Intelligent optimization method and monitoring device for 3D printing concrete forming quality

Abstract

The application relates to the field of 3D printing technology and discloses an intelligent optimization method and a monitoring device for the forming quality of 3D printing concrete. The method comprises the following steps: obtaining a target formula and benchmark printing parameters of concrete; printing concrete according to the target formula and the benchmark printing parameters, and obtaining real-time rheological parameters and morphological images in the concrete printing process; obtaining defect information of the concrete printing according to the real-time rheological parameters, the morphological images and the benchmark printing parameters; determining a target instruction according to the defect information, and controlling a concrete printing device to execute the target instruction. Through high-sensitivity real-time monitoring and accurate hierarchical error correction, the application effectively reduces defects such as uneven extrusion and interlayer cracking, improves the stability and consistency of the printing quality, reduces the printing failure rate, and improves the overall construction efficiency.

Description

Technical Field

[0001] This application relates to the field of 3D printing technology, and in particular to an intelligent optimization method and monitoring device for the molding quality of 3D printed concrete. Background Technology

[0002] With the deep penetration of additive manufacturing (3D printing) technology into the field of construction engineering, 3D printed concrete (3DCP) has been widely used in customized components (such as thin-walled structures and irregular joints), underground engineering support (such as tunnel lining), and emergency construction due to its advantages such as high design freedom, high material utilization, and short construction cycle. However, the 3DCP process has two major drawbacks: First, as a non-Newtonian fluid, concrete's rheological properties (yield stress YS, plastic viscosity PV) dynamically change with mix proportions and ambient temperature, easily leading to defects such as uneven extrusion and interlayer cracking. Second, during the printing process, parameters such as nozzle flow rate, Z-axis offset, and printing speed need to be matched with rheological properties in real time. Traditional methods relying on experience to determine parameters and manual inspection are insufficient to handle complex situations such as printing large components and temperature fluctuations.

[0003] Current 3D concrete printer sensors have significant shortcomings in error detection: First, their detection range is limited, only able to identify large-scale defects such as pipe blockage and material breakage, lacking sensitivity to minute morphological deviations such as insufficient extrusion caused by Z-axis offset as small as 1mm. Second, they lack real-time correction capabilities; the sensors can only detect defects and cannot dynamically adjust printing parameters such as flow rate and nozzle speed, requiring manual shutdown for debugging. Third, they are costly and have poor compatibility; dedicated sensors and amplifiers are expensive and difficult to adapt to different models of 3D concrete printers, resulting in low adoption.

[0004] Current mainstream visual monitoring solutions for printing defects have significant limitations: Single-camera solutions are commonly used, but they are mostly deployed independently without being linked to the printer. The lenses are fixed and easily obstructed by the print head, making it impossible to capture the shape of freshly extruded concrete in real time. This requires pausing printing for adjustments or reshooting, disrupting the construction schedule, reducing efficiency, and potentially causing new defects such as poor interlayer bonding due to the initial setting of the concrete. Multi-camera solutions, while providing multi-view coverage, are costly, prone to errors in data stitching from multiple lenses, and are sensitive to lighting conditions, leading to misjudgments. Structured light scanning solutions can perform 3D reconstruction to detect dimensional deviations, but are limited by scanning speed, resulting in poor real-time performance and detection lag.

[0005] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0006] The main purpose of this application is to provide an intelligent optimization method and monitoring device for the molding quality of 3D printed concrete, which aims to solve the problems of insufficient sensitivity of sensor monitoring and lagging detection results of visual monitoring in the 3D printing process of concrete, resulting in poor printing effect.

[0007] The first aspect of this application provides an intelligent optimization method for the molding quality of 3D printed concrete, the intelligent optimization method for the molding quality of 3D printed concrete includes the following steps:

[0008] Obtain the target concrete formula and baseline printing parameters;

[0009] Concrete is printed according to the target formula and the baseline printing parameters, and real-time rheological parameters and morphological images of the concrete are obtained during the 3D printing process.

[0010] Based on the real-time rheological parameters, the morphological image, and the baseline printing parameters, the defect information of the concrete printing is obtained;

[0011] The target instruction is determined based on the defect information, and the concrete printing equipment is controlled to execute the target instruction.

[0012] Optionally, in one embodiment of this application, the reference printing parameters include the extrusion speed and printing speed of the concrete printing equipment;

[0013] The acquisition of the target concrete formula and benchmark printing parameters specifically includes:

[0014] Obtain the performance requirements and operating conditions for concrete printing;

[0015] Based on the performance requirements and operating conditions, determine the initial concrete formula and corresponding rheological parameters;

[0016] If the rheological parameters meet the performance requirements, the initial formula is taken as the target formula, and the extrusion speed and printing speed of the concrete printing equipment are determined according to the rheological parameters that meet the performance requirements.

[0017] Optionally, in one embodiment of this application, the real-time rheological parameters include real-time yield stress and real-time plastic viscosity;

[0018] The acquisition of real-time rheological parameters and morphological images of concrete during the 3D printing process specifically includes:

[0019] Acquire sensor data from sensors and morphological images from industrial cameras during the 3D printing process of concrete.

[0020] Based on the sensor data, the real-time yield stress and real-time plastic viscosity of the concrete are calculated.

[0021] Optionally, in one embodiment of this application, the sensor data includes torque data;

[0022] The step of calculating the real-time yield stress and real-time plastic viscosity of concrete based on the sensor data specifically includes:

[0023] Based on the torque data and the geometric parameters of the screw, the shear stress borne by the concrete is calculated, and based on the real-time rotational speed and geometric parameters of the screw, the shear rate of the concrete is calculated.

[0024] Based on the shear stress and the shear rate, the initial yield stress and initial plastic viscosity are obtained by linear fitting using the least squares method.

[0025] If the initial yield stress and the initial plastic viscosity pass the verification, the initial yield stress is taken as the real-time yield stress, and the initial plastic viscosity is taken as the real-time plastic viscosity.

