Construction process for high-precision wear-resistant ground of super-large-area complex factory building

Abstract

The invention relates to the field of building construction, and discloses an ultra-large-area complex factory building high-precision wear-resistant ground construction process which comprises the following steps: establishing a predetermined height distribution model of a target ground based on topographic data and design requirements of a construction area, and estimating an actual construction height in real time in combination with sensing data acquired in a construction process; a construction error model used for reflecting the error between the target height and the actual construction height is established; and by taking the construction error model as input, performing nonlinear dynamic modeling operation on the construction equipment, establishing a multi-body dynamic model of the construction equipment, and extracting a control variable related to the motion state of the equipment as input of an equipment control strategy. By introducing a simulation optimization and real-time feedback mechanism, the dynamic optimization of a construction control strategy and material distribution is realized, the construction precision, the material utilization rate and the equipment efficiency are effectively improved, and the efficient and accurate execution of the construction process is ensured.

Description

Technical Field

[0001] This invention relates to the field of building construction, specifically to a construction process for high-precision wear-resistant flooring in large-area, complex factory buildings. Background Technology

[0002] In modern ground paving construction, precise control strategies and material distribution are crucial to project quality. Traditional construction methods often rely on manual operation and experience-based judgment, which not only leads to low construction accuracy but also easily causes material waste or excessive equipment load due to environmental changes or different equipment conditions.

[0003] Currently, some technologies have attempted to model the construction process using artificial intelligence or machine learning, optimizing it through data-driven methods. The main advantage of these technologies lies in their ability to predict error trends to some extent and make predictions and adjustments based on historical data. Furthermore, some systems have implemented basic construction path planning and material distribution scheduling, contributing to improved construction efficiency and accuracy.

[0004] However, existing technologies also have significant shortcomings. First, most technologies rely on a single predictive model and cannot adjust control strategies in real time according to changes in the construction environment, resulting in delayed error correction. Second, traditional material distribution methods are mostly based on rule-based allocation patterns, ignoring the impact of on-site changes on material usage, leading to resource waste or uneven material utilization.

[0005] Finally, existing simulation tools are usually just pre-construction verification tools and cannot be deeply integrated with actual construction instructions, resulting in a disconnect between simulation and reality. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a construction process for high-precision wear-resistant flooring in ultra-large-area complex factory buildings, solving the problem of lack of dynamic adaptability in construction control strategies and material distribution in existing technologies.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] The construction process for high-precision wear-resistant flooring in ultra-large and complex factory buildings includes the following steps:

[0009] Based on the terrain data and design requirements of the construction area, a predetermined height distribution model of the target ground is established. Combined with the sensor data collected during the construction process, the actual construction height is estimated in real time to establish a construction error model to reflect the error between the target height and the actual construction height.

[0010] Using the construction error model as input, nonlinear dynamic modeling is performed on the construction equipment to establish a multibody dynamic model of the construction equipment. Control variables related to the motion state of the equipment are extracted and used as inputs for the equipment control strategy.

[0011] Based on the aforementioned error model and control variables, a construction control strategy is established with the goal of minimizing the deviation between the target height and the actual height. An adaptive control mechanism is designed so that the control signal can be dynamically adjusted according to real-time error feedback.

[0012] While implementing control strategies, topology optimization of the distribution of construction materials is carried out based on the upper limit of total material usage and ground performance requirements;

[0013] The sensor data collected during construction is input into a deep learning model for feature extraction and time series learning in order to predict the error trend in future construction processes.

[0014] Based on the error prediction results output by the deep learning model and combined with the error information fed back in real time, the equipment control strategy and material distribution optimization scheme are dynamically adjusted.

[0015] The control strategy and material distribution results are numerically solved and dynamically optimized based on simulation optimization methods to generate the optimal construction control instruction set, which is then used to guide actual ground paving operations.

[0016] Preferably, the height distribution model includes:

[0017] The height distribution model includes a digital terrain model generated based on three-dimensional terrain scanning data, and determines the target construction height by combining the regional functions, load requirements and ground flatness requirements marked in the design drawings;

[0018] The model is built according to the principle of regional grid division, forming a predetermined height matrix composed of multiple cells, with each cell corresponding to a spatial positioning coordinate and a target elevation;

[0019] The height distribution model supports dynamic correction of target height values ​​based on real-time sensor data during construction, in order to cope with changes in construction terrain, settlement, and human adjustments.

[0020] Preferably, the construction error model includes:

[0021] The construction error model includes a spatial error distribution map built based on height difference data, which reflects the magnitude and direction of the deviation between the actual construction height and the target height in different areas;

[0022] The error model, combined with time series recording, models the evolution of errors as construction progresses, and identifies areas of error accumulation or repeated correction.

[0023] The model provides error feedback signals to the control system, which are used to adjust the equipment movement path, construction rhythm and material distribution in real time.

[0024] Preferably, the multibody dynamics model includes:

[0025] The multibody dynamics model includes physical modeling of key structural components of the construction equipment, such as the chassis, telescopic boom, transmission mechanism, and operating tools, taking into account their mass distribution, geometric connection relationships, and kinematic constraints.

[0026] The model defines the hinges and flexible connections between each connection point and embeds a simulation mechanism for the response behavior to the contact force on the construction ground.

[0027] The model uses a dynamic solver to calculate in real time the nonlinear relationship between the device's attitude change, displacement response, and control input, providing a basis for control signal optimization.

[0028] Preferably, the control strategy includes:

[0029] A control function is constructed with the objective of minimizing the sum of squared differences between the target ground height and the actual construction height, and the equipment movement speed, path change range and material spraying frequency are used as optimization variables.

