A smart construction system and method for laser-leveled high-temperature crack-resistant flooring integrating digital twin and AI decision-making

Abstract

This invention provides an intelligent construction system and method for laser-leveled high-temperature crack-resistant flooring that integrates digital twins and AI decision-making. The system uses a data platform as the data interaction center and connects three subsystems: an intelligent material preparation subsystem, used to monitor the physical properties of concrete raw materials in real time, obtain flooring batching schemes, and complete the preparation of flooring materials according to the batching schemes; an intelligent construction equipment cluster, including an intelligent laser screed machine, an automatic curing robot, and an AI vision acceptance robot, used for flooring construction and collecting construction data; and a digital twin construction management platform, used to combine a flooring construction digital twin model built based on Revit and Unity and construction data to map the flooring construction of the intelligent construction equipment cluster and assist in the flooring construction process.

Description

Technical Field

[0001] This invention belongs to the field of intelligent construction, specifically relating to an intelligent construction system and method for laser-leveled high-temperature crack-resistant flooring that integrates digital twins and AI decision-making. Background Technology

[0002] Currently, laser leveling floor construction still relies mainly on "traditional materials + manual control." Even with the integration of some digital technologies, the core problems of "data fragmentation, delayed decision-making, and low intelligence" remain.

[0003] 1. Material preparation relies on fixed formulas: The dynamic relationship between raw material characteristics (such as fluctuations in sand and gravel moisture content) and ambient temperature is not considered. At high temperatures, formula rigidity leads to a crack rate exceeding 20%, and the lack of digital traceability means that raw material quality problems cannot be quickly located. In traditional processes, formula adjustments require manual calculation, taking ≥2 hours, and the adjustment accuracy depends on experience, making it unable to adapt to batch differences in raw materials and real-time environmental changes.

[0004] 2. Construction parameters are largely based on manual experience: Laser screed machine parameters (travel speed, vibration frequency) are mostly set by workers based on experience, without real-time linkage with site temperature and floor flatness. At 35℃, the flatness compliance rate is only 80%. The construction process lacks digital mapping, making it impossible to predict changes in concrete fluidity due to temperature increases, often resulting in under-vibration or over-vibration, leading to a rework rate exceeding 15%.

[0005] 3. Low efficiency in maintenance and acceptance: Maintenance relies on manual watering, with humidity control accuracy of only ±10%. Low-humidity areas are prone to shrinkage cracks, while high-humidity areas are prone to mold growth. Acceptance relies on manual inspection using a 2m level, which takes 3 days for 10,000㎡. Data cannot be archived in real time, resulting in a lack of digital archives for later maintenance. Tracing the source of quality problems takes ≥48 hours.

[0006] 4. Fragmentation of Smart Technologies: While some projects have introduced BIM or IoT, a closed loop of "data-decision-execution" has not been formed. BIM is only used for design modeling and is not linked with construction equipment; IoT data is only stored and not used for parameter optimization, failing to realize the core value of smart construction. For example, IoT may detect an increase in the moisture content of sand and gravel, but it cannot automatically trigger formula adjustments, still requiring manual intervention. Summary of the Invention

[0007] To overcome the shortcomings of existing laser leveling flooring construction technologies, this invention provides an intelligent construction system and method for high-temperature crack-resistant laser leveling flooring that integrates digital twins and AI decision-making. Using a data platform as the data interaction center, along with an intelligent material preparation subsystem, an intelligent construction equipment cluster, and a digital twin construction management platform, it deeply integrates digital twins, AI decision-making, and IoT technologies with traditional flooring construction. This constructs a fully intelligent closed loop of "perception-decision-execution-feedback," breaking through the bottleneck of data fragmentation in various stages of traditional flooring construction and achieving a transformation from "experience-driven" to "data-driven."

[0008] According to one aspect of this specification, a laser-leveled high-temperature crack-resistant flooring intelligent construction system integrating digital twin and AI decision-making is provided, comprising:

[0009] The intelligent material preparation subsystem is used to monitor the physical properties of concrete raw materials in real time, obtain the flooring batching plan, and complete the preparation of flooring materials according to the flooring batching plan;

[0010] The intelligent construction equipment cluster, including intelligent laser screed machines, automatic maintenance robots, and AI vision acceptance robots, is used for floor construction and to collect construction data. The collected construction data includes: both the intelligent laser screed machines and AI vision acceptance robots are equipped with GNSS positioning modules to collect the positioning data of the AI ​​vision acceptance robot in real time, and are equipped with lidar to collect the three-dimensional coordinate data of the floor surface. The AI ​​vision acceptance robot is additionally equipped with a high-definition camera to capture images of the floor surface, and is equipped with an edge computing module that deploys the YOLOv8 crack recognition algorithm to mark crack information in the floor surface images.

[0011] The digital twin construction management platform is used to combine the digital twin model of floor construction built based on Revit and Unity with construction data to map the floor construction of the intelligent construction equipment cluster and assist in the floor construction process.

[0012] The data platform receives environmental temperature and humidity data and provides data exchange between the intelligent material preparation subsystem, the intelligent construction equipment cluster, and the digital twin construction management and control platform.

[0013] As a further technical solution, the intelligent material preparation subsystem includes:

[0014] The raw material IoT detection unit is used to monitor the physical properties of concrete raw materials, and the monitoring data is uploaded to the data platform in real time. The physical properties of the raw materials include, but are not limited to, the moisture content of sand and gravel, the purity of corundum, and the specific surface area of ​​cement.

[0015] The AI ​​dynamic batching model includes a deep neural network model built on the TensorFlow framework. Under the preset target compressive strength and target crack rate conditions, the deep neural network model outputs a flooring batching scheme based on the physical properties of the raw materials and the ambient temperature obtained from the data platform. The flooring batching scheme includes the gradation ratio of corundum, the dosage of expansion agent, and the mixing time.

[0016] The prefabrication process digital monitoring unit is used to complete the preparation of flooring materials according to the flooring batching plan and to obtain the aggregate dispersion in real time during the preparation process.

[0017] As a further technical solution, the prefabrication process digital monitoring unit includes: a feeding valve, a variable frequency mixer, and an aggregate dispersion laser detector; wherein, the automatic feeding valve feeds materials according to the floor material distribution plan, the variable frequency mixer starts mixing, and during the mixing process, the aggregate dispersion laser detector monitors the aggregate dispersion.

[0018] As a further technical solution, the AI ​​dynamic batching model incorporates an AI model to acquire monitoring data on ambient temperature and the physical properties of concrete raw materials. When the real-time physical properties of raw materials and / or ambient temperature fluctuations exceed a preset amplitude, the deep neural network is invoked to output an updated floor batching scheme. Additionally, the model acquires monitoring data on aggregate dispersion in the digital monitoring unit of the prefabrication process. When the aggregate dispersion does not meet the preset standard, a speed adjustment command is triggered to control the variable frequency mixer to increase its speed.

[0019] As a further technical solution, the process of building a digital twin model of floor construction based on Revit and Unity includes:

[0020] Import the CAD drawings of the flooring project into Revit, build the flooring BIM model, and annotate the flooring BIM model according to the target construction parameters of the flooring; the annotation content includes, but is not limited to, flooring zoning, design elevation, and material properties;

[0021] By adding a physics engine to the floor BIM model using Unity to simulate the high-temperature drying shrinkage process, a digital twin model of the floor construction is obtained.

[0022] As a further technical solution, the digital twin construction management platform assists in the construction of the floor slab, including:

[0023] It receives real-time positioning data from the intelligent laser screed and three-dimensional coordinate data of the floor surface, calculates the flatness deviation, and has an AI decision tree trained on gradient boosting tree. With temperature, flatness deviation and concrete initial setting time as input, it outputs the adjustment values ​​of the intelligent laser screed's walking speed and vibration frequency, and generates instructions for adjusting the intelligent laser screed's walking speed and vibration frequency.

[0024] The system receives environmental humidity data from the data platform, combines it with the built-in digital twin model of floor construction, generates a heat map of humidity distribution on the floor surface, and sends it to the automatic spraying maintenance robot.

[0025] The system receives real-time positioning data from the AI ​​vision inspection robot, three-dimensional coordinate data of the floor surface, and crack information. Based on the real-time positioning data and the three-dimensional coordinate data of the floor surface, it generates a floor flatness heat map, calculates the flatness pass rate, and combines the crack information in the floor surface image to obtain the acceptance result of the floor construction.

[0026] As a further technical solution, the automatic spraying maintenance robot is equipped with a path planning module to plan its travel route; a variable frequency sprayer; a humidity sensor to acquire real-time humidity data of the floor surface; and a built-in AI module to construct a differentiated spraying scheme based on the floor surface humidity distribution heat map. During the automatic spraying maintenance robot's movement, the spray volume of the variable frequency sprayer is adjusted based on the real-time floor surface humidity data and the differentiated spraying scheme to implement floor spraying maintenance. The differentiated spraying scheme divides the spraying area into high humidity, medium humidity, and low humidity zones, and adopts differentiated spray volumes for different spraying areas.