[0026] Optionally, in one embodiment of this application, the defect information includes the defect type and the abnormal probability distribution of key parameters;

[0027] The step of obtaining defect information for concrete printing based on the real-time rheological parameters, the morphological image, and the baseline printing parameters specifically includes:

[0028] The morphological image is processed to obtain defect features;

[0029] Based on the defect characteristics, the real-time yield stress, the real-time plastic viscosity, the extrusion speed, and the printing speed, the abnormal probability distribution of defect types and key parameters is obtained.

[0030] Optionally, in one embodiment of this application, processing the morphological image to obtain defect features specifically includes:

[0031] The morphological image is preprocessed to obtain a standardized image;

[0032] The standardized image is input into a trained neural network model, which outputs defect features.

[0033] Optionally, in one embodiment of this application, the abnormal probability distribution includes a first probability distribution of flow velocity, a second probability distribution of Z offset, and a third probability distribution of nozzle velocity;

[0034] The process of obtaining the abnormal probability distribution of defect types and key parameters based on the defect characteristics, the real-time yield stress, the real-time plastic viscosity, the extrusion speed, and the printing speed specifically includes:

[0035] A classification probability vector is generated based on the defect features, and the defect type is determined based on the classification probability vector.

[0036] The defect features, real-time yield stress, real-time plastic viscosity, extrusion speed, and printing speed are combined to obtain a comprehensive feature vector;

[0037] A first probability distribution of flow velocity, a second probability distribution of Z offset, and a third probability distribution of nozzle velocity are generated based on the comprehensive feature vector.

[0038] Optionally, in one embodiment of this application, the target instruction includes a first instruction from the screw servo motor, a second instruction from the vertical axis servo motor, and a third instruction from the horizontal motion system servo motor;

[0039] The step of determining the target instruction based on the defect information specifically includes:

[0040] Based on the defect type, the first probability distribution, the second probability distribution, and the third probability distribution, determine the anomaly classification corresponding to each key parameter;

[0041] Based on the anomaly classification corresponding to each key parameter, the first command of the screw servo motor, the second command of the vertical axis servo motor, and the third command of the horizontal motion system servo motor are obtained.

[0042] The second aspect of this application also provides a monitoring device for implementing the intelligent optimization method for the molding quality of 3D printed concrete as described in any of the above solutions, wherein the monitoring device is installed on a concrete printing device and includes a camera bracket, a scanner bracket, a laser scanner, an industrial camera, and a sensor assembly.

[0043] The camera bracket is mounted on the print head of the concrete printing equipment, the scanner bracket is connected to the camera bracket, the laser scanner is connected to the scanner bracket, the industrial camera is connected to the camera bracket, and the sensor assembly is connected to a component of the print head.

[0044] The industrial camera is used to acquire morphological images of concrete during the 3D printing process, the laser scanner is used to acquire the real-time offset of the printing nozzle in the Z-axis direction, and the sensor assembly is used to acquire the real-time rheological parameters of concrete during the 3D printing process.

[0045] Optionally, in one embodiment of this application, the sensor assembly includes a fixed base, a torque sensor, and a pressure sensor;

[0046] The torque sensor is connected to the motor shaft and the screw rod, and the torque sensor is connected to the fixed base, which is connected to the bottom of the motor; the pressure sensor is connected to the bottom of the print head.

[0047] The torque sensor is used to measure the torque that the screw needs to overcome to rotate, and the pressure sensor is used to measure the positive pressure generated when concrete is propelled forward in the pipe.

[0048] Beneficial Effects: This application provides an intelligent optimization method and monitoring device for the molding quality of 3D printed concrete. Before printing, the concrete formula is determined, and the rheological performance parameters of the new formula and corresponding printing parameters are output. During concrete pumping, several parameters are obtained through sensors to calculate the rheological performance data. During printing, data is collected through an industrial camera and scanner, and the probability of abnormal printing parameters is output. Finally, printing accuracy is ensured through three-level graded correction (light, medium, and heavy). This invention effectively reduces defects such as uneven extrusion and interlayer cracking through high-sensitivity real-time monitoring and precise graded error correction, improving the stability and consistency of printing quality. It also reduces printing interruptions, rework, and material waste caused by defects and quality problems, lowering the printing failure rate and improving overall construction efficiency. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 A flowchart illustrating a preferred embodiment of the intelligent optimization method for the molding quality of 3D printed concrete according to this application;

[0051] Figure 2 This is a flowchart illustrating step S101 in a preferred embodiment of the intelligent optimization method for the molding quality of 3D printed concrete according to this application.

[0052] Figure 3 This is a structural diagram of the sensor installation in a preferred embodiment of the monitoring device of this application;

[0053] Figure 4 This is a structural diagram showing the connection of the torque sensor in a preferred embodiment of the monitoring device of this application;

[0054] Figure 5 This is a structural diagram of the flat model pressure sensor in a preferred embodiment of the monitoring device of this application;

[0055] Figure 6This is a structural diagram of the flange nut seat in a preferred embodiment of the monitoring device of this application;

[0056] Figure 7 This is a structural diagram of an industrial camera mounted on a printhead in a preferred embodiment of the monitoring device of this application;

[0057] Figure 8 This is a flowchart illustrating step S103 in a preferred embodiment of the intelligent optimization method for the molding quality of 3D printed concrete according to this application.

[0058] Figure 9 This is a flowchart illustrating step S104 in a preferred embodiment of the intelligent optimization method for the molding quality of 3D printed concrete according to this application.

[0059] Explanation of reference numerals in the attached figures:

[0060] 11. Camera mount; 12. Scanner mount; 13. Laser scanner; 14. Industrial camera;

[0061] 15. Fixed base; 16. Torque sensor; 17. Pressure sensor; 18. Nut flange; 19. First double thin-plate coupling; 20. Second double thin-plate coupling; 21. Motor shaft end; 22. Helical rod top;

[0062] 23. Sensor flange; 24. Bolt hole; 25. Sensor diaphragm end nut; 26. Internal thread nut; 27. Flange section; 28. Welded end.