[0030] The control strategy adopts a hierarchical control architecture, including a global path planning module and a local error compensation module. The former is used to generate the overall construction path, and the latter is used to correct the current error in real time.

[0031] Error information is collected in real time by sensors deployed on construction equipment, and the information is fed back to the control module to dynamically adjust control commands, thereby realizing a closed-loop control mechanism.

[0032] Preferably, the optimal distribution scheme includes:

[0033] The optimal distribution scheme includes factors such as compressive strength, abrasion resistance coefficient, hydration reaction rate, and temperature and humidity conditions of the construction environment, forming a material selection library;

[0034] By using a multi-objective topology optimization algorithm, the material ratio, pouring sequence and local material addition intensity of each grid cell are determined while ensuring construction accuracy and ground performance.

[0035] The optimization results are output in the form of two-dimensional or three-dimensional maps, providing the types, quantities and layout strategies of materials required for each construction sub-area, to guide the actual material placement operation.

[0036] Preferably, the feature extraction and temporal learning include:

[0037] Feature extraction and temporal learning include using convolutional neural networks to extract spatial features from height maps, error maps, and equipment status images collected by sensors at the construction site;

[0038] By combining long short-term memory neural networks, the changing trend of error values ​​within a continuous construction period is modeled, and its temporal characteristics and evolution patterns are learned.

[0039] The model has the ability to identify error growth, predict error peaks, and provide early warnings of high-risk areas, which can be used to enhance the control system's ability to cope with construction in complex locations.

[0040] Preferably, the error trend includes:

[0041] Error trends include the average slope of error growth over time, the identification of local error burst areas, and the propagation trajectory of error from high-value areas to the surrounding areas;

[0042] The trend modeling results combine construction path, equipment load and environmental change factors to predict and classify the spatiotemporal range of error exceeding the limit.

[0043] Error trend data serves as input variables for equipment motion path optimization, material distribution reconfiguration, and control strategy reconstruction.

[0044] Preferably, the construction control instruction set includes:

[0045] The construction control instruction set includes equipment movement path control instructions, operation tool execution parameters, material spraying rhythm and ratio settings, and error correction strategy call identifiers.

[0046] The instruction set is integrated into the central control system in a structured format, and is distributed to various intelligent devices through the construction scheduling platform, and supports real-time updates and conflict resolution mechanisms;

[0047] The control instruction set also incorporates a traceability mechanism to record the status feedback and error response before and after the execution of each instruction, which is used for subsequent construction quality assessment and optimization model training.

[0048] Preferably, the deep learning model includes:

[0049] The deep learning model includes a multi-layer convolutional structure at the front end to extract spatial feature patterns from construction status images, error distribution maps, and material laying maps;

[0050] The intermediate layer uses a gated recurrent neural network to improve the model's prediction accuracy regarding the direction and rate of error evolution.

[0051] This invention provides a construction process for high-precision wear-resistant flooring in ultra-large, complex factory buildings. It offers the following advantages:

[0052] 1. This invention employs a technical solution of "error prediction and real-time feedback linkage control based on deep learning," achieving the technical effect of real-time dynamic correction of control parameters and material distribution during construction. Compared to existing control methods that rely on static preset parameters or single model prediction, this invention solves the problems of being unable to adapt to rapid changes in errors and delayed adjustment responses during construction.

[0053] 2. By introducing a technical solution of "simulation optimization method to jointly solve control strategy and material distribution," this invention achieves globally optimal scheduling of construction path and material usage plan. Compared with traditional phased optimization or manual experience-based configuration methods, this avoids the disconnect between control strategy and material distribution, and solves the problems of material waste and equipment coordination imbalance.

[0054] 3. This invention constructs a complete closed-loop control system based on a three-dimensional coupling mechanism of "error prediction + feedback correction + simulation optimization," enabling construction control commands to be dynamically generated and executed according to actual conditions. Traditional methods often neglect the real-time disturbance response of the construction site; this technology effectively solves the technical shortcomings of slow response to environmental changes and rigid control.

[0055] 4. A path control mechanism that integrates high-fidelity simulation models and numerical optimization algorithms ensures that the generated control commands are not only optimal but also implementable, truly achieving an engineering closed loop from modeling to control. Existing technologies mostly remain at the offline simulation level, lacking command closed-loop verification. This invention achieves breakthroughs in both executability and accuracy, solving the problem of the disconnect between simulation and practical application. Attached Figure Description

[0056] Figure 1 This is a flowchart of the process steps of the present invention. Detailed Implementation

[0057] The technical solutions in 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, and 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.

[0058] Please see the appendix Figure 1 This invention provides a construction process for high-precision wear-resistant flooring in large-area, complex factory buildings, including the following steps:

[0059] S1. Based on the terrain data and design requirements of the construction area, establish a predetermined height distribution model of the target ground, and combine it with the sensor data collected during the construction process to estimate the actual construction height in real time, so as to establish a construction error model to reflect the error between the target height and the actual construction height.

[0060] First, a predetermined height distribution model of the target ground needs to be established based on the terrain data and design requirements of the construction area. The accuracy of this model is the foundation for subsequent construction precision control. Specifically, the terrain data of the construction area is obtained through high-precision 3D scanning, and combined with the design requirements for ground flatness, bearing capacity, and load distribution, a target ground height distribution model is formed. The establishment of this model provides a clear target standard for the entire construction process.