[0027] As a further technical solution, the data platform also stores quality parameters and equipment safety parameter thresholds for early warning. Quality parameters include flatness deviation and crack width, while equipment safety parameters include the amplitude of the intelligent laser leveling machine and the power consumption of the automatic spray maintenance robot. When the quality parameters and / or equipment safety parameters exceed the threshold, an early warning message is automatically sent to the management personnel.

[0028] According to another aspect of this specification, a smart construction method for laser-leveled high-temperature crack-resistant flooring integrating digital twins and AI decision-making is provided. This method is implemented based on a smart construction system for laser-leveled high-temperature crack-resistant flooring integrating digital twins and AI decision-making, and includes the following steps:

[0029] After the raw materials for flooring construction arrive on site, the intelligent material preparation subsystem completes the preparation of the flooring materials, including: after the raw materials arrive on site, the raw material IoT detection unit collects the physical characteristics of the raw materials. After the collection is completed, the physical characteristic data of the raw materials is uploaded to the data platform in real time. The data platform inputs the physical characteristic data of the raw materials and the real-time ambient temperature into the AI ​​dynamic batching model and outputs the adjusted flooring batching scheme. The prefabrication process digital monitoring unit is used to feed and mix the materials according to the flooring batching scheme to complete the preparation of the flooring materials.

[0030] The digital twin construction management platform issues construction tasks to the intelligent laser screed. The laser radar inside the intelligent laser screed starts scanning, generating point cloud data of the local surface of the floor. It compares this data with the digital twin model of the floor construction and calculates the flatness deviation of the initial construction area. The AI ​​decision tree takes temperature, flatness deviation, and concrete initial setting time as inputs and outputs adjustment values ​​for the intelligent laser screed's walking speed and vibration frequency, generating intelligent construction and adjustment instructions for the intelligent laser screed's walking speed and vibration frequency. The intelligent laser screed executes the floor screed construction according to the walking speed and vibration frequency adjustment instructions.

[0031] After the floor leveling construction is completed, the automatic spraying maintenance robot starts its operation, executes a differentiated spraying plan, and completes the spraying maintenance of the lawn. After the preset maintenance cycle is met, the AI ​​vision acceptance robot starts its operation. The lidar in the AI ​​vision acceptance robot collects the three-dimensional coordinate data of the floor surface, generates a flatness heat map based on the flatness data in the digital twin construction control platform, calculates the flatness compliance rate, and takes images of the floor surface through a camera. The edge computing unit of the YOLOv8 crack recognition algorithm is used to obtain crack information and obtain the acceptance results of the floor construction.

[0032] As a further technical solution, the method also includes: pre-simulation based on a digital twin model of the floor construction:

[0033] Import historical temperature and humidity data of the area to be constructed, as well as the physical property parameters of the raw materials to be used, input the AI ​​dynamic batching model, and output several flooring batching schemes;

[0034] Several sets of intelligent laser screed machine walking speed and vibration frequency parameters are preset and combined with several floor material preparation schemes. Based on the digital twin model of floor construction, the floor screed construction of different combinations is simulated, and the floor crack rate and flatness compliance rate of each combination are obtained. The combination with the best comprehensive results of floor crack rate and flatness compliance rate is selected as the initial floor material preparation scheme and initial construction parameters.

[0035] Compared with existing technologies, the beneficial effects of this invention are as follows: By using a data platform as the data interaction center to connect the intelligent material preparation subsystem, intelligent construction equipment cluster, and digital twin construction management platform, it deeply integrates digital twin, AI decision-making, and IoT technologies with traditional flooring construction, constructing a full-process intelligent closed loop of "perception-decision-execution-feedback." This breaks through the bottleneck of data fragmentation in various stages of traditional flooring construction, achieving a transformation from "experience-driven" to "data-driven," and realizing intelligent, precise, and efficient flooring construction. The intelligent material preparation subsystem provides dynamic correlation considering the physical properties of raw materials and ambient temperature, quickly providing flooring material mixing solutions, and providing real-time updates based on AI models to better adapt to batch differences in raw materials and real-time environmental changes. The digital twin construction management platform maps and assists in mapping flooring construction, considering real-time linkage of site temperature and floor flatness, providing more productive intelligent laser screed machine parameters (walking speed, vibration frequency), resulting in better screeding construction effects. Assisted by automatic spray curing robots and AI visual acceptance robots, it improves the efficiency of flooring maintenance and acceptance. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 A schematic diagram of the structure of a laser-leveled high-temperature crack-resistant flooring intelligent construction system integrating digital twin and AI decision-making, provided in an embodiment of the present invention;

[0038] Figure 2 This invention provides a schematic diagram illustrating the training of the AI ​​dynamic ingredient dispensing model in an embodiment of the invention.

[0039] Figure 3 A flowchart illustrating a smart construction method for laser-leveled high-temperature crack-resistant flooring that integrates digital twin and AI decision-making, provided as an embodiment of the present invention;

[0040] Figure 4 This is a flowchart illustrating an intelligent construction example of laser-leveled high-temperature crack-resistant flooring that integrates digital twin and AI decision-making, as provided in an embodiment of the present invention. Detailed Implementation

[0041] It should be noted that:

[0042] The terms “comprising” and “having”, and any variations thereof, in the specification, claims, and accompanying drawings of this invention are intended to cover a non-exclusive inclusion, such as a process, method, system, product, or apparatus that includes a series of steps or units, not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0043] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices. The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be decomposed, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0044] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. In addition, the technical features of the various embodiments or individual embodiments provided by the present invention can be arbitrarily combined to form new technical solutions. Such combinations are not bound by the order of steps and / or structural composition patterns, but must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0045] like Figure 1 As shown, a smart construction system for laser-leveled high-temperature crack-resistant flooring, integrating digital twins and AI decision-making, uses a data platform as the data exchange center, connecting three subsystems:

[0046] The intelligent material preparation subsystem is used to monitor the physical properties of concrete raw materials in real time, obtain the flooring batching plan, and complete the preparation of flooring materials according to the flooring batching plan;

[0047] The intelligent construction equipment cluster, including intelligent laser screed machines, automatic maintenance robots, and AI vision acceptance robots, is used for floor construction and to collect construction data. The collected construction data includes: both the intelligent laser screed machines and AI vision acceptance robots are equipped with GNSS positioning modules to collect the positioning data of the AI ​​vision acceptance robot in real time, and are equipped with lidar to collect the three-dimensional coordinate data of the floor surface. The AI ​​vision acceptance robot is additionally equipped with a high-definition camera to capture images of the floor surface, and is equipped with an edge computing module that deploys the YOLOv8 crack recognition algorithm to mark crack information in the floor surface images.

[0048] The digital twin construction management platform is used to combine the digital twin model of floor construction built based on Revit and Unity with construction data to map the floor construction of the intelligent construction equipment cluster and assist in the floor construction process.

[0049] Firstly, the intelligent material preparation subsystem includes the following core modules:

[0050] (a1) Raw material IoT detection unit, used to monitor the physical properties of concrete raw materials, including but not limited to sand and gravel moisture content, corundum purity and cement specific surface area.

[0051] Optionally, the raw material IoT detection unit includes a sand and gravel moisture content sensor, a cement specific surface area analyzer, and a corundum purity analyzer. The sand and gravel moisture content sensor adopts a high-precision capacitive detection principle with a detection accuracy of ±0.1%, and can collect data in real time during sand and gravel transportation without manual sampling. Based on the Blaine method, the detection time is ≤5 minutes, and the detection result error is ≤2%. The corundum purity analyzer uses X-ray fluorescence spectroscopy technology with an accuracy of ±0.5%, and can identify impurities (such as iron and silicon) in corundum, preventing impurities from affecting the wear resistance of the flooring. All detection data is uploaded to the data platform in real time via a 5G network with an upload latency of ≤1 second.

[0052] (a2) The AI ​​dynamic batching model is a deep neural network model built on the TensorFlow framework. Under the premise of target compressive strength and target crack rate, it outputs the flooring batching scheme according to the physical properties of raw materials and ambient temperature.

[0053] The flooring mix design includes the proportion of corundum gradation, the amount of expanding agent, and the mixing time.

[0054] Specifically, the AI ​​dynamic batching model structure includes an input layer, three hidden layers, and an output layer. Optionally, the input layer has six parameters: sand and gravel moisture content, cement specific surface area, corundum purity, ambient temperature, target compressive strength, and target crack rate. The first three parameters are physical characteristic parameters of concrete raw materials, which can be changed and / or added or removed according to the actual physical parameter characteristics considered, and the input layer can be modified accordingly. Each hidden layer has 64 neurons and uses the ReLU activation function to enhance the model's nonlinear fitting ability. The output layer has five key parameters: coarse:medium:fine gradation of corundum, expansion agent dosage, and mixing time.