[0063] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0064] To make the objectives, technical solutions, and effects of this application clearer and more explicit, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. The described embodiments are only possible technical implementations of this application and not all possible implementations. Based on the embodiments in this application, those skilled in the art can obtain other embodiments without creative effort, and these embodiments are also within the protection scope of this application.

[0065] First, let's introduce the terms used in the embodiments of this application:

[0066] 3DCP, 3D Concrete Printing;

[0067] PICNN, Physics-Informed Convolutional Neural Network, is the PICNN model;

[0068] MHN, Multi-Head Neural Network, also known as the MHN model;

[0069] CNN stands for Convolutional Neural Network, also known as the CNN model.

[0070] RF stands for Random Forest, which is the RF model.

[0071] YS, Yield Stress, is the yield stress (YS).

[0072] PV, Plastic Viscosity.

[0073] ES, Extrusion Speed;

[0074] PS stands for Printing Speed.

[0075] W / B, Water-Binder Ratio, water-binder ratio (W / B);

[0076] S / B, Sand-Binder Ratio.

[0077] PC, Portland Cement;

[0078] GGBFS, Ground Granulated Blast Furnace Slag, is granulated blast furnace slag powder, also known as slag powder (GGBFS).

[0079] MK, Metakaolin, is a type of kaolin clay.

[0080] SP, Superplasticizer, High-efficiency water-reducing agent (SP);

[0081] SR, Set Retarder;

[0082] TH stands for Thickener.

[0083] PN, Printing Nozzle Size, is the nozzle size (PN).

[0084] PH, Printing Layer Height, is the height of the printed layer (PH).

[0085] Q, Flow Rate, flow (Q);

[0086] τ, Shear Stress;

[0087] Shear Rate );

[0088] R², Coefficient of Determination;

[0089] Profinet, Process Field Net, is a process fieldbus protocol.

[0090] Pa, Pascal, is a unit of pressure.

[0091] Pa·s, Pascal·second, is a unit of viscosity.

[0092] RS485, Recommended Standard 485, refers to the RS485 interface.

[0093] In related technologies, the accuracy of rheological property prediction is low; the adaptability of printing parameters to material formulations and real-time operating conditions is poor. Existing monitoring methods are not sensitive to minute defects and can only detect problems, not automatically solve them; error correction relies on manual intervention and is severely delayed. There is a lack of coordination and intelligent decision-making between material formulations, rheological states, process parameters, and equipment actions. Dedicated systems are costly, difficult to adapt to different printer models and diverse materials and operating conditions, and rely on human experience intervention.

[0094] To address the problems of low rheological property prediction accuracy, poor adaptability of printing parameters, lag in process monitoring and error correction, insufficient multi-parameter collaborative control, weak versatility and robustness, and reliance on manual intervention in existing 3D printed concrete technologies, this invention proposes an integrated intelligent control solution covering the entire 3D printed concrete chain. This solution comprises four core modules covering the entire concrete printing process: Before printing, a PICNN is used to iteratively generate a concrete formula that meets the current printing requirements and environment, and outputs the rheological property parameters and corresponding printing parameters of the new formula; during concrete pumping, several parameters are obtained through sensors, and rheological property data is calculated from this data; during printing, data is collected through industrial cameras and scanners, processed by CNN, and correlated with parameters by MHN to output the probability of abnormal printing parameters; finally, three-level graded correction (light, medium, and heavy) ensures printing accuracy.

[0095] This invention proposes an intelligent control technology solution that integrates four modules: formula fine-tuning, rheological monitoring, defect prediction, and hierarchical error correction. It utilizes a Physical Information Fusion Model (PICNN) and a Multi-Head Neural Network (MHN) as its core algorithms, supported by multi-sensor collaborative perception. This achieves precise matching of the formula to requirements before printing, real-time monitoring of the rheological state during printing, accurate location of defects and parameter anomalies, and subsequent hierarchical error correction based on probability. Simultaneously, the accumulation of data during the printing process provides targets for optimizing the formula for the next print run, forming a closed loop of printing, feedback, and optimization.

[0096] The first module is pre-printing formula adaptation and fine-tuning. After iterating through the Physical Information Convolutional Neural Network (PICNN) to obtain the suitable formula, it predicts rheological parameters such as concrete yield strength (YS) and plastic viscosity (PV), and obtains the optimal extrusion speed (ES) by combining random forest (RF). Finally, it calculates the optimal printing speed through the printing speed (PS) equation to achieve precise adaptation between the formula and printing requirements.

[0097] Module 2 is the rheological monitoring function during the conveying process. By installing torque and pressure sensors in the pumping system, operating parameters are obtained and the rheological properties of concrete are calculated in reverse, so as to monitor the material conveying status (rheological properties) in real time.

[0098] Module 3 is the printing process prediction function. It uses a laser scanner and an industrial camera to collect concrete shape data. After preprocessing by the image processing unit, it extracts features and judges shape defects through a convolutional neural network (CNN). Then, a multi-head neural network (MHN) associates parameters such as flow rate, Z offset, nozzle speed, and additive ratio, and outputs parameter level probabilities.

[0099] Module four is graded error correction, which combines sensing data and verification results to automatically perform graded processing of mild errors (no adjustment), moderate errors (automatic correction), and severe errors (manual alarm) to ensure the accuracy of error correction.

[0100] This invention enables intelligent control of the entire process of 3D printed concrete, from formula adaptation and process monitoring to real-time error correction. It improves the convenience of rheological prediction, parameter adaptability and printing quality stability, enhances the versatility for different printing needs, working conditions and materials, reduces the cost of manual intervention, effectively reduces material waste and printing failure rate, and improves the engineering applicability and reliability of 3D printed concrete technology.

[0101] The technical solutions of this application will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0102] The intelligent optimization method for the molding quality of 3D printed concrete described in the preferred embodiment of this application, such as... Figure 1 As shown, the intelligent optimization method for the molding quality of 3D printed concrete includes the following steps:

[0103] In step S101, the target concrete formula and benchmark printing parameters are obtained.