[0061] After establishing a predetermined height distribution model for the target ground, we also need to estimate the actual construction height in real time and, based on this, construct a construction error model to reflect the difference between the target height and the actual construction height. The error model is designed to dynamically adjust the distribution of construction equipment and materials, thereby maintaining the flatness and wear resistance of the ground throughout the construction process and ensuring construction quality.

[0062] In this embodiment, topographic data of the construction area is first obtained through 3D terrain scanning and analysis of design requirements. Topographic data typically includes height information, terrain undulation, settlement, and ground functional requirements based on design needs (such as load-bearing capacity and ground flatness level). This data, collected by sensors, is input into a computational model to further generate a target ground height distribution model.

[0063] Based on the specific shape of the construction area, the target ground elevation distribution model divides the area into multiple sub-regions, each corresponding to a precise elevation calibration value. These elevation values ​​form a predetermined elevation matrix, representing the target construction elevation for each region. These target elevations are determined based on ground functional requirements, load requirements, and accuracy requirements in the design drawings. In the actual implementation process, digital terrain modeling technology is used, combined with various geological conditions of the region, to optimize the distribution of target elevations, making them as accurate as possible to meet the design requirements.

[0064] Alternatively, the target ground height distribution model can be meshed into multiple grid cells. Each grid cell corresponds to a specific target height value. Based on real-time sensor data acquired during construction, the model also supports dynamic adjustment of the height value to adapt to possible changes during construction, such as local settlement or equipment relocation.

[0065] After establishing the target ground height distribution model, the next step is to construct a construction error model. In this embodiment, the construction error model dynamically calculates the error value by monitoring the deviation between the actual construction height and the target ground height in real time. To achieve this, various high-precision sensors are installed on the construction equipment during the construction process. These sensors are used to collect ground height data of the construction area in real time and compare it with the predetermined target height to calculate the error between the target height and the actual height.

[0066] The construction error model not only records the error at each point during construction but also models the evolution of the error over time. Specifically, the error model identifies the trend and accumulation of errors through time series analysis. In some embodiments, the error model, based on regression analysis or other numerical optimization methods, further refines error prediction by modeling the clustering effect of multiple error points. This enables the model to not only reflect immediate errors during construction but also predict potential error trends.

[0067] The error model is constructed using the following formula:

[0068] ;

[0069] in, Location and time Construction errors at any given moment; This indicates the actual ground height measured during construction. Indicates in the target area The predetermined target altitude of the location.

[0070] By continuously monitoring and updating error values, the construction system can dynamically adjust equipment control strategies and material distribution schemes, thereby minimizing errors and achieving higher precision construction.

[0071] In some embodiments, the construction error model is not only used to adjust equipment behavior in real time, but also incorporates multi-level optimization of equipment and materials. Through real-time feedback from the error model, the equipment control system can adjust the equipment movement path, material application amount, and construction speed based on the current error value, thereby improving the accuracy and efficiency of the construction process.

[0072] Specifically, when the error model detects that the error in a certain construction area exceeds the predetermined range, the system will automatically issue a warning and make necessary adjustments. For example, if the error value in a certain area is too large, the system will instruct the construction equipment to adjust its path or increase the amount of materials used to compensate. In this way, the construction process can adapt to environmental changes and correct deviations in a timely manner, ensuring that the ground ultimately meets the predetermined target requirements.

[0073] Based on the output of the error model, the system further incorporates an adaptive control mechanism. When the error deviation reaches a certain set threshold, the system activates the error correction module. This module dynamically adjusts the equipment control signals according to the current error value, changing the movement path of the construction equipment or the working mode of the tools to reduce the impact of the error. This control signal is continuously updated through a closed-loop feedback mechanism until the error reaches an acceptable range.

[0074] For example, when errors accumulate in localized areas, the equipment control system will adjust the equipment's operating trajectory or the pressure of the tools, and redistribute materials to quickly correct the errors. This close integration of the error model and control system allows for effective control of errors during construction, thereby maximizing the accuracy of ground construction.

[0075] S2. Using the construction error model as input, perform nonlinear dynamic modeling on the construction equipment to establish a multibody dynamic model of the construction equipment, and extract control variables related to the motion state of the equipment as input for the equipment control strategy.

[0076] Precise control of construction equipment is crucial for achieving construction accuracy. To optimize equipment behavior and improve construction efficiency, this invention employs nonlinear dynamics modeling to model the construction equipment and further establishes a multibody dynamics model. Specifically, the construction error model serves as input, allowing for the precise extraction of the equipment's motion state and control variables, which are then used as input to the equipment control strategy. This process is key to achieving high-precision equipment operation.

[0077] In this process, the construction error model provides feedback information related to equipment behavior. After nonlinear dynamic modeling, the multibody dynamics model of the construction equipment adjusts the equipment's motion trajectory and operating parameters based on this feedback. This method ensures that the construction equipment automatically adjusts itself according to real-time error feedback at different construction stages, thereby minimizing construction errors.

[0078] In this embodiment, the dynamic characteristics of the construction equipment are first nonlinearly modeled using the real-time output of the construction error model. Construction equipment typically consists of multiple interconnected components, such as a chassis, telescopic boom, transmission device, and working tools. The motion between these components is affected by interaction forces and constraints; therefore, the overall motion characteristics of the equipment exhibit nonlinearity.

[0079] When performing nonlinear dynamic modeling, it is first necessary to define the physical parameters of each component of the equipment, such as mass, stiffness, and damping, and consider the connection relationships between the components. The motion state of each component can be described by physical quantities such as position, velocity, and acceleration. In order to accurately reflect the various nonlinear dynamic effects that the equipment may generate during construction, nonlinear forces, such as elastic forces, frictional forces, and feedback forces generated by contact with the ground, need to be introduced into the model.