[0055] Specifically, the training dataset for the AI ​​dynamic batching model includes over 1000 sets of historical data, covering environments ranging from 25-40℃, three batches of corundum aggregate, and two types of cement. Preferably, the AI ​​dynamic batching model supports real-time learning, receiving daily production data and completing one iteration every 24 hours, with the model error consistently controlled within 4.2% after iteration. Figure 2 As shown, the training process includes:

[0056] Data collection: Collected physical property parameters of raw materials (moisture content, specific surface area, purity), environmental parameters (temperature, humidity), flooring batching scheme parameters (gradation, admixture), and target material properties (target compressive strength, target crack rate), totaling 1200 sets of data;

[0057] Data preprocessing: missing values ​​were filled using the mean method, outliers were removed using the 3σ principle (removal rate 5%), and Min-Max standardization was used to map the data to the [0,1] interval to ensure the stability of model training;

[0058] Model training: The training set (80%) and validation set (20%) were split. A TensorFlow model was built (6 neurons in the input layer, 3×64 neurons in the hidden layer, and 5 neurons in the output layer). The Adam optimizer was used (learning rate 0.001), and the model was iterated 1000 times. The accuracy on the training set was 95%, and the accuracy on the validation set was 92%.

[0059] Model validation: Test the model with 50 sets of new data. If the error is >5%, return to the data collection stage to supplement data. If the error is ≤5%, proceed to model deployment.

[0060] Model Deployment and Iteration: After the model is deployed, real-time production data is received every 24 hours and an iteration is completed. After the iteration, the model error is reduced by 0.1-0.3%, and the accuracy is continuously optimized.

[0061] Preferably, the AI ​​dynamic batching model provides the function of continuously optimizing the accuracy of the formula. For example, when the purity of a certain batch of corundum increases to 99.5%, the proportion of coarse aggregate is automatically reduced by 1% to ensure that the aggregate packing density remains unchanged.

[0062] (a3) Prefabrication process digital monitoring unit (or dynamic batching execution unit) is used to complete the preparation of flooring materials according to the flooring batching plan.

[0063] The prefabrication process digital monitoring unit (dynamic batching execution unit) includes: an automatic feeding valve, a variable frequency mixer, and an aggregate dispersion laser detector. Optionally, the automatic feeding valve is controlled by a servo motor with a control accuracy of ±0.5kg, automatically feeding materials according to the flooring batching plan, ensuring precise control of the amount of each raw material in the formula; the variable frequency mixer has a speed adjustment range of 0-200r / min, supports stepless speed regulation, and can adjust the speed according to the aggregate type and formula (built-in mixing speed sensor can collect speed data in real time within the 0-200r / min range); the aggregate dispersion laser detector uses line laser scanning technology with a scanning frequency of 1 scan / 10s, which can detect the uniformity of aggregate dispersion during the mixing process in real time.

[0064] The automatic feeding valve feeds materials according to the floor material batching plan, and the frequency converter starts mixing. During the mixing process, the aggregate dispersion laser detector monitors the aggregate dispersion in real time.

[0065] Furthermore, the AI ​​dynamic batching model incorporates an AI model to acquire monitoring data on ambient temperature and the physical properties of concrete raw materials. When the real-time physical properties of raw materials and / or ambient temperature fluctuations exceed a preset amplitude, the deep neural network model is invoked to output an updated floor batching scheme. Additionally, the model acquires monitoring data on aggregate dispersion in the prefabrication process digital monitoring unit. When the aggregate dispersion does not meet the preset standard, a speed adjustment command is triggered to control the variable frequency mixer to increase its speed.

[0066] Optionally, when the moisture content of raw materials fluctuates by ±1% and / or the ambient temperature fluctuates by ±2℃, the model can output a new formula within 30 seconds. For example, when the moisture content of sand and gravel is 6%, the model will automatically increase the proportion of corundum coarse aggregate to 32% and extend the mixing time by 10 seconds to ensure that the aggregate dispersion is ≥95%.

[0067] The preset standard for aggregate dispersion is set to be no less than 95%. When the proportion of fine aggregate is high, if the dispersion is detected to be <95%, it will be automatically fed back to the AI ​​dynamic batching model and output a speed adjustment command within 30 seconds to increase the speed of the variable frequency mixer by 5-10 r / min to avoid agglomeration until the dispersion meets the standard. Finally, the speed of the variable frequency mixer is increased to 180 r / min.

[0068] In one specific embodiment, the operation process of the smart material preparation subsystem includes:

[0069] After the raw materials arrive on site, the raw material IoT detection unit collects the physical characteristics of the raw materials. After the collection is completed, the physical characteristic data of the raw materials is uploaded to the data platform in real time. The data platform inputs the physical characteristic data of the raw materials and the real-time ambient temperature into the AI ​​dynamic batching model and outputs the adjusted flooring batching scheme. The prefabrication process digital monitoring unit (dynamic batching execution unit) is used to feed and mix the materials according to the flooring batching scheme to complete the preparation of the flooring material.

[0070] In a specific embodiment, at a temperature of 35°C, when the raw material IoT detection unit collected data showing that the sand and gravel moisture content was 6.2% and the cement specific surface area was 345㎡ / kg, the AI ​​dynamic batching model output the flooring batching scheme: coarse corundum aggregate accounting for 32%, medium corundum accounting for 50%, fine corundum accounting for 18%, and expansion agent dosage of 4.2%.

[0071] The prefabrication process digital monitoring unit feeds materials according to the flooring batching plan and starts mixing. During the mixing process, the laser detector of the prefabrication process digital monitoring unit monitors the aggregate dispersion in real time. When the dispersion is initially measured at 92% (not meeting the ≥95% qualification standard), the AI ​​dynamic batching model issues a mixing speed adjustment command, increasing the speed of the variable frequency mixer from 150r / min to 180r / min. After 30 seconds, the dispersion is retested and reaches 96%, meeting the qualification requirements.

[0072] Secondly, the intelligent construction equipment cluster includes:

[0073] (b1) Intelligent laser screed machine, the actuator receives instructions from the digital twin platform to adjust the walking speed range and vibration frequency range for lawn screed construction; equipped with GNSS positioning module to collect the positioning data of intelligent laser screed machine in real time, and equipped with lidar to collect the three-dimensional coordinate data of the ground surface.

[0074] Preferably, the intelligent laser screed is equipped with an edge computing unit to autonomously identify obstacles in the construction area and make fine-tuning of the path.

[0075] Optionally, in one specific embodiment, the intelligent laser screed is equipped with an edge computing unit, a GNSS positioning module, and an actuator. The first edge computing unit uses an industrial-grade CPU, supports local data processing, reduces data transmission latency, and can autonomously identify obstacles such as steel stirrups and embedded parts in the construction area. The identification algorithm is based on YOLOv7 training, with an accuracy rate of ≥95%. After identifying an obstacle, it completes path fine-tuning within 10 seconds to avoid equipment collisions. The GNSS positioning module receives positioning data in real time to ensure that the intelligent laser screed constructs according to the planned path, automatically correcting when the deviation exceeds 3cm. The laser radar ranging accuracy (scanning resolution) reaches ±0.1mm, and the acquisition frequency... The cycle time is 1 time / 30s, which can generate a 3D point cloud model of the floor surface for calculating flatness deviation: the flatness deviation within a radius of 2m is calculated, and the calculation result error is ≤0.2mm; the actuator is hydraulically driven and can receive instructions from the digital twin platform to adjust the walking speed range to 0.8-1.2m / min and the vibration frequency range to 60-70Hz, with an adjustment response time of ≤5s. For example, after the platform issues the instruction of "walking speed 0.85m / min, vibration frequency 69Hz", the actuator completes the parameter adjustment within 5s.

[0076] (b2) An automatic spraying maintenance robot is equipped with a path planning module to plan the robot's route; a variable frequency sprayer; a humidity sensor to acquire real-time humidity data of the floor surface; and a built-in AI module to construct a differentiated spraying scheme based on the floor surface humidity distribution heat map. During the automatic spraying maintenance robot's movement, the spray volume of the variable frequency sprayer is adjusted based on the real-time floor surface humidity data and the differentiated spraying scheme to implement floor spraying maintenance. The differentiated spraying scheme divides the spraying area into high humidity, medium humidity, and low humidity zones, and adopts differentiated spray volumes for different spraying areas.

[0077] Optionally, in the differentiated spraying scheme, a standard spray volume is preset in the medium humidity zone, the spray volume in the low humidity zone is increased by 20% compared to the medium humidity zone, and the spray volume in the high humidity zone is reduced by 30% compared to the medium humidity zone.

