of the two curves showed that when only water was added after dry mixing (control group), the temperature change rate in the stable segment from 295 s to 684 s was 4.251 × 10⁻⁶. -4 K / s; with the addition of water and PCS, the temperature change rate during the steady-state period of 284s-512s is 1.137×10 K / s. -4 K / s; the temperature change rate of both is less than 0.001 K / s, the adaptive control system judges it to be in the slow hydration stage and issues a command to inject accelerator.

[0084] Figure 3 (d) Figure 4 (d) Figure 5 The figures in (d) show the changes in visual texture index, temperature, and instantaneous torque of the spindle over time after the mixture containing PCS reached a stable state and was then treated with an aluminum sulfate-based alkali-free accelerator (hereinafter referred to as accelerator). The addition of the accelerator caused a significant short-term fluctuation in the visual texture index, indicating that the adaptive control system is responsive to rapid microstructural changes related to rapid hardening. Figure 4 The addition of the medium-speed setting agent (d) caused a brief, transient temperature drop in the mixture of about 10 seconds, followed by a rapid, linear increase. During the stable period of 530-792 seconds, the temperature change rate was 2 × 10⁻⁶. -3 K / s. Comparing the temperature characteristics of the three stages, the temperature change rate differs by an order of magnitude, providing strong experimental evidence for the logic gate separation threshold proposed in this method.

[0085] Figure 5 (a), (b), (c), and (d) in the figure represent torque monitoring graphs for the following stages: dry mixing of raw materials, addition of water to dry-mixed raw materials (control group), addition of water and water-reducing agent to dry-mixed raw materials, and addition of accelerator after the mixture with PCS reaches a stable state. During the dry mixing stage ( Figure 5 In (a), the torque rises sharply after the mixer starts, and then stabilizes at a constant baseline level. Figure 5 In cases (b) and (c), after the addition of the liquid phase, the torque curves exhibit typical instantaneous fluctuations, reflecting the resistance changes during the wetting process, before rapidly decreasing to a new steady state. In contrast, Figure 5After adding the accelerator in (d), the torque suddenly increased and then continued to rise slowly, indicating that the macrostructure was continuously forming. Linear regression analysis was performed on the stable growth segment of the torque-time curve. Figure 5 The stable growth segments (b), (c), and (d) occurred after 648 s, 512 s, and 792 s, respectively, showing a rapid decrease in torque and regional stabilization after the addition of the liquid phase. After the addition of the accelerator, consistent with the thermal analysis results, the torque-time curve showed a significant positive slope of 0.186 Nm / s.

[0086] The torque is monitored in real time. When the static yield stress is greater than or equal to 1.17 kPa, that is, when the torque reaches 1.580 kN·m, the adaptive control system immediately issues a discharge command.

[0087] Printing tests were conducted using the concrete mixture obtained above. The target width of the printed product was 200 mm, the extruder nozzle diameter was 200 mm, and the extruder nozzle moving speed was 150 mm / s. The concrete was successfully pumped and extruded during the printing test, and no significant shape change was observed after extrusion. The vertical deformation rate of the printed product was less than 2%, demonstrating that the concrete material obtained using this adaptive control method and system can resist the self-weight of 3D printed concrete. Furthermore, within a 1-meter printing distance, the width of the printed product was 200 ± 2 mm, with a maximum width fluctuation of 1%, indicating that the concrete material obtained using this adaptive control method and system has a uniform component distribution.

[0088] Preliminary experiment

[0089] The process of calibrating the target rheological window / target calibration range of torque using torque values ​​and rheological parameters is as follows:

[0090] (1) Preparation of benchmark test samples: According to the concrete raw material components and mix proportions described in the example, the initial dry mixing and liquid phase mixing are carried out in a JW2000B single-shaft vertical mixer to ensure that the initial material state is consistent.

[0091] (2) Synchronous acquisition of online torque and offline sampling: After the accelerator is added to the mixing chamber, the concrete material enters the rapid hydration stage, and the yield stress begins to increase continuously. During this hardening process, the stable torque value of the mixing spindle is recorded at each preset time node (e.g., every 30 seconds). (Unit: kN·m), and immediately stop the mixer, simultaneously extracting concrete samples from different depths in the mixing chamber.

[0092] (3) Offline rheological parameter testing: The obtained concrete sample is quickly loaded into the test container of the offline rheometer (TR-CRI), and the standard quasi-static shear rheological test protocol is executed to accurately measure the static yield stress of the concrete at the corresponding sampling time. (Unit: kPa) and dynamic yield stress.