[0080] The multibody dynamics model of the construction equipment is derived based on the Lagrange method. The equations of motion for the multibody system can be expressed as:

[0081] ;

[0082] in, This is the mass matrix of the equipment, reflecting the mass distribution of each component; These are the generalized coordinates of the various components of the equipment, representing the degrees of freedom of the equipment's motion; The acceleration of each component of the equipment; The Coriolis force matrix is ​​used to account for non-inertial forces during the equipment's movement. The term represents gravity, describing the gravitational effects on each component. To control torque, it represents the force applied to each component.

[0083] By solving the multibody dynamics equations, we can obtain the precise motion trajectory of the equipment, including the position, velocity, and acceleration information of each component. This information will serve as the basis for subsequent adjustments to the control strategy.

[0084] In the multibody dynamics model of construction equipment, in addition to basic parameters such as mass and torque, it is also necessary to extract control variables related to the motion state of the equipment. These control variables typically include parameters such as the equipment's moving speed, path, operating angle of the working tools, and pressure.

[0085] In this embodiment, the extraction of control variables is mainly based on the real-time motion state of the equipment during construction. These motion states are obtained through sensor feedback, including data such as the equipment's position, attitude, speed, and acceleration. In actual construction, the equipment senses information such as ground height and construction load through sensors, and integrates this feedback with data in the construction error model to obtain the precise motion state of the equipment in real time.

[0086] Alternatively, the extraction of control variables can be optimized by coupling them with the equipment dynamics model. For example, in a multibody dynamics model, changes in the equipment's attitude and the load of the working tools will affect the equipment's motion state. By analyzing these influencing factors, the selection and adjustment of control variables can be further refined. Specifically, control variables can be expressed as:

[0087] ;

[0088] in, This represents the control strategy vector of the device; The moving speed of the equipment; The working angle of the equipment and tools; The force applied to a work tool.

[0089] The equipment motion trajectory calculated using a multibody dynamics model and the extracted control variables will serve as inputs to the equipment control strategy. The goal of the equipment control strategy is to minimize construction errors by adjusting the equipment's motion behavior. Specifically, when the construction error model detects an error in a certain construction area, the control system will automatically adjust the equipment's motion path, the pressure of the work tools, and other control variables to minimize the error.

[0090] The optimization of the control strategy is based on a dynamic feedback mechanism. The equipment control system will continuously adjust the control input according to real-time feedback information, thereby ensuring that the construction equipment always maintains the optimal working state throughout the construction process.

[0091] For example, if the construction error measured in a certain area is large, the control system will adjust the working speed of the equipment or the pressure of the working tools based on the equipment motion data output by the multibody dynamics model to ensure that the equipment can effectively compensate the construction ground.

[0092] S3. Based on the error model and control variables, a construction control strategy is established with the goal of minimizing the deviation between the target height and the actual height, and an adaptive control mechanism is designed so that the control signal can be dynamically adjusted according to real-time error feedback.

[0093] In the aforementioned embodiments, a target ground predetermined height distribution model and a real-time construction error model have been constructed. A multibody dynamics model of the construction equipment has been established through nonlinear dynamics modeling, thereby extracting control variables related to the equipment's motion state. Based on this, to further improve ground construction accuracy, it is necessary to design a construction control strategy aimed at minimizing the deviation between the target height and the actual construction height, and to construct an adaptive control mechanism that can dynamically respond to error changes. This control strategy integrates the construction error model with the equipment control variables and continuously adjusts the equipment behavior through real-time updates of control signals, maintaining a high degree of consistency between the construction surface and the target surface.

[0094] In this embodiment, the construction control strategy is based on a construction error model. As one of the input variables, combined with the control variables of the construction equipment. Dynamically optimize control output.

[0095] Generally, the error model includes the target height. Compared to actual height The differences between them constitute the optimization objective of the control strategy.

[0096] Specifically, the optimization objective of the control strategy can be expressed as:

[0097] ;

[0098] in, Let be the cost function, representing the entire construction area. The sum of squares of the deviations between the internal target height and the actual height; The control input is the set of control variables used to regulate the motion state of construction equipment. This indicates the actual ground height measured during construction. Indicates in the target area The predetermined target altitude of the location; This is the construction area; For the region Tiny regional elements on the surface.

[0099] As an option, control variables It can be further refined into the following form:

[0100] ;

[0101] in, Indicates the speed of the equipment; For the angle of the work tool; This indicates the pressure or flow control parameters applied to the material.

[0102] Under this control strategy, the system gradually approximates the target height surface by adjusting the control variables in multiple dimensions, thereby achieving overall fitting of the construction surface.

[0103] To ensure that the control strategy can respond to error changes in real time, this invention further designs a feedback-based adaptive control mechanism. This mechanism dynamically adjusts the control commands through error feedback signals, thereby correcting equipment behavior and suppressing the spread and accumulation of errors.

[0104] In one possible implementation, the adaptive control mechanism constructs an adjustment law based on Lyapunov stability theory and introduces a gain function to adjust the control parameters in real time. The system control law can be expressed as:

[0105] ;

[0106] in, This is the final output control signal; As the baseline control trajectory; This represents the construction error at the current location and time. This is the derivative of the error with respect to time, i.e., the rate of change of the error; The proportional gain and differential gain parameters are obtained through empirical adjustment; The differential gain coefficient is obtained through empirical adjustment.