[0078] Preferably, the automatic spray maintenance robot also has a power management function. When the power is lower than the preset low power threshold, it will automatically return to the charging area to recharge, ensuring continuous maintenance operations.

[0079] Optionally, in one specific embodiment, the automated spraying maintenance robot achieves autonomous path planning based on the SLAM algorithm, with a path planning accuracy of ±10cm, 100% coverage, and a repetition rate of ≤5%, avoiding missed spraying or repeated spraying. The robot is equipped with a humidity sensor and a variable frequency spray pump. The humidity sensor has a detection range of 0-100%, which can collect real-time humidity data of the floor surface; the variable frequency spray pump has a spray volume adjustment range of 15-20L / min. The humidity distribution heat map generated by the digital twin construction management module divides the spraying area into high humidity zone (humidity > 90%), medium humidity zone (humidity 80-90%), and low humidity zone (humidity < 80%), and adopts a differentiated spraying strategy: the spray volume is 18-20L / min in the low humidity zone, 15-17L / min in the medium humidity zone, and 10-12L / min in the high humidity zone, ensuring that the humidity of the floor surface is uniformly maintained at 80-90%, reducing the risk of drying shrinkage cracks. In addition, the robot has a power management function. When the power is below 20%, it will automatically return to the charging area to recharge. Charging for 30 minutes can restore 50% of the power, ensuring continuous maintenance operations.

[0080] (b3) AI vision inspection robot for floor inspection: equipped with LiDAR to collect three-dimensional coordinate data of the floor surface; equipped with a high-definition camera to capture images of the floor surface; equipped with an edge computing module and deployed YOLOv8 crack recognition algorithm to mark crack information in the floor surface images, including cracks and their location, length and width.

[0081] Preferably, during the acceptance process, the AI ​​vision acceptance robot scans along a preset path and periodically generates phased acceptance reports. The reports include flatness deviation data, a crack list, and the acceptance pass rate calculated based on the flatness deviation data and crack list. The data is uploaded to the data platform in real time, allowing engineers to view and review it online.

[0082] The acceptance pass rate is calculated based on the flatness deviation threshold and the maximum crack size within the preset area.

[0083] Optionally, in one specific embodiment, the AI-powered inspection robot is equipped with a LiDAR, a 4K high-definition camera, and an edge computing module. The LiDAR has a ranging accuracy of ±0.1mm and can collect three-dimensional coordinate data of the floor surface to calculate the flatness deviation within a 2m range, with a calculation error ≤0.2mm. The 4K high-definition camera captures images at 30fps, clearly capturing minute cracks on the floor surface, with an image resolution of 3840×2160. An edge computing module is deployed with the YOLOv8 crack recognition algorithm, which has been trained on over 10,000 crack images, achieving a recognition accuracy of ≥98%. It can automatically label the location, length, and width of cracks, with a width recognition accuracy of 0.1mm. During the inspection process, the robot scans along a preset path at a speed of 0.5m / s, generating a phased inspection report every 100㎡ scanned.

[0084] Preferably, the intelligent construction equipment cluster supports remote operation and maintenance. The data platform allows real-time monitoring of the operational status of intelligent laser screed machines, automatic spraying maintenance robots, and AI vision acceptance robots, including but not limited to battery level, fault codes, and operation duration. When equipment malfunctions, the data platform's AI model automatically analyzes the cause of the fault through its diagnostic module and pushes the most suitable repair solution to the on-site engineer based on historical maintenance data. This repair solution includes disassembly steps, required spare parts models, and precautions, which can reduce repair time by 40%.

[0085] Third, the digital twin construction management platform incorporates an AI decision tree trained using XGBoost, which takes temperature, flatness deviation, and concrete initial setting time as inputs and outputs adjustment values ​​for the intelligent laser screed's walking speed and vibration frequency, generating instructions for adjusting the intelligent laser screed's walking speed and vibration frequency. A digital twin model of the floor construction is built based on Revit and Unity for 3D visualization of the construction process.

[0086] In one specific implementation, the training data for the AI ​​decision tree includes 500+ sets of on-site construction data, covering different temperature ranges from 25-40℃, with a decision response time of ≤30s.

[0087] Preferably, the AI ​​decision tree also has a temperature prediction function. By inputting historical meteorological data from the same period and the current real-time temperature, it obtains a prediction of future temperature change trends, and then outputs an adjustment plan for the intelligent laser screed's walking speed and vibration frequency in advance. In a preferred embodiment, the AI ​​decision tree predicts the temperature change trend 2 hours in advance based on meteorological data from the same period over the past 5 years and real-time environmental parameters, achieving a prediction accuracy of 92%. Based on this, when it predicts that the temperature will rise by 2°C in 1 hour, the AI ​​decision tree, combining the predicted temperature with the flatness deviation and the initial setting time of the concrete, outputs an adjustment plan in advance: "reduce the walking speed by 0.05 m / min and increase the vibration frequency by 1 Hz," avoiding fluctuations in construction quality caused by temperature increases.

[0088] In the digital twin construction management platform, the process of building a digital twin model of floor construction based on Revit and Unity includes:

[0089] Import the CAD drawings of the flooring project into Revit, build the flooring BIM model, and annotate the flooring BIM model according to the target construction parameters of the flooring; the annotation content includes, but is not limited to, flooring zoning, design elevation, and material properties;

[0090] By adding a physics engine to the floor BIM model using Unity to simulate the high-temperature drying shrinkage process, a digital twin model of the floor construction is obtained.

[0091] Furthermore, the digital twin construction management platform assists in the construction of the flooring, including:

[0092] It receives real-time positioning data from the intelligent laser screed and three-dimensional coordinate data of the floor surface, calculates the flatness deviation, and has an AI decision tree trained on gradient boosting tree. With temperature, flatness deviation and concrete initial setting time as input, it outputs the adjustment values ​​of the intelligent laser screed's walking speed and vibration frequency, and generates instructions for adjusting the intelligent laser screed's walking speed and vibration frequency.

[0093] The system receives environmental humidity data from the data platform, combines it with the built-in digital twin model of floor construction, generates a heat map of humidity distribution on the floor surface, and sends it to the automatic spraying maintenance robot.

[0094] The system receives real-time positioning data from the AI ​​vision inspection robot, three-dimensional coordinate data of the floor surface, and crack information. Based on the real-time positioning data and the three-dimensional coordinate data of the floor surface, it generates a floor flatness heat map, calculates the flatness pass rate, and combines the crack information in the floor surface image to obtain the acceptance result of the floor construction.

[0095] Specifically, the digital twin model of the floor construction enables 3D visualization of the construction process, with a scene scale of 1:1. It can display the construction progress, equipment status, and quality indicators in real time: the construction progress is displayed as a block-by-block completion rate. For example, in the E1 plant, block ③ has a completion rate of 65% and is expected to be completed in 1.5 hours. The equipment status displays the location, walking speed, vibration frequency, and battery level of the intelligent laser screed. In case of failure, it displays the fault code and preliminary diagnosis results. The quality indicators are displayed in the form of a heat map to show the flatness deviation. The red area indicates a deviation >2mm, the yellow area indicates a deviation 1-2mm, and the green area indicates a deviation <1mm. At the same time, crack risk warning areas are marked. The warning areas are calculated based on the temperature change predicted by AI and the concrete drying shrinkage coefficient. For example, an area with a sudden temperature drop of 5℃ will be marked as a high-risk area to remind engineers to monitor it closely.

[0096] The data platform internally stores real-time environmental temperature and humidity data from field sensor units and data from various subsystems. It also provides storage for preset quality and equipment safety thresholds and connects to various subsystems via API interfaces, providing data interoperability.

[0097] The environmental sensor data includes real-time ambient temperature and humidity, with optional ambient temperature detection accuracy of ±0.5℃ and ambient humidity detection accuracy of ±2%, updated every 10 seconds to ensure real-time reflection of changes in the construction environment.

[0098] Specifically, the data platform uses a time-series database (InfluxDB) to store data for the entire construction process. This database supports high-concurrency writing and fast querying, with a data retention period of ≥5 years. It can be searched in multiple dimensions by "project name-construction block-time node". For example, querying the construction data of "Hikvision Project E1 Plant ③ Block July 18, 2025 19:00" can retrieve information such as temperature, laser leveling machine parameters, and flatness deviation within 10 seconds.

[0099] Preferably, the data delay is ≤1s. For example, the formula data of the intelligent material preparation subsystem can be sent to the dynamic batching execution unit in real time, and the construction instructions of the digital twin platform can be pushed to the intelligent laser leveling machine in real time. At the same time, it supports two-way data interaction. The working status data of the intelligent equipment (such as the fault code of the laser leveling machine) can be uploaded to the platform in real time. The platform adjusts the instructions according to the data to form a closed-loop control.