[0104] In one possible implementation, the baseline printing parameters include the extrusion speed and printing speed of the concrete printing equipment.

[0105] Step S101 specifically includes: obtaining the performance requirements and operating conditions for concrete printing; determining the initial concrete formula and corresponding rheological parameters based on the performance requirements and operating conditions; if the rheological parameters meet the performance requirements, then using the initial formula as the target formula, and determining the extrusion speed and printing speed of the concrete printing equipment based on the rheological parameters that meet the performance requirements.

[0106] Specifically, in step S101, the formula before printing is fine-tuned to form a concrete formula that adapts to past data and new printing requirements, as well as printing data baseline parameters.

[0107] Furthermore, such as Figure 2 As shown, step S101 specifically includes:

[0108] Step S11: Input printing requirements and operating conditions: Rheological performance requirements for the component (e.g., large components require a yield stress YS = 200Pa-400Pa, thin-walled components require a plastic viscosity PV = 1Pa·s-2Pa·s). Strength requirements for the component, ensuring it can stably withstand daily use loads and the effects of outdoor temperature and humidity changes, as well as damp environments (specifically, it must meet a cubic compressive strength ≥ 40MPa after 28 days of standard curing). External operating conditions during component printing (ambient temperature and humidity).

[0109] Step S12: Retrieve concrete formulas (including Portland cement (PC), granulated blast furnace slag powder (GGBFS), metakaolin (MK), water-cement ratio (W / B), sand-cement ratio (S / B), high-efficiency water-reducing agent (SP), retarder (SR), thickener (TH)) and printing parameters from the database (including literature collection data and equipment historical printing formulas).

[0110] Input the above formula data into a physical information convolutional neural network:

[0111] PICNN, this model uses rheology-related physical information equations (PIEs, such as...) YS=K 0 +K 1 (S / B)-K 2 ·(W / B)+ K 3 (GGBFS / PC)+K 4 (MK / PC)-K 5 (SP / PC (Embedding the CNN loss function to constrain the prediction results to conform to physical laws).

[0112] The iterative formulation and its corresponding rheological properties are then output. If YS and PV deviate from the target range, the formulation is adjusted (e.g., the water-cement ratio or the content of water-reducing agent) and the prediction is repeated until the target is met.

[0113] Step S13: Input the YS and PV predicted by PICNN into the Random Forest (RF) model, then output the optimal extrusion speed (ES), and based on the target flow rate (Q), print nozzle size (PN), and target printed layer height (PH), use the formula: ;in, PS Printing speed; Q :flow; PN Print nozzle size; PH : Print layer height; Calculate printing speed ( PS ).

[0114] In this invention, requirements and verification are introduced before printing to ensure that the starting state is optimized. Qualitative experience is transformed into quantitative prediction, using PICNN combined with physical equations to replace manual experience, and RF models to replace manual experience values, making the decision-making process quantifiable, replicable, and optimizable. It is understood that the ES and PS output in step S101 are planned values ​​used for comparison during subsequent real-time monitoring and diagnosis. All subsequent corrections are based on confirming the deviation of the real-time state from the initial optimal plan.

[0115] In step S102, concrete is printed according to the target formula and the benchmark printing parameters, and real-time rheological parameters and morphological images of the concrete during the printing process are obtained.

[0116] In one possible implementation, the real-time rheological parameters include real-time yield stress and real-time plastic viscosity.

[0117] Step S102 specifically includes: acquiring sensor data from the sensor and morphological images from the industrial camera during the concrete printing process; and calculating the real-time yield stress and real-time plastic viscosity of the concrete based on the sensor data.

[0118] In one possible implementation, the sensor data includes torque data.

[0119] The specific implementation of the steps for calculating real-time yield stress and real-time plastic viscosity is as follows: Based on the torque data and the geometric parameters of the screw rod, the shear stress borne by the concrete is calculated, and based on the real-time rotational speed and geometric parameters of the screw rod, the shear rate of the concrete is calculated; based on the shear stress and the shear rate, a linear fit is performed using the least squares method to obtain the initial yield stress and the initial plastic viscosity; if the initial yield stress and the initial plastic viscosity pass the verification, the initial yield stress is taken as the real-time yield stress, and the initial plastic viscosity is taken as the real-time plastic viscosity.

[0120] Furthermore, step S102 specifically includes:

[0121] Step S21: Real-time monitoring of the rheological properties of concrete during the printing process.

[0122] like Figure 3 , Figure 4 , Figure 5 and Figure 6 As shown, this application employs several sensors to monitor 3D-printed concrete in real time. The sensors include: a torque sensor 16 (which directly reflects the shear stress of the concrete and provides core data for calculating the yield stress) and a pressure sensor 17 (which assists in calculating the shear stress distribution and verifies the validity of the data from the torque sensor 16).

[0123] See Figure 3 and Figure 4The torque sensor 16 is connected to the motor shaft and the screw rod via an aluminum alloy double-thin-plate coupling. The specific connection sequence from top to bottom is: motor shaft end 21, first double-thin-plate coupling 19, dynamic torque sensor 16, second double-thin-plate coupling 20, and screw rod top 22. The connection method involves using a double-thin-plate coupling with customized shaft holes (based on the diameters of the motor shaft, sensor shaft, and screw rod shaft), and securing the upper and lower shafts of the coupling with M5 set screws (the first double-thin-plate coupling 19 connects the motor shaft and sensor shaft, and the second double-thin-plate coupling 20 connects the sensor shaft and screw rod shaft). Since the screw rod needs to rotate continuously, the dynamic torque sensor 16 needs to be fixed using a dedicated mounting base 15. This mounting base 15 connects to the bottom of the motor and the sensor body, ensuring that the sensor does not rotate synchronously with the screw rod. The sensor's data transmission cable is also fixed to the base, effectively preventing wire tangling during motor shaft operation.