[0107] As a supplement, the adaptive mechanism can also construct an exponentially weighted cumulative error by introducing a forgetting factor:

[0108] ;

[0109] in, (0,1) represents the forgetting factor; This indicates the degree of influence of the error history on the current control strategy; This represents the construction error at the current location and time. To control torque, it represents the force applied to each component; This refers to the current moment.

[0110] This structure enables the control mechanism to take into account the impact of historical errors on the system state while avoiding excessive system response caused by long-term accumulation.

[0111] In some embodiments, to further enhance the robustness of the control mechanism, the adaptive control strategy incorporates a nonlinear prediction module to make short-term predictions of error trends and correct the control signal in advance. This prediction module can be implemented through recursive modeling of the error time series, for example, using parameter recursion formulas based on least squares estimation.

[0112] ;

[0113] in, This represents the short-term predicted value of the error; , , These are the fitting coefficients, which can be dynamically updated based on historical data. For the current moment Error value; For the previous moment The error value represents the error of the device at that moment.

[0114] This prediction can be used to construct feedforward control components, which can be superimposed on feedback control structures to enhance the foresight of the response.

[0115] In this embodiment, the control strategy module, error estimation module, and multibody dynamics module are coupled at a data level. Real-time error information generated by the error model is transmitted to the core control strategy module via the data bus, triggering updates to the control quantity calculation and execution module. The control signal is ultimately received by the device execution unit and fed back to the control module for closed-loop adjustment.

[0116] In terms of system structure, the control strategy component, the sensing and acquisition system, the dynamic solver, and the motion controller form a high-frequency collaborative execution loop, which completes error acquisition, parameter calculation, and control output in each cycle.

[0117] Specifically, during equipment operation, if the error in a certain area deviates from the threshold range, the control system will adjust the equipment speed, tool angle and pressure in the corresponding area, redistribute the amount of material used and the work rhythm until the error returns to the stable range.

[0118] In some implementation scenarios, this mechanism is particularly suitable for complex ground structures, heterogeneous foundations, or highly variable construction environments with localized rebound, demonstrating good dynamic adaptability.

[0119] S4. While implementing the control strategy, topology optimization of the distribution of construction materials is carried out based on the upper limit of the total amount of materials used and the ground performance requirements.

[0120] In the foregoing embodiments, precise control of the construction equipment has been achieved through the implementation of multibody dynamics modeling and control strategies. However, there is still room for optimization regarding material usage during construction, particularly in balancing the upper limit of material usage with ground performance requirements. Therefore, this invention further proposes a material distribution topology optimization scheme based on the upper limit of total material usage and ground performance requirements. This scheme aims to meet construction accuracy and performance requirements while rationally controlling material usage to reduce material waste and improve construction economy.

[0121] In this embodiment, the material distribution topology optimization problem is described as optimizing the distribution of construction materials based on an upper limit on the total amount of materials used, while meeting construction accuracy requirements. By combining ground performance requirements with material usage constraints, an optimization method for intensive allocation of construction materials is formed.

[0122] Specifically, the goal is to minimize material usage while meeting ground performance requirements (such as strength and uniformity) through the rational allocation of construction materials. To this end, the material demand function for the construction area is first defined. This represents the quantity of material required per unit area. Assume... Indicates the location The objective function for optimizing the allocated material quantity can be expressed as:

[0123] ;

[0124] in, The goal is to optimize the total amount of materials used; For position Material demand function at the location; The amount of material allocated to this location; This is the construction area; For the region Tiny regional elements on the surface.

[0125] In addition, the optimization scheme needs to meet the upper limit constraint of the total material usage, that is:

[0126] ;

[0127] in, This is the upper limit for the total amount of materials used; The amount of material allocated to this location; This is the construction area; For the region Tiny regional elements on the surface.

[0128] As an alternative, ground performance requirements should also be considered during the optimization process to ensure that the post-construction ground performance does not fall below the predetermined standards. This assumes that ground performance in certain areas is achieved through parameters... The following performance constraints must be met during the optimization process:

[0129] ;

[0130] in, Ground performance at the location; The target ground performance requirements.

[0131] In this embodiment, the topology optimization problem is solved by numerical optimization methods. The construction area is discretized using the finite element method (FEM) or other suitable numerical methods, and the optimal material distribution is obtained by solving the optimization problem.

[0132] In one possible implementation, the optimization problem can be solved using a gradient-based optimization algorithm (e.g., the finite difference method or gradient descent). The optimization algorithm continuously adjusts the material distribution at various locations within the region, calculates the objective function value, and updates the material distribution based on gradient information until the minimization objective and constraints are met.

[0133] When performing topology optimization for material distribution, the construction control strategy module and the optimization module need to be tightly coupled. Within each optimization cycle, the control strategy module first generates control signals based on the current construction error information and transmits them to the equipment for motion adjustment. Simultaneously, the material distribution optimization module optimizes the material allocation scheme in real time based on construction accuracy requirements and material usage limitations.

[0134] Specifically, the control signals generated in the construction control strategy will affect the material allocation process. For example, when construction equipment needs to make precise adjustments to a certain area, the control system may require more material to be allocated to that area to ensure that the ground flatness meets the requirements. At this time, the topology optimization module will dynamically adjust the amount of material used in the corresponding area based on the feedback of the control signals, ensuring that the optimization result can meet the ground performance requirements, while controlling the total amount of material used to not exceed the upper limit.

[0135] To further improve the accuracy and adaptability of the optimization results, this invention introduces an adaptive feedback mechanism. This mechanism can dynamically adjust the parameters during the optimization process based on real-time construction data, enabling material distribution optimization to respond in real time to changes in construction errors, equipment status, and ground performance.