[0100] The data platform also stores quality parameters and equipment safety parameter thresholds for early warning: quality parameters include flatness deviation and crack width, and equipment safety parameters include laser leveling machine amplitude and maintenance robot power. When quality and / or equipment safety parameters exceed the threshold, an early warning message is automatically sent to the management personnel.

[0101] In one specific embodiment, quality thresholds are set including flatness deviation ≥2mm and crack width ≥0.15mm, while equipment safety thresholds include intelligent laser screed amplitude ≥0.5mm and automatic spray maintenance robot battery level ≤10%. When a threshold is exceeded, an early warning message is automatically sent to the mobile terminal of the management personnel, with a response time ≤1min. Preferably, a data traceability function is also provided, supporting reverse traceability of "quality problem - construction parameters - raw materials". For example, if a crack width of 0.18mm is found in a certain area, clicking on the crack location will display the temperature (36.5℃), laser screed parameters (walking speed 0.85m / min), and raw material batch (sand and gravel moisture content 6.2%) for the corresponding construction period, helping to quickly locate the cause and reducing the traceability time from the traditional 48 hours to 5 minutes.

[0102] Preferably, the data platform incorporates an AI data analysis module for AI data analysis: employing association rule mining algorithms, it automatically analyzes the correlation between raw material characteristics, environmental parameters, and floor quality. For example, by analyzing over 1000 sets of data, it discovers the pattern that "for every 1% increase in the moisture content of sand and gravel, the dosage of expanding agent needs to be increased by 0.2% to maintain a crack rate ≤3%", and generates an optimal construction parameter template based on this pattern, with a template adaptability ≥90%. Simultaneously, the module can also predict the long-term performance of the floor. Based on construction data and material characteristics, it predicts the wear resistance and crack development trend of the floor after one year, with a prediction accuracy ≥85%, providing a reference for subsequent operation and maintenance.

[0103] Preferably, the data platform also includes automatically generated digital quality archives, which contain raw material testing reports, dynamic curves of construction parameters, flatness heat maps, crack photos, and acceptance results. The platform allows both the construction company and the supervision company to view and review these archives online.

[0104] This invention proposes an intelligent construction method for laser-leveled high-temperature crack-resistant flooring that integrates digital twins and AI decision-making. It is implemented based on an intelligent construction system for laser-leveled high-temperature crack-resistant flooring that integrates digital twins and AI decision-making. The steps include:

[0105] Step 1: After the raw materials for flooring construction arrive on site, the intelligent material preparation subsystem completes the preparation of the flooring materials, including: after the raw materials arrive on site, the raw material IoT detection unit collects the physical characteristics of the raw materials. After the collection is completed, the physical characteristic data of the raw materials is uploaded to the data platform in real time. The data platform inputs the physical characteristic data of the raw materials and the real-time ambient temperature into the AI ​​dynamic batching model and outputs the adjusted flooring batching scheme. The prefabrication process digital monitoring unit (dynamic batching execution unit) is used to feed and mix the materials according to the flooring batching scheme to complete the preparation of the flooring materials.

[0106] Step 2: Intelligent leveling construction based on digital twin mapping: The digital twin construction management platform issues construction tasks to the intelligent laser leveling machine. The lidar inside the intelligent laser leveling machine starts scanning, generating point cloud data of the floor surface, and comparing it with the BIM model to calculate the flatness deviation of the initial construction area. The AI ​​decision tree takes temperature, flatness deviation, and concrete initial setting time as inputs, and outputs adjustment values ​​for the intelligent laser leveling machine's walking speed and vibration frequency, generating intelligent construction and adjustment instructions for the intelligent laser leveling machine's walking speed and vibration frequency. The intelligent laser leveling machine executes the floor leveling construction according to the walking speed and vibration frequency adjustment instructions.

[0107] Step 3, AI-powered closed-loop maintenance and acceptance: After the floor leveling construction is completed, the automatic spraying maintenance robot starts its operation, executes a differentiated spraying plan, and completes the spraying maintenance of the lawn; after the preset maintenance cycle is met, the AI ​​vision acceptance robot starts its operation. The lidar in the AI ​​vision acceptance robot collects the three-dimensional coordinate data of the floor surface, generates a flatness heat map based on the flatness data in the digital twin construction control platform, calculates the flatness compliance rate, and takes images of the floor surface through a camera. Crack information is obtained through the edge computing unit of the YOLOv8 crack recognition algorithm to obtain the acceptance result of the floor construction.

[0108] Furthermore, the method also includes: pre-simulation based on a digital twin model of the floor construction:

[0109] Import historical temperature and humidity data of the area to be constructed, as well as the physical property parameters of the raw materials to be used, input the AI ​​dynamic batching model, and output several flooring batching schemes;

[0110] Several sets of intelligent laser screed machine walking speed and vibration frequency parameters are preset and combined with several floor material preparation schemes. Based on the digital twin model of floor construction, the intelligent construction of different combinations is simulated. According to the results of floor crack rate and flatness compliance rate, the initial floor material preparation scheme and initial construction parameters are determined.

[0111] The digital twin construction management platform supports full-stage mapping from design to construction to acceptance. During the design phase, it simulates the risk of floor cracks under different temperatures with a simulation accuracy of ±0.1mm. During the construction phase, it compares the digital model with the on-site point cloud data in real time. When the deviation exceeds 2mm, it automatically triggers an early warning, which is simultaneously pushed to the on-site engineer's mobile device and the digital twin platform. The engineer must respond within 15 minutes. During the acceptance phase, it automatically generates a digital quality report, which includes a flatness heat map, crack location markings, and parameter traceability tables for each construction stage.

[0112] This invention provides a specific embodiment, based on the "Hikvision Wuhan Smart Industrial Park (Phase II)" project, and using the "Hikvision Wuhan Smart Industrial Park (Phase II) Section 1 E1 Factory Flooring Construction" as the application scenario (construction area 15,000㎡, construction period July, average daily temperature 35℃), deploying a laser-leveled high-temperature crack-resistant flooring intelligent construction system that integrates digital twin and AI decision-making, as detailed below:

[0113] Hardware deployment:

[0114] Intelligent material preparation subsystem: Deploy sand and gravel moisture content sensor (model SensirionSHT35), cement specific surface area meter (model Bettersize3000), and dynamic batching system (model Sany Heavy Industry SDP-100) in the prefabrication plant. All equipment is connected to the project's 5G network with a network bandwidth of ≥100Mbps to ensure stable data transmission.

[0115] The digital twin construction management and control subsystem uses a Dell PowerEdge R750 server, configured with 2×Intel Xeon Gold 6348 CPUs, 128GB of memory, and a 2TB SSD, to meet the needs of multi-source data processing and model calculation. Revit 2024 is deployed for BIM modeling, Unity 2023 for 3D visualization, and InfluxDB 2.0 for time-series data storage.

[0116] Intelligent construction equipment subsystem: equipped with 2 intelligent laser leveling machines (Wirtgen SP1500, with Beidou + GPS dual-mode positioning module), 3 automatic spraying maintenance robots (DJI RoboMaster S1, with customized SLAM path planning module and humidity sensor), and 1 AI vision inspection robot (Robotech RS-LIDAR-M1, equipped with 4K high-definition camera and YOLOv8 algorithm).

[0117] Data platform: Deployed on Alibaba Cloud server (ECSg7.xlarge instance, configured with 4 cores and 16GB memory), it connects to various subsystems through RESTful API interface, supports access from web (Chrome browser) and mobile (Android / iOS APP), with web response time ≤2s and mobile response time ≤3s.

[0118] Software debugging:

[0119] The AI ​​dynamic batching model was trained using 1200 sets of historical data (high-temperature flooring construction data in Wuhan from 2020 to 2024), with 80% used for training and 20% for validation. After 1000 iterations, the model error was 4.8%. Validated with 50 sets of new data, the model's output formula achieved a 98% material performance compliance rate. During debugging, the simulated sand and gravel moisture content increased from 5% to 7%, and the model output a formula adjustment command within 28 seconds, increasing the expansion agent dosage from 4% to 4.4%, validating the model's real-time response capability.

[0120] Digital twin construction control subsystem: (1) Revit-based BIM model: After importing the E1 factory building BIM model, the error was ≤1cm compared with the actual size on site, which meets the accuracy requirements. The accuracy of the simulated laser radar point cloud data superimposed on the BIM model is ≤2mm, which can accurately calculate the flatness deviation. The AI ​​decision tree was tested with 500 sets of on-site data. The average decision response time was 26s, the temperature prediction accuracy was 92%, and the flatness deviation was reduced by an average of 0.8mm after parameter adjustment. (2) Intelligent equipment collaboration: The instruction response time of the intelligent laser leveling machine and the digital twin model of the floor construction was tested. The average was 25s, which meets the requirement of ≤30s. The automatic spray maintenance robot had a path coverage of 100% and a repetition rate of 4%, which meets the requirement of ≤5%. The AI ​​visual acceptance robot was tested with 100 known cracks. The recognition rate was 99%, and the width measurement error was ≤0.02mm, which verified the acceptance accuracy.