[0093] (4) Establishing a mapping model: After multiple sampling tests, multiple sets of torque-yield stress corresponding data pairs were obtained within the complete hardening cycle. The linear least squares regression algorithm was used to fit the data set, establishing a linear mapping equation between online torque and offline static yield stress:

[0094] ;

[0095] In the formula, The mapping slope constant is... is the intercept constant. The fitting results show that the linear model has extremely high coefficients of determination, proving that the instantaneous torque of the spindle can be used as a reliable online proxy index for characterizing rheological properties.

[0096] (5) Solving the target rheological window: It is planned to print a single layer of concrete with a layer height of 80mm on a horizontal ground. The vertical plane of the printed product is rectangular, and its printable static yield stress satisfies: Static yield stress =1.17 kPa. Based on the above 3D printing process requirements, the theoretical minimum static yield stress threshold required to prevent the bottom material from collapsing is... Substituting into the established linear mapping equation, the corresponding target spindle torque critical value is calculated to be 1.580 kN·m. To ensure safety redundancy and smooth material pumping during the 3D printing process and to eliminate occasional signal fluctuation interference, the target calibration range of torque (i.e., the target rheological window) is set to [1.580 kN·m, 1.650 kN·m]. This calibration range is then written into the decision algorithm of the adaptive control system as the hardware execution benchmark for triggering the material discharge command in step S3.

[0097] Verification Example

[0098] Synchronous testing was conducted using an offline rheometer:

[0099] Referring to the mixing process in the mixing chamber, liquid phase (water or water and water-reducing agent) and accelerator were sequentially added to the dry mixture, and rheological parameters were monitored using an offline rheometer. For ease of description, group B1 represents the result of adding water to the dry mixture, group B2 represents the structure of adding water and water-reducing agent to the dry mixture, and group C represents the result of adding accelerator after the mixture with added water and water-reducing agent has reached a stable state. The results are as follows: Figure 6 As shown, Figure 6In Figures (a), (b), and (c), the static yield stress, dynamic yield stress, and plastic viscosity of groups B1, B2, and C, respectively. The fitted curves for static and dynamic yield stresses are completely consistent with the observed results of the instantaneous torque of the online spindle. The cumulative variation of plastic viscosity in each group is less than 1%, which is negligible. Linear regression analysis was used to establish the mapping relationship between static yield stress, dynamic yield stress, and instantaneous spindle torque. Figure 7 To establish the relationship between yield stress and torque, the fitting results show that groups B2 and C have high coefficients of determination above 0.99, indicating that the instantaneous torque of the spindle can be used as a reliable substitute index for real-time yield stress, and real-time monitoring of torque can achieve real-time monitoring of yield stress changes. Furthermore, comparing groups B1 and B2, it is shown that the fitting curves for adding different liquid phases exhibit significantly different slopes, reflecting the sensitivity of this adaptive control system to formula changes, and making it applicable to the production of printed concrete with different formulas.

[0100] Comparative Example

[0101] This comparative example uses conventional methods to prepare printed concrete, and all raw materials are exactly the same as those in the example.

[0102] The traditional preparation method is as follows:

[0103] An open-loop timed stirring control mode with a preset fixed time is adopted. The specific steps are as follows:

[0104] (1) Add cement, fine aggregate, coarse aggregate and hydroxypropyl methylcellulose (HPMC) to the mixing chamber and dry mix at a fixed speed of 22 rpm for 3 minutes; (2) Add the amount of water and polycarboxylate superplasticizer (PCS) in the formula at one time, and continue to mix at a fixed speed for 3 minutes; (3) Finally add aluminum sulfate-based alkali-free quick-setting agent, and after mixing for 3 minutes, regardless of the actual uniformity and rheological state of the material, the system directly issues a discharge command.

[0105] Referring to the embodiment, a printing test was conducted on the comparative sample. The target width of the printed product was 200 mm, the extruder nozzle diameter was 200 mm, and the extruder nozzle moving speed was 150 mm / s. The results showed that although the concrete in the comparative sample could achieve initial pumping and extrusion, the extruded strip exhibited significant shape collapse. In the vertical direction, the average deformation rate of the printed product was as high as 50%, proving that the concrete material obtained using this traditional timing control method failed to establish sufficient initial yield stress and was unable to resist the self-weight of the 3D printed structure's bottom layer. Furthermore, within a 1-meter distance of continuous printing, the strip width of the comparative sample's printed product fluctuated drastically within the range of 200 ± 85 mm, with a maximum width fluctuation rate reaching 42.5%. This further confirms that due to the lack of adaptive monitoring of the material state, the internal component distribution of the concrete material obtained by the traditional method is extremely uneven, resulting in severe instability in rheological properties. Compared to the comparative sample, the deformation rate of the printed product in the embodiment was reduced by 96%, and the width fluctuation rate was reduced by 97.6%, proving that the adaptive control method of this application can effectively improve the rheological consistency and batch stability of concrete.