[0124] See Figure 5 and Figure 6 A flat diaphragm pressure sensor 17 is selected to monitor the pressure in the concrete extrusion section. This sensor is connected to the pipe wall of the extrusion section via a flange nut seat. The flange nut seat consists of two parts: a nut section and a flange section 27. Bolt holes 24 are provided on the sensor flange 23, and one end of the sensor flange 23 is a sensor diaphragm end nut 25. An internal thread nut 26 is provided on the inner side of the flange section 27, and a welding end 28 is provided on the other end of the flange section 27. Specific installation steps: First, drill a 38mm diameter hole (the exact diameter depends on the actual situation) 50mm from the end of the screw rod in the extrusion section pipe. Then deburr the hole opening and grind the surface of the welding area to ensure the stability of the subsequent connection. Second, insert the flat end of the flange nut seat into the hole, aligning the pipe opening with the inner wall of the pipe. It is strictly forbidden to protrude from the pipe surface to avoid material blockage. Third, symmetrically spot weld the outer side of the connection between the nut and the pipe wall, then use gas shielded welding to continuously fillet weld around the perimeter, with a weld leg height of approximately 4mm. Fourth, apply a small amount of medium-strength threadlocker to the external threads of the sensor and slowly screw it into the threaded pipe of the flange nut seat until the sensor flange contacts the flange of the flange nut seat. At this point, the end face of the sensor diaphragm should theoretically be flush with the thin wall of the pipe. Finally, insert the six hexagonal bolts into the six bolt holes of the flange nut seat flange in sequence, and tighten them diagonally alternately after using spring washers to ensure that the flange surface is evenly pressed and without warping.

[0125] In another embodiment of this application, a waterproof and dustproof temperature sensor is installed on a structure such as an industrial camera bracket base. Its signal transmission is via a twisted-pair cable, sharing the same metal corrugated conduit as the gigabit network cable of the industrial camera and scanner. The sensor is laid along the same path inside the frame column, secured side-by-side with nylon cable ties every 20cm, and then introduced into the computing unit.

[0126] Step S22: To ensure the effectiveness of subsequent rheological performance calculations, it is crucial that each set of parameters is acquired at the same time during the multi-parameter synchronous acquisition process. Therefore, a synchronizer is integrated into the computing unit, outputting a trigger pulse every 400ms (2.5Hz) as the reference time source for all sensor acquisitions, avoiding data misalignment due to sampling time differences. After receiving the trigger pulse from the synchronizer, the sensor system simultaneously sends acquisition commands to the three sensors. The torque sensor reads the real-time torque value through the RS485 interface and records the timestamp of the data acquisition moment (accurate to milliseconds). The pressure sensor reads the extrusion section positive pressure through the acquisition module and synchronously binds the timestamp. The temperature sensor records synchronously along with the above parameters, with each set of data bound to the same timestamp.

[0127] Step S23: The computing unit (Raspberry Pi) receives the raw monitoring data transmitted by the sensor at 2.5Hz, and combines it with the screw rod structural parameters (diameter D, effective length L), converting the raw data into the core parameters required for rheological performance using physical formulas: shear stress ( The torque (T) collected by the torque sensor is calculated using the following formula: ;in, Shear stress; Torque; : Effective length of the screw rod; : Screw diameter; Shear rate is calculated based on screw rotation speed (n, unit: r / min) using the following formula: ;in, Shear rate; n Screw speed; D : Screw diameter; Ten data points are continuously stored to form ( , ) sequence, in Bingham model ( τ=YS+PV· Perform least squares fitting. PV The slope of the corresponding linear relationship, YS The corresponding linear intercept. From the formula: ; ;in, Plastic viscosity (unit: Pa·s) is a parameter describing the internal viscous resistance of 3D printed concrete under shear flow conditions. Yield stress is the minimum stress (unit: Pa) that 3D printed concrete must overcome to change from a static state to a flowing state. : refers to the i Shear rate measurements at each experimental data point (unit: s) - ¹); : refers to the iShear stress measurements (unit: Pa) at each experimental data point; plastic viscosity PV and yield stress YS were calculated, with results refreshed every 2.5 Hz.

[0128] If the shear stress is calculated based on the torque If the pressure increases but the pressure data does not increase synchronously, it may be due to a mechanical failure of the screw rod (such as bearing wear) rather than a change in the rheological properties of the concrete, thus avoiding misdiagnosis. Changes in normal pressure and shear force are positively correlated. Only when the shear resistance changes synchronously with the normal pressure, and - The regression correlation coefficient R of the curve 2 A value ≥0.85 indicates normal operation, and the newly calculated YS and PV are then output downstream.

[0129] Step S24: Provide early warning feedback for abnormal rheological data. If YS > 400 Pa or PV > 5 Pa·s, an early warning is triggered. If the data is within the normal range, the calculation unit synchronizes the real-time YS, PV data and related data to Module3 at a frequency of 2.5 Hz (step S103 Real-time prediction and defect identification).

[0130] This invention eliminates the need to wait for laboratory results, obtaining rheological data directly during the printing process, enabling real-time control. Through cross-validation of torque and pressure, it can intelligently distinguish between problems with the material itself and malfunctions in the conveying equipment, significantly improving the reliability of system diagnosis and preventing false alarms. The real-time YS and PV output in step S102 are key evidence for the MHN model's defect root cause analysis in step S103. For example, the MHN model learns that a bulge (visual defect) has appeared, and simultaneously detects an abnormal increase in PV (internal evidence), determining that it is highly likely caused by excessive flow rate, thus making a more accurate diagnosis.

[0131] In step S103, defect information of concrete printing is obtained based on the real-time rheological parameters, the morphological image, and the reference printing parameters.

[0132] In one possible implementation, the defect information includes the defect type and the abnormal probability distribution of key parameters.

[0133] Step S103 specifically includes: processing the morphological image to obtain defect features; and obtaining the abnormal probability distribution of defect type and key parameters based on the defect features, the real-time yield stress, the real-time plastic viscosity, the extrusion speed, and the printing speed.

[0134] Specifically, the step of determining defect features involves: preprocessing the morphological image to obtain a standardized image; inputting the standardized image into a trained neural network model to output defect features.