[0121] After deploying a smart construction system for laser-leveled high-temperature crack-resistant flooring that integrates digital twins and AI decision-making, a case study of smart construction for laser-leveled high-temperature crack-resistant flooring integrating digital twins and AI decision-making was completed using this system. Figure 4 Specifically, it includes:

[0122] Step 1: Digital twin simulation and parameter optimization (supported by the digital twin construction management platform).

[0123] 1.1 Model Construction: Import the CAD drawings of the floor slab to be constructed into Revit to build a BIM model of the floor slab. The model is divided into 18 construction blocks according to the expansion reinforcement zone. Each block is labeled with the design elevation (±0.000), flatness requirement (4mm / 2m), and reinforcement density (12@200). Import the BIM model into the Unity engine, add a physics engine, and set the concrete drying shrinkage coefficient in the physics engine to 1.2×10⁻. 5 / ℃, simulating the drying shrinkage deformation process of concrete at high temperatures, to obtain a digital twin model of floor construction based on Revit and Unity;

[0124] 1.2 Data Input: Import meteorological data for Jiangxia District, Wuhan City from July 2020 to 2025. This data shows an average daily temperature of 35℃, a maximum temperature of 38℃, and humidity of 60±5%. Simultaneously, input the parameters of the raw materials to be used, including 99% purity of corundum, cement type P.O42.5R (specific surface area 340±10㎡ / kg), and sand and gravel moisture content of 5-7%.

[0125] 1.3 Pre-construction Optimization: Based on the digital twin model of the floor construction, different floor material mix schemes and construction parameter combinations were simulated, such as 30:50:20 and 32:50:18 gradations of corundum, 4% and 4.2% of expanding agent, and 0.9m / min and 1.0m / min of laser leveling speed. The pre-construction simulation revealed that when the corundum gradation was 32:50:18, the expanding agent content was 4.2%, the laser leveling speed was 0.9m / min, and the vibration frequency was 68Hz, the floor crack rate was 3.2%, and the flatness compliance rate was 98.5%, exhibiting the best overall performance. This was determined as the initial floor material mix scheme and initial construction parameters. Simultaneously, the AI ​​decision tree, based on meteorological data predicting that the temperature would rise to 36.5℃ around 19:00 on July 18th, proactively output a contingency plan to "reduce the walking speed to 0.85m / min and increase the vibration frequency to 69Hz" to avoid flatness deviations caused by temperature increases.

[0126] Step 2: AI-driven intelligent material preparation (supported by an intelligent material preparation subsystem).

[0127] 2.1 Raw material testing is conducted within the raw material IoT testing unit: When sand and gravel arrive at the site, their moisture content is automatically collected, and the result is 6.2%, exceeding the upper limit of the preset 5-7% range by 0.2%; when cement arrives, the specific surface area of ​​the cement is measured by a specific surface area analyzer to be 345㎡ / kg, meeting the requirement of 340±10㎡ / kg; when corundum arrives, the purity analyzer measures a purity of 99.2%, exceeding the design requirement of 99%. All physical property monitoring data for raw materials are uploaded to the data platform via 5G, with a data upload delay of 0.8s.

[0128] 2.2 AI Flooring Mixing Scheme Adjustment: The data platform inputs the physical properties of the raw materials and the ambient temperature into the AI ​​dynamic mixing model. The AI ​​dynamic mixing model, based on a deep neural network built using the TensorFlow framework, combines the current ambient temperature of 35℃ with preset target compressive strength and target crack rate to output the adjusted flooring mixing scheme (the original flooring mixing scheme corresponds to the initial flooring mixing scheme in step 1.3): 32% coarse aggregate (carborundum), 50% medium aggregate, 18% fine aggregate, 4.2% expansion agent, and a mixing time of 110s (10s longer than the initial formula). The model also outputs the reason for the formula adjustment: the sand and gravel moisture content increased by 0.2%, requiring an increase in the proportion of coarse aggregate by 2% to reduce moisture adsorption, and extending the mixing time by 10s to ensure uniform aggregate dispersion.

[0129] 2.3 Intelligent Preparation: The dynamic batching system automatically feeds materials according to the adjusted flooring batching plan. Each batch contains 43.2 kg of coarse corundum, 67.5 kg of medium corundum, 24.3 kg of fine corundum, 210 kg of cement, 8.8 kg of expanding agent, and 95 kg of water. The variable frequency mixer starts at 160 r / min. During mixing, a laser detector scans the aggregate dispersion in real time. The initial dispersion is 92% (not meeting the 95% pass standard). The data is fed back to the AI ​​model built into the AI ​​dynamic batching model in real time, which outputs a speed adjustment command within 25 seconds, increasing the speed to 180 r / min. A second dispersion test is conducted after 30 seconds, yielding 96%, meeting the pass requirements. After preparation, the system automatically labels the batch with batch number (2025071801) and formula parameters. The materials are then transported to the construction site in a temperature-controlled tanker (temperature controlled at 25-30℃) over 1.2 hours. During this time, the tanker's temperature sensor uploads temperature data in real time to ensure stable material performance.

[0130] Step 3: Intelligent leveling construction under digital twin mapping.

[0131] 3.1 Construction Preparation: One environmental sensor will be deployed every 200㎡ within Block ③, for a total of six sensors, to ensure real-time collection of temperature and humidity data. The digital twin construction management platform will issue construction tasks to the intelligent laser screed, prioritizing Block ③ of the E1 factory building, and simultaneously issue the BIM model fragment and initial construction parameters (walking speed 0.9m / min, vibration frequency 68Hz) for this block. The automatic spraying maintenance robot will receive the boundary coordinates of Block ③ from the platform and generate the optimal path based on the SLAM algorithm. The path will cover all areas of the block without duplication or omission.

[0132] 3.2 Real-time Positioning and Scanning: The intelligent laser screed uses GNSS to position itself at the starting point of block ③, with a positioning error of 2.5cm, meeting the accuracy requirement of ±3cm. The lidar initiates scanning, generating point cloud data of the floor surface, which is uploaded to the data platform every 30 seconds. The digital twin construction management platform obtains the point cloud data of the floor surface from the data platform and compares it with the BIM model to calculate the flatness deviation of the initial construction area. The result is 1.2mm, meeting the acceptable requirements.

[0133] 3.3 Dynamic Adjustment of Construction Parameters: When the laser screed reached the middle of block ③, the environmental sensor detected a temperature rise to 36.5℃, 1.5℃ higher than the initial temperature. Simultaneously, the lidar detected a flatness deviation of 2.8mm in this area, exceeding the acceptable threshold of 2mm. The AI ​​decision tree of the digital twin construction management platform immediately analyzed the data and determined that the deviation was caused by the increased temperature leading to reduced concrete fluidity, requiring adjustment of construction parameters. Within 28 seconds, the decision tree output adjustment instructions: reduce the walking speed from 0.9m / min to 0.85m / min and increase the vibration frequency from 68Hz to 69Hz. The instructions were transmitted to the laser screed via 5G. The actuator completed the parameter adjustment within 4 seconds, and 1 minute later, the lidar re-measured the flatness deviation of the area, finding it had decreased to 1.5mm, restoring it to acceptable level.

[0134] 3.4 Obstacle Recognition and Path Adjustment: During construction, the edge computing unit of the intelligent laser screed identified two protruding steel stirrups (5cm high) within the block, with a 100% accuracy rate. The system adjusted its path within 8 seconds, bypassing the protruding areas, and simultaneously uploaded the obstacle locations to the digital twin construction management platform, marking the area as a "key acceptance area" to remind users to focus on flatness inspection during later acceptance checks. Furthermore, the vibration sensor of the intelligent laser screed detected an amplitude of 0.55mm, exceeding the 0.5mm safety threshold, and immediately sent an alert to the data platform. The platform's AI diagnostic module analyzed the data and determined it to be minor wear on the vibratory motor bearing, pushing a repair plan to the on-site engineer. The engineer replaced the bearing according to the plan, taking 40 minutes, a 60% reduction compared to traditional manual inspection (1.5 hours).

[0135] 3.5 Process Monitoring: The digital twin platform displays the construction progress of block ③ in real time, updating every 30 seconds. When the construction reaches 65% of the block area, it displays the remaining construction time of 1.5 hours. The platform also displays environmental data and equipment status, with the temperature stable at 36.2-36.5℃, humidity at 62%, and the intelligent laser leveling machine's battery at 85%, with no fault alarms, ensuring continuous construction.

[0136] Step 4: Maintenance and acceptance of the AI ​​closed loop.