[0136] In some embodiments, the feedback mechanism is designed to adjust the material allocation strategy through real-time error feedback. After the equipment completes construction in a certain area, the error feedback module evaluates the construction effect in that area and adjusts the material allocation for subsequent areas based on the evaluation results. For example, if the error exceeds a predetermined range, it may be necessary to increase the amount of material in that area or reallocate the material to ensure that construction accuracy and ground performance requirements are met.

[0137] S5. Input the sensor data collected during construction into the deep learning model to perform feature extraction and time series learning in order to predict the error trend in the future construction process.

[0138] In the aforementioned embodiments, construction accuracy control has been achieved through multibody dynamics modeling and error control strategies. However, in actual construction, the error trend may change over time. Therefore, this invention further proposes a prediction mechanism based on a deep learning model. This mechanism can predict future error trends by analyzing real-time sensor data during construction, performing feature extraction and temporal learning. This method can provide early warning and dynamic adjustment basis for subsequent control strategies, further improving construction accuracy and efficiency.

[0139] For the sensor data collected during construction, a deep learning model is used for feature extraction and time-series learning. By analyzing the sensor data, the model can learn the error variation patterns during construction and predict future error trends, thus providing data support for real-time construction control.

[0140] Typically, during construction, sensor data such as position, velocity, acceleration, pressure, and temperature are collected in real time. This data includes the motion state of the construction equipment and its error information relative to the target ground. By learning from this time-series data, deep learning models can automatically extract latent features from the data and predict future construction errors based on historical data.

[0141] Alternatively, Long Short-Term Memory (LSTM) networks or Gated Recurrent Unit (GRU) networks can be used as deep learning models. These models excel at handling time-series signals and can effectively capture the temporal dependence of error changes. Specifically, the model's input is the time-series information of historical construction data, and the output is a prediction of future construction error trends.

[0142] In this embodiment, the training process of the deep learning model includes four stages: data preprocessing, feature extraction, model training, and evaluation. First, the data preprocessing module cleans, normalizes, and performs feature engineering on the collected sensor data to remove noise and redundant information, providing clean input data for subsequent feature extraction.

[0143] While implementing deep learning error prediction, the prediction results are closely integrated with the previous control strategy module. Specifically, the deep learning model predicts future construction error trends. This information will be fed back to the construction control system as input. By analyzing the predicted error trends, the control system can perform error corrections and strategy adjustments in advance, thereby achieving more efficient construction control.

[0144] For example, when it is predicted that construction errors in a certain area will deviate significantly in the future, the control system can adjust the control parameters of the construction equipment in advance, such as speed, pressure, and tool angle, based on the prediction results. In this way, the system can more accurately control the construction process and avoid the accumulation and expansion of errors.

[0145] In some embodiments, to improve the accuracy of prediction results, deep learning models can also be combined with other technical modules for data augmentation and optimization. For example, error correction methods based on physical models can be introduced, and hybrid modeling can be performed by combining physical models and data-driven methods, thereby further improving the accuracy and robustness of prediction results.

[0146] The error trends predicted by the deep learning model can be used as input parameters for dynamic optimization and combined with existing construction control strategies. For example, when the model predicts that the error trend in a certain area will exceed a predetermined threshold, the system can respond by adjusting parameters such as the operating rhythm of the equipment, pressure, or material distribution.

[0147] In one possible implementation, based on the prediction results, the control system will recalculate the control signals during the construction process. Assume that at time... The system obtains the prediction error. The control system adjusts the control quantity using the following formula. :

[0148] ;

[0149] in: The adjusted control signal; The future error value predicted by the deep learning model; This is the current control signal; The adjustment function adjusts the current control signal based on the prediction error.

[0150] This adjustment mechanism enables the system to be dynamically optimized during construction, reducing error accumulation and ensuring construction accuracy.

[0151] S6. Based on the error prediction results output by the deep learning model and combined with the error information fed back in real time, dynamically adjust the equipment control strategy and material distribution optimization scheme.

[0152] By combining the error prediction results of a deep learning model with real-time construction error information, a closed-loop feedback control system is formed. This system can dynamically adjust the operating parameters of the equipment and the distribution of materials based on real-time data during the construction process, thereby achieving the best construction results.

[0153] Generally, the error prediction results output by deep learning models The results are based on time-series learning using sensor data from past moments, reflecting potential error trends in the construction process at a future point in time or over a given period. Real-time error feedback, on the other hand, comes from the construction control system's monitoring of the current construction status, typically obtained through sensors and feedback modules. By combining these two approaches, the system can perform error correction and predictive adjustments during construction.

[0154] Alternatively, real-time feedback information can be processed through error feedback control algorithms. For example, at a certain construction stage, the predicted error value might indicate that the equipment's trajectory will deviate from the predetermined target in the future, while the real-time feedback information provides the actual current error. By comparing the predicted error with the actual error, the control system can adjust the equipment's operating parameters, such as speed, acceleration, and operating angle, to ensure that the construction process does not deviate beyond the error tolerance.

[0155] In implementing this invention, the equipment control strategy and the material distribution optimization scheme are mutually influential. Specifically, adjustments to the equipment control strategy directly affect the material requirements during construction, while material distribution optimization, in turn, affects the operational accuracy and efficiency of the equipment. Therefore, this invention achieves efficient system coordination by jointly adjusting the equipment control strategy and the material distribution optimization scheme.