[0135] In one possible implementation, the anomaly probability distribution includes a first probability distribution of flow velocity, a second probability distribution of Z-offset, and a third probability distribution of nozzle velocity.

[0136] The specific implementation of the step of determining the defect type and the abnormal probability distribution of key parameters is as follows: generating a classification probability vector based on the defect features, and determining the defect type based on the classification probability vector; concatenating the defect features, the real-time yield stress, the real-time plastic viscosity, the extrusion speed and the printing speed to obtain a comprehensive feature vector; and generating a first probability distribution of flow velocity, a second probability distribution of Z offset and a third probability distribution of nozzle speed based on the comprehensive feature vector.

[0137] Specifically, in step S103, real-time prediction and defect identification (Module 3).

[0138] Furthermore, step S103 specifically includes:

[0139] Step S31: The equipment's vision system employs an industrial camera and a laser scanner. The scanner acquires nozzle Z-direction offset data at a frequency of 2.5Hz, while the industrial camera captures 2D surface (horizontal) morphological images of the concrete at a frequency of 2.5Hz, recording defects such as line breaks (length > 2mm), bulges (width > 1.2 × design value), and overflow. Figure 7 As shown, the industrial camera 14 is mounted directly behind the printing direction via a bracket. The camera is placed on an adjustable bracket 200mm from the axis of the auger, with the lens axis forming a 60° angle with the surface of the printed layer. The center of the lens faces 1cm behind the nozzle extrusion, capturing the shape of the freshly extruded concrete lines. The laser scanner 13 is mounted on the industrial camera bracket 11 via a custom bracket assembly. This assembly consists of a bracket rod vertically connected to the camera bracket 11 rod via a cross-shaped clamp. The other end of the bracket is connected to the laser scanner 13 via a vertical mounting base 15. After installation, the scanner angle and height can be adjusted via the cross-shaped clamp. The scanner's position (left or right in the printing direction) can also be changed by adjusting the cross-shaped clamp as needed. Furthermore, the scanner's pitch and rotation angles can be fine-tuned by adjusting the clamp and scanner base, ultimately calibrating the scanner lens axis to a 60° angle with the surface of the printed layer.

[0140] Step 32: As Figure 8As shown, image data received from an industrial camera is preprocessed using a Raspberry Pi: a 320×320 pixel focus area is cropped based on the nozzle pixel coordinates; RGB channels are normalized (using the pixel mean / standard deviation of the Step 1 recipe dataset); and dust noise is removed. The preprocessed 224×224 pixel image (224×224 pixels is the standard size for inputting a CNN model, balancing image feature preservation and computational efficiency, which is an industry consensus) is synchronized to the computing unit.

[0141] Step S33: CNN extracts features and determines the defect type. The preprocessed image is input using a CNN with a ResNet18 architecture. Local features (such as the texture density of bulges and the edge contours of broken edges) are extracted through 3×3 convolution kernels, and the defect type (bulge, broken lines, overflow, no defect) is output.

[0142] Step S34: Correlate defects with parameters through MHN and output anomaly probabilities. MHN receives CNN defect feature data, rheological parameters (YS, PV) collected in Module 2, and printing baseline parameters from Module 1. MHN first concatenates the four-dimensional defect encoding (bulge, line break, overflow, no defect) output by CNN, the two-dimensional rheological parameters (YS, PV) processed by Module 2, and the two-dimensional printing baseline parameters (ES, PS) from Module 1 into an 8-dimensional input vector. Then, through a shared feature extraction layer consisting of two fully connected layers (hidden unit number 64→32), nonlinear fusion encoding is performed on this vector to autonomously learn the specific correlation rules between defects, rheology, and equipment parameters (such as the potential feature that "bulge and high PV" correspond to "excessive flow rate"), and outputs a 32-dimensional fusion feature vector rich in global correlation information. Next, three structurally identical but parameter-independent output heads (flow velocity, z-offset, and nozzle movement speed) receive the fused feature. Through a fully connected layer and a Softmax activation function, they calculate and output the low, normal, and high probability distributions for flow velocity, z-offset, and nozzle speed (the sum of the three probabilities is 1). Finally, the defect type and parameter anomaly probabilities are packaged into a standardized format (JSON format: lightweight, easy to parse, and multi-platform compatible) and synchronized to Module4 at a frequency of 2.5Hz.

[0143] This invention combines real-time material state (YS, PV) with original design goals (ES, PS), making the diagnostic results more comprehensive and reliable. For example, for the same "bulge" defect, MHN may diagnose it as "excessive Z-axis offset" and "excessive flow rate" when PV is normal and when PV is high, respectively. The output of this step is a probability rather than a hard judgment, which is more in line with the uncertainty of the real world. Different levels of measures can be flexibly taken according to the probability threshold (step S104), avoiding system overreaction or jitter caused by minor model misjudgments.

[0144] In step S104, a target instruction is determined based on the defect information, and the concrete printing equipment is controlled to execute the target instruction.

[0145] In one possible implementation, the target instructions include a first instruction from the screw servo motor, a second instruction from the vertical axis servo motor, and a third instruction from the horizontal motion system servo motor.

[0146] Step S104 specifically includes: determining the anomaly classification corresponding to each key parameter based on the defect type, the first probability distribution, the second probability distribution, and the third probability distribution; and obtaining the first command of the screw servo motor, the second command of the vertical axis servo motor, and the third command of the horizontal motion system servo motor based on the anomaly classification corresponding to each key parameter.

[0147] Specifically, step S104 involves graded correction.