[0137] 4.1 Intelligent Maintenance: After the construction of Block ③ is completed (the intelligent laser screed is positioned at the end of Block ③ via GNSS), the data platform receives a "completion signal" and immediately instructs the automatic spray maintenance robot to start operation. The automatic spray maintenance robot enters Block ③ according to the preset path. The humidity sensor collects the surface humidity in real time, initially detecting 75%, which is lower than the requirement of 80%. The AI ​​module determines that this area is a low-humidity area and automatically increases the spray rate from 15L / min to 18L / min. After 20 minutes, the humidity sensor detects that the humidity has risen to 82%, and the AI ​​module adjusts the spray rate to 16L / min to maintain stable humidity. During the maintenance period, the data platform generates a maintenance report every hour, which includes humidity change curves, robot working status, and spray rate adjustment records. Engineers can view these reports in real time via the web interface.

[0138] 4.2 AI Acceptance: After 14 days of curing, the AI ​​vision acceptance robot entered block ③ and scanned along a preset path at a speed of 0.5 m / s. LiDAR collected flatness data and generated a flatness heatmap for block ③, showing a maximum deviation of 3.2 mm / 2 m and a flatness compliance rate of 99.2%. A high-definition camera captured surface images, and the YOLOv8 crack detection algorithm automatically detected cracks, finding two cracks with widths of 0.12 mm and 0.1 mm respectively, both ≤0.2 mm, achieving a 100% recognition rate. The acceptance robot simultaneously correlated crack locations with construction data, such as the temperature of the area with the 0.12 mm crack being 36.5℃ and the laser leveling machine's travel speed being 0.85 m / min during construction, providing a basis for subsequent quality analysis.

[0139] 4.3 Data Archiving: Upon completion of acceptance, the data platform automatically generates a digital quality archive for Block ③. The archive includes raw material testing reports, dynamic curves of construction parameters, a flatness heat map, crack photos, and acceptance results. The archive is stored along the path "Hikvision Project - E1 Plant - Block ③ - 20250718," allowing the construction and supervision parties to view and review it online through the smart construction site platform. Simultaneously, the data platform's AI data analysis module mines the construction data for Block ③, discovering a pattern that "for every 1°C increase in temperature, the flatness deviation increases by an average of 0.3mm." This pattern is incorporated into the parameter template for use in subsequent block construction.

[0140] This invention also provides comparative examples. Two blocks of the same area (2000㎡) were selected in the E1 factory building, and construction was carried out using the intelligent method of this invention (experimental group) and the traditional process (control group), respectively. The comparison results are as follows:

[0141] Comparison Indicators | Control Group (Traditional Process) | Experimental Group (This Invention) | Improvement Rate |

[0142] |Materials preparation time| 2.5h / batch| 1.7h / batch| 32%|

[0143] |Number of construction workers|15 people / block|9 people / block|40% (reduction)|

[0144] |Smoothness compliance rate (4mm / 2m)|78.3%|99.2%|26.7%|

[0145] |Crack rate (≥0.1mm)|22.5%|3.2%|85.8% (reduced)|

[0146] |Inspection time|8h / 2000㎡|2h / 2000㎡|75%|

[0147] |Rework loss rate|16.8%|3.5%|80% (reduced)|

[0148] |Quality traceability time|48h|5min|99.8% (reduced)|

[0149] Through the system and method proposed in this invention, the construction of the E1 factory floor achieves the following improvements: Quality Enhancement: 99.2% flatness compliance rate (traditional 80%), 3.2% crack rate (traditional 22.5%), and a Mohs hardness of 8.2 (traditional 6.5), all exceeding design requirements; Efficiency Enhancement: 30% improvement in material preparation efficiency (AI formula adjustment replaces manual calculation, reducing time from 2 hours to 30 seconds), 25% improvement in construction efficiency (intelligent equipment replaces 10 on-site workers, reducing the construction period for 15,000㎡ from 30 days to 22 days), and 80% improvement in acceptance efficiency (15,000㎡). Acceptance takes only 1 day, compared to 3 days traditionally; cost reduction: raw material loss rate is reduced from 5% to 2% (AI-precise material batching reduces waste), labor costs are reduced by 40% (reducing the number of workers by 10, with an average daily wage of 300 yuan, saving 60,000 yuan in labor costs for 15,000 square meters of construction), and operation and maintenance costs are reduced by 30% (digital archives facilitate later traceability, and troubleshooting time is reduced by 80%); intelligent enhancement: achieving "no human intervention" in material preparation and construction parameter adjustment, with AI decision-making accounting for more than 90%, and humans only need to handle abnormal situations, which is in line with the development trend of "less human and unmanned" intelligent construction.

[0150] It is evident that this invention is significantly superior to traditional processes in terms of efficiency, quality, cost, and traceability, and it achieves full digitalization and intelligentization of the construction process, which aligns with the development trend of intelligent construction.

[0151] The core innovative value of this invention lies in:

[0152] Technological integration and innovation: For the first time, digital twin, AI decision-making, and IoT technologies are deeply integrated with traditional flooring construction to build a full-process intelligent closed loop of "perception-decision-execution-feedback", breaking through the bottleneck of data fragmentation in each link of "materials-construction-maintenance" and realizing the transformation from "experience-driven" to "data-driven".

[0153] Deepening AI Decision-Making: AI is not only used for parameter adjustment, but also has functions such as real-time learning, risk prediction, and fault diagnosis. For example, the AI ​​model iterates every 24 hours to continuously optimize the formula; the AI ​​decision tree predicts temperature changes 2 hours in advance to avoid quality fluctuations; the AI ​​diagnostic module quickly locates equipment faults, shortens maintenance time, and significantly improves the system's intelligence.

[0154] With broad application value, it provides a feasible solution for intelligent construction of industrial floors in high-temperature areas, and can be extended to smart industrial parks, electronics factories, high-end warehouses, and other scenarios. Through system expansion and parameter template adjustment, it can quickly adapt to the characteristics of raw materials and environmental conditions of different projects, with an adaptability of ≥90%, driving the transformation of the construction industry from traditional construction to intelligent construction.

[0155] Key innovations and technological effects:

[0156] AI-driven dynamic material adaptation technology breaks through the limitations of traditional fixed formulations, constructing an AI-driven ingredient formulation closed loop of "detection-decision-execution-feedback". By detecting the characteristics of raw materials and environmental parameters through the Internet of Things, the AI ​​model adjusts the formulation in real time, solving the problem of unstable material performance caused by "raw material fluctuations + high temperature environment".

[0157] Digital twin full-process mapping and AI prediction technology: This is the first time that digital twins and AI prediction have been combined, spanning the entire process from "pre-construction to acceptance," enabling visualization of the construction process, traceability of parameters, and early warning of problems. In the pre-construction phase, based on historical meteorological data and a physics engine, construction parameters are optimized in advance to avoid blind construction. In the construction phase, an AI decision tree predicts temperature changes two hours in advance and outputs parameter adjustment plans. For example, if a 2°C temperature increase is predicted, walking speed is reduced by 0.05 m / min in advance to prevent the flatness deviation from expanding. In the acceptance phase, AI visual recognition and data association enable rapid tracing of quality problems. In this embodiment, the risk of cracks in the corner of block ③ was identified in advance during the pre-construction phase. During construction, the vibration frequency in this area was increased by 1 Hz, ultimately resulting in no cracks in that area. In the construction phase, through AI prediction and parameter adjustment, the flatness compliance rate increased from the traditional 80% to 99.2%, and rework losses decreased from 16.8% to 3.5%.

[0158] The intelligent equipment cluster AI collaborative technology constructs an AI collaborative operation system consisting of an "intelligent laser screed machine + automatic maintenance robot + AI acceptance robot." The devices communicate and coordinate commands through a data platform, eliminating the need for manual intervention. After completing a section of work, the laser screed machine automatically sends a "completion signal" to the maintenance robot, which then enters the area within 10 minutes to begin maintenance, reducing the time by 30 minutes compared to manual notification. The maintenance robot performs differentiated spraying based on humidity distribution, increasing spray volume in low-humidity areas by 20%, raising the floor humidity uniformity from the traditional 70% to 90%. The acceptance robot automatically links the identified crack locations to construction data, reducing traceability time from 48 hours to 5 minutes. Furthermore, the intelligent equipment possesses autonomous fault diagnosis and remote maintenance capabilities. When the laser screed machine malfunctions, the AI ​​diagnostic module pushes repair solutions, reducing repair time by 40% and significantly improving equipment utilization.