[0156] In some embodiments, based on the error prediction results of the deep learning model and real-time feedback information, the system first optimizes and adjusts the equipment control strategy. For example, when the model predicts an increase in future errors, the control system can adjust the equipment's movement path or operating rhythm in advance to avoid error accumulation. Simultaneously, based on the real-time error feedback, the system can also dynamically adjust the equipment's operating parameters, such as operating pressure, operating speed, and angle, thereby enabling the equipment to perform construction with higher precision.

[0157] As an alternative, the material distribution optimization scheme will also be dynamically adjusted based on error prediction results and real-time feedback. For example, when it is predicted that the error in a certain construction area will exceed the tolerance range, the system will increase the amount of material in that area to ensure the construction accuracy of that area. At the same time, the material allocation will be fine-tuned in conjunction with real-time feedback during the construction process to ensure that the material usage in each area meets the construction accuracy requirements without exceeding the upper limit of the total material quantity.

[0158] In this embodiment, the system combines deep learning predictions with real-time feedback information to achieve closed-loop adjustment of construction control strategies and material distribution optimization. Specifically, after receiving the error prediction results from the deep learning model, the control system pre-adjusts the equipment's operating parameters based on these results and further adjusts them based on the actual feedback errors.

[0159] In some embodiments, the control system can not only correct the control signals of the equipment in real time, but also dynamically update the optimization strategy for material distribution. For example, based on the predicted error trend and the real-time error value, the system determines which areas may have lower construction accuracy than required, and adjusts the material distribution in these areas in real time to avoid the expansion of construction errors.

[0160] S7. Based on the simulation optimization method, numerical solution and dynamic optimization of control strategy and material distribution results are performed to generate the optimal construction control instruction set, which is then used to guide actual ground paving construction operations.

[0161] Building upon deep learning models for error trend prediction and real-time feedback for dynamic adjustment of equipment control and material distribution, this invention further proposes a simulation-based optimization method for numerically solving and dynamically optimizing control strategies and material distribution schemes to enhance the accuracy and scheduling efficiency of construction control. By constructing a high-fidelity simulation model of the construction environment, the system can generate the optimal set of construction control instructions for the current working conditions based on prediction and feedback information. This enables full-process simulation and refined execution of construction actions, effectively connecting multiple functional modules such as error prediction, control feedback, and material optimization, and ultimately applying it to actual ground paving construction operations.

[0162] In this embodiment, a multiphysics simulation model reflecting the characteristics of the construction area, equipment motion model, material physical properties, and ground response behavior is first constructed. This model covers multiple factors related to error evolution during construction and can be used to verify the rationality of the control and adjustment strategy and material distribution results after the fusion of deep learning prediction and feedback data.

[0163] Generally, the simulation optimization process uses deep learning predictions and real-time feedback errors as input constraints, while also introducing multiple constraint variables such as equipment operating parameters, material usage limits, construction schedule requirements, and ground compaction targets to construct an objective optimization function. Under these multi-dimensional constraints, the system optimizes its control strategy. With material distribution Perform collaborative optimization.

[0164] Alternatively, the objective function can be a weighted combination of the squared integral of the error, the material distribution uniformity index, and the equipment energy consumption function, in the following form:

[0165]

[0166] in, To optimize the target value globally; This is the construction error distribution function; This represents the gradient of material distribution in space; This is the energy consumption function of the device. , , The weighting coefficients for each item are set according to the actual project requirements; For the region Tiny area elements on; This is the construction area.

[0167] In one possible implementation, the objective function is numerically solved by combining global optimization and local search strategies such as genetic algorithms, simulated annealing algorithms, or gradient descent, to obtain the optimal set of control variables that satisfy the constraints, including equipment trajectory adjustment amount, speed scheduling parameters, and local material increase / decrease strategies.

[0168] In some embodiments, to ensure the feasibility of the optimization results, the system also aligns the simulation results with the actual equipment operating constraints. These constraints include equipment turning radius, maximum load capacity, material spraying resolution, etc. By constructing a constraint mapping function, the optimization instructions in the simulation space are mapped to executable instructions for the equipment.

[0169] In this embodiment, the final generated optimal control instruction set includes a sequence of control parameters over a continuous time series:

[0170] ;

[0171] in, This refers to the instruction set used for construction control. In time The control commands executed at all times include parameters such as control angle, speed, pressure, and material delivery rate.

[0172] In a specific application, this control command set is transmitted to the actual construction equipment via the controller interface module to guide it to operate according to the optimized path and parameters along the laying path. The system synchronously receives status feedback information from the equipment to ensure that the actual execution effect is consistent with the simulation results, thereby achieving a feedback loop.

[0173] Furthermore, in areas with complex terrain or abrupt changes, to enhance the robustness of the control strategy, this invention also introduces a robust optimization mechanism based on disturbance response analysis. This mechanism performs perturbation analysis on the control results during simulation, assessing the fluctuation range of system performance under different external conditions, thereby selecting a control instruction set that possesses both optimal performance and high stability.

[0174] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A construction process for high-precision wear-resistant flooring in ultra-large, complex factory buildings, characterized by: Includes the following steps: Based on the terrain data and design requirements of the construction area, a predetermined height distribution model of the target ground is established. Combined with the sensor data collected during the construction process, the actual construction height is estimated in real time to establish a construction error model to reflect the error between the target height and the actual construction height. Using the construction error model as input, nonlinear dynamic modeling is performed on the construction equipment to establish a multibody dynamic model of the construction equipment. Control variables related to the motion state of the equipment are extracted and used as inputs for the equipment control strategy. Based on the aforementioned error model and control variables, a construction control strategy is established with the goal of minimizing the deviation between the target height and the actual height. An adaptive control mechanism is designed so that the control signal can be dynamically adjusted according to real-time error feedback. While implementing control strategies, topology optimization of the distribution of construction materials is carried out based on the upper limit of total material usage and ground performance requirements; The sensor data collected during construction is input into a deep learning model for feature extraction and time series learning in order to predict the error trend in future construction processes. Based on the error prediction results output by the deep learning model and combined with the error information fed back in real time, the equipment control strategy and material distribution optimization scheme are dynamically adjusted. The control strategy and material distribution results are numerically solved and dynamically optimized based on simulation optimization methods to generate the optimal construction control instruction set, which is then used to guide actual ground paving operations.