[0148] Furthermore, step S104 specifically includes:

[0149] like Figure 9 As shown in step S41: After MHN outputs the probability of each state parameter, it will classify the probability of each state to refine the execution of the operation. When the probability of a parameter being "normal" is ≥70%, it is determined to be in a normal state, and the system maintains the real-time parameter and continuously monitors it at a frequency of 2.5Hz. If the probability of "normal" is <70%, it will be further classified according to the highest abnormal probability of the parameter being "high" or "low" (taking the maximum of the two): the abnormal probability between 30% and 49% is a slight abnormality, and the first fine adjustment is made with a fixed step size (Z-axis offset ±2%, printing speed ±4%, flow rate ±4%). The abnormal probability between 50% and 79% is a moderate abnormality, and the adjustment range is 1.5 times that of the previous probability range (30%-49%) (Z-axis offset ±3%, printing speed ±6%, flow rate ±6%). The abnormal probability ≥80% is a serious abnormality, and the shutdown procedure will be initiated immediately and an audible and visual alarm will be triggered to notify manual intervention.

[0150] For example, when the flow velocity is determined to be too high, reaching 108%, the system will adjust the rotational speed of the screw to reduce the flow velocity by 4% to 104%. The reason for adjusting the flow velocity by changing the rotational speed is based on the formula relating rotational speed to flow velocity under actual conditions, taking rheological characteristics into account:

[0151] ;in, : Actual volumetric flow rate per unit time (m³ / s); D : Diameter of the screw rod (m); P : Helical pitch (m); η Filling efficiency (0.7-0.9); N : Screw speed (r / min); YS Material yield stress (Pa); PV Material plastic viscosity (Pa·s); , : Rheological correction coefficients (experimentally calibrated, units are m³ / (s·Pa) and m³ / (s·Pa·s), respectively), reflecting YS , PV (The extent of the impact on traffic loss).

[0152] It is known that the rotational speed of the screw directly affects the flow rate and rheological properties of concrete during operation. Therefore, after MHN judges that the parameters are abnormal, the flow rate is adjusted by rotating the screw.

[0153] Step S42: After the grading results are determined, the central control unit (Raspberry Pi logic controller) automatically generates specific adjustment instructions (such as reducing the flow rate by 2% and increasing the Z offset by 0.03mm) according to preset rules, and converts them into control signals in the Profinet protocol.

[0154] Signals are distributed to the corresponding actuators via an industrial bus. The screw motor receives flow rate adjustment commands, and the robotic arm's vertical axis servo motor responds to Z-offset commands to precisely adjust the nozzle height. The horizontal drive system receives nozzle speed commands and changes its movement rate.

[0155] This invention balances automation and safety through a three-level response mechanism: automatic fine-tuning for minor anomalies handles daily fluctuations, demonstrating intelligence; enhanced adjustment for moderate anomalies addresses clear faults, demonstrating decisiveness; and shutdown alarm for severe anomalies handles high-risk situations, demonstrating absolute safety—the bottom line for industrial system design. This invention employs fixed step sizes and probability-based amplitude grading to avoid repeated oscillations between over-adjustment and correction caused by excessively large single adjustments, ensuring a smooth control process.

[0156] Following step S104, to provide data support for subsequent iterative optimization of the printing recipe, the printing data should be archived and organized. The system will organize the rheological monitoring data (YS, PV change curves over time) from step S102, the defect parameter correlation records from step S103, and the correction logs from step S104. The archived data will then be added to the PICNN training set to correct the rheological parameter prediction model, providing a clear direction for the next recipe adjustment.

[0157] It is understood that this invention employs a combination of the "PICNN model" and "physics equation embedded in CNN" logic, but is not limited to this; a novel architecture embedding the physical equation into "LSTM" can also be used. In another embodiment of this invention, a physical scraper plate can be installed, with a pressure sensor mounted on the scraper plate. Based on the data fed back by the sensor, it can be determined whether there is a deviation in the printing layer, thereby achieving deviation correction.

[0158] Based on the above embodiments, this application also provides a monitoring device for implementing the intelligent optimization method for the molding quality of 3D printed concrete described in the above scheme, such as... Figures 3-7 As shown, the monitoring device is installed on the concrete printing equipment, and the monitoring device includes a camera bracket 11, a scanner bracket 12, a laser scanner 13, an industrial camera 14, and a sensor assembly;

[0159] The camera bracket 11 is mounted on the print head of the concrete printing equipment, the scanner bracket 12 is connected to the camera bracket 11, the laser scanner 13 is connected to the scanner bracket 12, the industrial camera 14 is connected to the camera bracket 11, and the sensor assembly is connected to a component of the print head.

[0160] The industrial camera is used to acquire morphological images of concrete during the 3D printing process, the laser scanner is used to acquire the real-time offset of the printing nozzle in the Z-axis direction, and the sensor assembly is used to acquire the real-time rheological parameters of concrete during the 3D printing process.

[0161] In one embodiment of this application, the sensor assembly includes a fixed base 15, a torque sensor 16, and a pressure sensor 17;

[0162] The torque sensor 16 is connected to the motor shaft and the screw rod, and the torque sensor 16 is connected to the fixed base 15, which is connected to the bottom of the motor; the pressure sensor 17 is connected to the bottom of the print head through the nut flange 18.

[0163] The torque sensor is used to measure the torque that the screw needs to overcome to rotate, and the pressure sensor is used to measure the positive pressure generated when concrete is propelled forward in the pipe.

[0164] This invention embeds torque and pressure sensors in the printing equipment, and the computing unit obtains real-time concrete rheological performance data. This sensor monitoring data and rheological performance data are used both to predict printing parameter deviations in the MHN model and as reference values ​​to feed back into the PICNN model for optimizing formulations and predicting material physical properties.

[0165] This invention deeply integrates the PICNN and MHN models to form a dual-model collaborative verification mechanism, avoiding prediction bias caused by a single model. Simultaneously, this solution broadens the application of the MHN model from thermoplastic materials to concrete, deeply linking printing parameters, printing morphology, and concrete rheological properties, thereby achieving collaborative optimization and high adaptability.

[0166] This invention integrates analyzed printing defects (images captured by laser scanners and industrial cameras), baseline values ​​of printing parameters, and real-time rheological data collected by sensors. It then uses the MHN model to correlate these data and derive the probability of abnormal printing parameters. Subsequently, based on this probability, correction levels are assigned to achieve tiered correction.