[0159] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention 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; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A smart construction system for laser-leveled high-temperature crack-resistant flooring integrating digital twin and AI decision-making, characterized in that, The data platform serves as the data interaction center, connecting three subsystems: The intelligent material preparation subsystem is used to monitor the physical properties of concrete raw materials in real time, obtain the flooring batching plan, and complete the preparation of flooring materials according to the flooring batching plan; The intelligent construction equipment cluster, including intelligent laser screed machines, automatic maintenance robots, and AI vision inspection robots, is used for floor construction and data collection. The collected data includes: both the intelligent laser screed and the AI ​​vision inspection robot are equipped with GNSS positioning modules to collect real-time positioning data from the AI ​​vision inspection robot, and are equipped with LiDAR to collect three-dimensional coordinate data of the floor surface; the AI ​​vision inspection robot is additionally equipped with a high-definition camera to capture images of the floor surface and an edge computing module deploying the YOLOv8 crack recognition algorithm to annotate crack information in the floor surface images. The digital twin construction management platform is used to combine the digital twin model of floor construction built based on Revit and Unity with construction data to map the floor construction of the intelligent construction equipment cluster and assist in the floor construction process. The data platform also receives real-time environmental temperature and humidity data from the field sensor units.

2. The intelligent construction system for laser-leveled high-temperature crack-resistant flooring integrating digital twin and AI decision-making as described in claim 1, characterized in that, The intelligent material preparation subsystem includes: The raw material IoT detection unit is used to monitor the physical properties of concrete raw materials, and the monitoring data is uploaded to the data platform in real time. The physical properties of the raw materials include, but are not limited to, the moisture content of sand and gravel, the purity of corundum, and the specific surface area of ​​cement. The AI ​​dynamic batching model includes a deep neural network model built on the TensorFlow framework. Under the preset target compressive strength and target crack rate conditions, the deep neural network model outputs a flooring batching scheme based on the physical properties of the raw materials and the ambient temperature obtained from the data platform. The flooring batching scheme includes the gradation ratio of corundum, the dosage of expansion agent, and the mixing time. The prefabrication process digital monitoring unit is used to complete the preparation of flooring materials according to the flooring batching plan and to obtain the aggregate dispersion in real time during the preparation process.

3. The intelligent construction system for laser-leveled high-temperature crack-resistant flooring integrating digital twin and AI decision-making as described in claim 2, characterized in that, The prefabrication process digital monitoring unit includes: a feeding valve, a variable frequency mixer, and an aggregate dispersion laser detector; wherein, the automatic feeding valve feeds materials according to the floor material distribution plan, the variable frequency mixer starts mixing, and the aggregate dispersion laser detector monitors the aggregate dispersion during the mixing process.

4. The intelligent construction system for laser-leveled high-temperature crack-resistant flooring integrating digital twin and AI decision-making as described in claim 3, characterized in that, The AI ​​dynamic batching model incorporates an AI model to acquire monitoring data on ambient temperature and the physical properties of concrete raw materials. When the real-time physical properties of raw materials and / or ambient temperature fluctuations exceed a preset amplitude, the deep neural network is invoked to output an updated floor batching scheme. It also acquires monitoring data on aggregate dispersion in the digital monitoring unit of the prefabrication process. When the aggregate dispersion does not meet the preset standard, a speed adjustment command is triggered to control the variable frequency mixer to increase its speed.

5. The intelligent construction system for laser-leveled high-temperature crack-resistant flooring integrating digital twin and AI decision-making as described in claim 1, characterized in that, The process of building a digital twin model of floor construction based on Revit and Unity includes: Import the CAD drawings of the flooring project into Revit, build the flooring BIM model, and annotate the flooring BIM model according to the target construction parameters of the flooring; the annotation content includes, but is not limited to, flooring zoning, design elevation, and material properties; By adding a physics engine to the floor BIM model using Unity to simulate the high-temperature drying shrinkage process, a digital twin model of the floor construction is obtained.

6. The intelligent construction system for laser-leveled high-temperature crack-resistant flooring integrating digital twin and AI decision-making as described in claim 1, characterized in that, The digital twin construction management platform assists in the construction of flooring, including: It receives real-time positioning data from the intelligent laser screed and three-dimensional coordinate data of the floor surface, calculates the flatness deviation, and has an AI decision tree trained on gradient boosting tree. With temperature, flatness deviation and concrete initial setting time as input, it outputs the adjustment values ​​of the intelligent laser screed's walking speed and vibration frequency, and generates instructions for adjusting the intelligent laser screed's walking speed and vibration frequency. The system receives environmental humidity data from the data platform, combines it with the built-in digital twin model of floor construction, generates a heat map of humidity distribution on the floor surface, and sends it to the automatic spraying maintenance robot. The system receives real-time positioning data from the AI ​​vision inspection robot, three-dimensional coordinate data of the floor surface, and crack information. Based on the real-time positioning data and the three-dimensional coordinate data of the floor surface, it generates a floor flatness heat map, calculates the flatness pass rate, and combines the crack information in the floor surface image to obtain the acceptance result of the floor construction.

7. The intelligent construction system for laser-leveled high-temperature crack-resistant flooring integrating digital twin and AI decision-making as described in claim 6, characterized in that, The automated spraying maintenance robot is equipped with a path planning module to plan its travel route; a variable frequency sprayer; and a humidity sensor to acquire real-time humidity data of the floor surface. A built-in AI module constructs a differentiated spraying scheme based on a heat map of the floor surface humidity distribution. During the robot's movement, the spray volume of the variable frequency sprayer is adjusted based on the real-time surface humidity data and the differentiated spraying scheme to implement floor spraying maintenance. The differentiated spraying scheme divides the spraying area into high-humidity, medium-humidity, and low-humidity zones, and uses differentiated spray volumes for different spraying zones.

8. The intelligent construction system for laser-leveled high-temperature crack-resistant flooring integrating digital twin and AI decision-making as described in claim 1, characterized in that, The data platform also stores quality parameters and equipment safety parameter thresholds for early warning. Quality parameters include flatness deviation and crack width, while equipment safety parameters include the amplitude of the intelligent laser leveling machine and the power consumption of the automatic spray maintenance robot. When the quality parameters and / or equipment safety parameters exceed the threshold, an early warning message is automatically sent to the management personnel.

9. A method for intelligent construction of laser-leveled high-temperature crack-resistant flooring integrating digital twin and AI decision-making, implemented based on the intelligent construction system for laser-leveled high-temperature crack-resistant flooring integrating digital twin and AI decision-making as described in any one of claims 1-8, comprising the following steps: After the raw materials for flooring construction arrive on site, the intelligent material preparation subsystem completes the preparation of the flooring materials, including: after the raw materials arrive on site, the raw material IoT detection unit collects the physical characteristics of the raw materials. After the collection is completed, the physical characteristic data of the raw materials is uploaded to the data platform in real time. The data platform inputs the physical characteristic data of the raw materials and the real-time ambient temperature into the AI ​​dynamic batching model and outputs the adjusted flooring batching scheme. The prefabrication process digital monitoring unit is used to feed and mix the materials according to the flooring batching scheme to complete the preparation of the flooring materials. The digital twin construction management platform issues construction tasks to the intelligent laser screed. The laser radar inside the intelligent laser screed starts scanning, generating point cloud data of the local surface of the floor. It compares this data with the digital twin model of the floor construction and calculates the flatness deviation of the initial construction area. The AI ​​decision tree takes temperature, flatness deviation, and concrete initial setting time as inputs and outputs adjustment values ​​for the intelligent laser screed's walking speed and vibration frequency, generating intelligent construction and adjustment instructions for the intelligent laser screed's walking speed and vibration frequency. The intelligent laser screed executes the floor screed construction according to the walking speed and vibration frequency adjustment instructions. After the floor leveling construction is completed, the automatic spraying maintenance robot starts its operation, executes a differentiated spraying plan, and completes the spraying maintenance of the lawn. After the preset maintenance cycle is met, the AI ​​vision acceptance robot starts its operation. The lidar in the AI ​​vision acceptance robot collects the three-dimensional coordinate data of the floor surface, generates a flatness heat map based on the flatness data in the digital twin construction control platform, calculates the flatness compliance rate, and takes images of the floor surface through a camera. The edge computing unit of the YOLOv8 crack recognition algorithm is used to obtain crack information and obtain the acceptance results of the floor construction.

10. A smart construction method for laser-leveled high-temperature crack-resistant flooring integrating digital twin and AI decision-making, further comprising: Pre-simulation based on digital twin model of floor construction; Import historical temperature and humidity data of the area to be constructed, as well as the physical property parameters of the raw materials to be used, input the AI ​​dynamic batching model, and output several flooring batching schemes; Several sets of intelligent laser screed machine walking speed and vibration frequency parameters are preset and combined with several floor material preparation schemes. Based on the digital twin model of floor construction, the floor screed construction of different combinations is simulated, and the floor crack rate and flatness compliance rate of each combination are obtained. The combination with the best comprehensive results of floor crack rate and flatness compliance rate is selected as the initial floor material preparation scheme and initial construction parameters.

Images