[0106] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. An adaptive control method for a concrete mixing process, characterized in that: Includes the following steps: S1. Real-time acquisition of visual image sequences of material surface in the mixing chamber, monitoring of material thermal spectrum evolution, and acquisition of instantaneous torque of the main shaft; S2. Feature extraction and physical mapping are performed on the three original signals collected in S1 to obtain the linear mapping equation between visual uniformity, temperature change rate, and instantaneous torque of the spindle and offline yield stress. Visual uniformity is obtained by extracting the ROI from the video frames captured by the camera, removing interference areas, converting the color image to a grayscale image, and calculating the standard deviation of the pixel grayscale histogram; the temperature change rate is obtained by dynamically extracting the highest temperature on the material surface as a representative value, fitting the temperature data within a preset sliding window using a linear least squares regression algorithm, and calculating its slope. The steps for determining the linear mapping equation between the instantaneous torque of the spindle and the offline yield stress, and the target calibration interval, include: acquiring multiple sets of instantaneous spindle torques at different times during concrete mixing and simultaneously sampling and measuring offline yield stress data; establishing a linear mapping equation between the instantaneous torque of the spindle and the offline yield stress using a linear regression algorithm; and, based on the static yield stress threshold requirement for the bottom material to not collapse under the 3D printing process, substituting the theoretical minimum static yield stress threshold required for the bottom material to not collapse into the linear mapping equation to back-calculate the target spindle torque critical value, and combining it with a preset safety redundancy range to determine the target calibration interval. visual uniformity The formula is as follows: ； In the formula, This represents the maximum value of the visual texture index within the test interval. This represents the final stable value of the visual texture index. This refers to the real-time visual texture index; among which, , It was obtained by fitting the test results of the real-time visual texture index; Temperature change rate The formula is: ； In the formula, This represents the temperature increment, indicating the difference between the real-time measured temperature and the ambient temperature, where n is the number of data points in the window. It is all within the window The average value of the data points The average value of the time series. In order to be in Temperature increment over time; S3. Based on the visual uniformity, temperature change rate, and instantaneous spindle torque, adaptive control is performed on the concrete mixing process: using the visual uniformity threshold as a basic constraint, and in conjunction with the temperature change rate as a chemical arbitration signal, the current mixing process stage is autonomously determined; if the visual uniformity reaches 95% or higher, and the temperature change rate... If the flow rate is greater than 0.01 K / s, the dry mixing stage is considered complete, and an instruction is issued to add water premixed with the water-reducing agent; if the visual uniformity reaches 95% or more and the temperature change rate is... A value less than 0.001 K / s indicates the presence of low thermal activity characteristic of the induction period, classifying it as a slow hydration state, and instructing for a one-time injection of the quick-setting agent; if the visual uniformity reaches 95% or more and 0.001 K / s ≤ If the instantaneous torque of the spindle reaches ≤0.01K / s, it is determined to be a rapid hydration state. If the instantaneous torque of the spindle reaches the target calibration range, it is determined to have reached the target rheological window and the unloading signal is activated; otherwise, return to step S1.

2. The adaptive control method for the printing concrete mixing process according to claim 1, characterized in that: In step S1, the sampling frequency is 1Hz, and the three data sampling frequencies are strictly synchronized.

3. A control system applying the adaptive control method for the printing concrete mixing process as described in claim 1, characterized in that: include: Monitoring module: Used to collect multimodal raw signals of vision, thermal sensing and instantaneous torque of the main shaft in real time during the stirring process; The computing center is used to perform feature mapping on the original signal, run a phased logic gate fusion algorithm to identify the process stage, and generate decision instructions for feeding or discharging materials based on the real-time status of the materials. Control unit: Used to receive the decision instructions, drive the feeding execution mechanism or the discharging mechanism to operate, and realize closed-loop control of the mixing process.

4. The control system according to claim 3, characterized in that: The monitoring module includes a 4K industrial camera installed at the center of the top of the mixing chamber, an infrared thermal sensor deployed inside the mixing drum, and a dynamic torque sensor installed on the mixing spindle; a sunshade is provided on the top of the mixing chamber to shield the top of the 4K industrial camera.

Images