[0167] This invention archives all data from the entire 3D concrete printing process (rheological monitoring curves, defect and parameter correlation records, and correction logs) and supplements them into the training set of the "PICNN model." This provides iterative and timely data support for printing data optimization and concrete formulation improvement.

[0168] The monitoring device provided in this application is used to realize the intelligent optimization method for the molding quality of 3D printed concrete described in the above scheme, thereby having all the above-mentioned beneficial effects, which will not be repeated here.

[0169] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0170] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0171] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0172] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable storage medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable storage medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable storage medium could be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0173] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0174] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0175] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0176] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

[0177] It should be understood that the application of this application is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

[0178] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for intelligently optimizing the molding quality of 3D printed concrete, characterized in that, The intelligent optimization method for the molding quality of 3D printed concrete includes: Obtain the target concrete formula and baseline printing parameters; Concrete is printed according to the target formula and the baseline printing parameters, and real-time rheological parameters and morphological images of the concrete are obtained during the 3D printing process. Based on the real-time rheological parameters, the morphological image, and the baseline printing parameters, the defect information of the concrete printing is obtained; The target instruction is determined based on the defect information, and the concrete printing equipment is controlled to execute the target instruction. The baseline printing parameters include the extrusion speed and printing speed of the concrete printing equipment; The acquisition of the target concrete formula and benchmark printing parameters specifically includes: Obtain the performance requirements and operating conditions for concrete printing; Based on the performance requirements and operating conditions, determine the initial concrete formula and corresponding rheological parameters; If the rheological parameters meet the performance requirements, the initial formula is taken as the target formula, and the extrusion speed and printing speed of the concrete printing equipment are determined according to the rheological parameters that meet the performance requirements. The real-time rheological parameters include real-time yield stress and real-time plastic viscosity; The acquisition of real-time rheological parameters and morphological images of concrete during the 3D printing process specifically includes: Acquire sensor data from sensors and morphological images from industrial cameras during the 3D printing process of concrete. Based on the sensor data, the real-time yield stress and real-time plastic viscosity of the concrete are calculated. The defect information includes the defect type and the abnormal probability distribution of key parameters; The step of obtaining defect information for concrete printing based on the real-time rheological parameters, the morphological image, and the baseline printing parameters specifically includes: The morphological image is processed to obtain defect features; Based on the defect characteristics, the real-time yield stress, the real-time plastic viscosity, the extrusion speed, and the printing speed, the abnormal probability distribution of defect types and key parameters is obtained; The abnormal probability distribution includes a first probability distribution of flow velocity, a second probability distribution of Z-offset, and a third probability distribution of nozzle velocity; The process of obtaining the abnormal probability distribution of defect types and key parameters based on the defect characteristics, the real-time yield stress, the real-time plastic viscosity, the extrusion speed, and the printing speed specifically includes: A classification probability vector is generated based on the defect features, and the defect type is determined based on the classification probability vector. The defect features, real-time yield stress, real-time plastic viscosity, extrusion speed, and printing speed are combined to obtain a comprehensive feature vector; A first probability distribution of flow velocity, a second probability distribution of Z offset, and a third probability distribution of nozzle velocity are generated based on the comprehensive feature vector.

2. The intelligent optimization method for the molding quality of 3D printed concrete according to claim 1, characterized in that, The sensor data includes torque data; The step of calculating the real-time yield stress and real-time plastic viscosity of concrete based on the sensor data specifically includes: Based on the torque data and the geometric parameters of the screw, the shear stress borne by the concrete is calculated, and based on the real-time rotational speed and geometric parameters of the screw, the shear rate of the concrete is calculated. Based on the shear stress and the shear rate, the initial yield stress and initial plastic viscosity are obtained by linear fitting using the least squares method. If the initial yield stress and the initial plastic viscosity pass the verification, the initial yield stress is taken as the real-time yield stress, and the initial plastic viscosity is taken as the real-time plastic viscosity.

3. The intelligent optimization method for the molding quality of 3D printed concrete according to claim 1, characterized in that, The process of processing the morphological image to obtain defect features specifically includes: The morphological image is preprocessed to obtain a standardized image; The standardized image is input into a trained neural network model, which outputs defect features.

4. The intelligent optimization method for the molding quality of 3D printed concrete according to claim 1, characterized in that, The target commands include a first command from the screw servo motor, a second command from the vertical axis servo motor, and a third command from the horizontal motion system servo motor. The step of determining the target instruction based on the defect information specifically includes: Based on the defect type, the first probability distribution, the second probability distribution, and the third probability distribution, determine the anomaly classification corresponding to each key parameter; Based on the anomaly classification corresponding to each key parameter, the first command of the screw servo motor, the second command of the vertical axis servo motor, and the third command of the horizontal motion system servo motor are obtained.

5. A monitoring device for implementing the intelligent optimization method for the molding quality of 3D printed concrete according to any one of claims 1 to 4, characterized in that, The monitoring device is installed on the concrete printing equipment and includes a camera bracket, a scanner bracket, a laser scanner, an industrial camera, and sensor components. The camera bracket is mounted on the print head of the concrete printing equipment, the scanner bracket is connected to the camera bracket, the laser scanner is connected to the scanner bracket, the industrial camera is connected to the camera bracket, and the sensor assembly is connected to a component of the print head. The industrial camera is used to acquire morphological images of concrete during the 3D printing process, the laser scanner is used to acquire the real-time offset of the print head in the Z-axis direction, and the sensor assembly is used to acquire the real-time rheological parameters of concrete during the 3D printing process.

6. The monitoring device according to claim 5, characterized in that, The sensor assembly includes a fixed base, a torque sensor, and a pressure sensor; The torque sensor is connected to the motor shaft and the screw rod, and the torque sensor is connected to the fixed base, which is connected to the bottom of the motor; the pressure sensor is connected to the bottom of the print head. The torque sensor is used to measure the torque that the screw needs to overcome to rotate, and the pressure sensor is used to measure the positive pressure generated when concrete is propelled forward in the pipe.

Images