2. The construction process for high-precision wear-resistant flooring in ultra-large area complex factory buildings according to claim 1, characterized in that, The height distribution model includes: The height distribution model includes a digital terrain model generated based on three-dimensional terrain scanning data, and determines the target construction height by combining the regional functions, load requirements and ground flatness requirements marked in the design drawings; The model is built according to the principle of regional grid division, forming a predetermined height matrix composed of multiple cells, with each cell corresponding to a spatial positioning coordinate and a target elevation; The height distribution model supports dynamic correction of target height values ​​based on real-time sensor data during construction, in order to cope with changes in construction terrain, settlement, and human adjustments.

3. The construction process for high-precision wear-resistant flooring in ultra-large area complex factory buildings according to claim 1, characterized in that, The construction error model includes: The construction error model includes a spatial error distribution map built based on height difference data, which reflects the magnitude and direction of the deviation between the actual construction height and the target height in different areas; The error model, combined with time series recording, models the evolution of errors as construction progresses, and identifies areas of error accumulation or repeated correction. The model provides error feedback signals to the control system, which are used to adjust the equipment movement path, construction rhythm and material distribution in real time.

4. The construction process for high-precision wear-resistant flooring in ultra-large area complex factory buildings according to claim 1, characterized in that, The multibody dynamics model includes: The multibody dynamics model includes physical modeling of key structural components of the construction equipment, such as the chassis, telescopic boom, transmission mechanism, and operating tools, taking into account their mass distribution, geometric connection relationships, and kinematic constraints. The model defines the hinges and flexible connections between each connection point and embeds a simulation mechanism for the response behavior to the contact force on the construction ground. The model uses a dynamic solver to calculate in real time the nonlinear relationship between the device's attitude change, displacement response, and control input, providing a basis for control signal optimization.

5. The construction process for high-precision wear-resistant flooring in ultra-large area complex factory buildings according to claim 1, characterized in that, The control strategy includes: A control function is constructed with the objective of minimizing the sum of squared differences between the target ground height and the actual construction height, and the equipment movement speed, path change range and material spraying frequency are used as optimization variables. The control strategy adopts a hierarchical control architecture, including a global path planning module and a local error compensation module. The former is used to generate the overall construction path, and the latter is used to correct the current error in real time. Error information is collected in real time by sensors deployed on construction equipment, and the information is fed back to the control module to dynamically adjust control commands, thereby realizing a closed-loop control mechanism.

6. The construction process for high-precision wear-resistant flooring in ultra-large area complex factory buildings according to claim 1, characterized in that, The optimal distribution scheme includes: The optimal distribution scheme includes factors such as compressive strength, abrasion resistance coefficient, hydration reaction rate, and temperature and humidity conditions of the construction environment, forming a material selection library; By using a multi-objective topology optimization algorithm, the material ratio, pouring sequence and local material addition intensity of each grid cell are determined while ensuring construction accuracy and ground performance. The optimization results are output in the form of two-dimensional or three-dimensional maps, providing the types, quantities and layout strategies of materials required for each construction sub-area, to guide the actual material placement operation.

7. The construction process for high-precision wear-resistant flooring in ultra-large area complex factory buildings according to claim 1, characterized in that, The feature extraction and temporal learning include: Feature extraction and temporal learning include using convolutional neural networks to extract spatial features from height maps, error maps, and equipment status images collected by sensors at the construction site; By combining long short-term memory neural networks, the changing trend of error values ​​within a continuous construction period is modeled, and its temporal characteristics and evolution patterns are learned. The model has the ability to identify error growth, predict error peaks, and provide early warnings of high-risk areas, which can be used to enhance the control system's ability to cope with construction in complex locations.

8. The construction process for high-precision wear-resistant flooring in ultra-large area complex factory buildings according to claim 1, characterized in that, The error trend includes: Error trends include the average slope of error growth over time, the identification of local error burst areas, and the propagation trajectory of error from high-value areas to the surrounding areas; The trend modeling results combine construction path, equipment load and environmental change factors to predict and classify the spatiotemporal range of error exceeding the limit. Error trend data serves as input variables for equipment motion path optimization, material distribution reconfiguration, and control strategy reconstruction.

9. The construction process for high-precision wear-resistant flooring in ultra-large area complex factory buildings according to claim 1, characterized in that, The construction control instruction set includes: The construction control instruction set includes equipment movement path control instructions, operation tool execution parameters, material spraying rhythm and ratio settings, and error correction strategy call identifiers. The instruction set is integrated into the central control system in a structured format, and is distributed to various intelligent devices through the construction scheduling platform, and supports real-time updates and conflict resolution mechanisms; The control instruction set also incorporates a traceability mechanism to record the status feedback and error response before and after the execution of each instruction, which is used for subsequent construction quality assessment and optimization model training.

10. The construction process for high-precision wear-resistant flooring in ultra-large area complex factory buildings according to claim 1, characterized in that, The deep learning model includes: The deep learning model includes a multi-layer convolutional structure at the front end to extract spatial feature patterns from construction status images, error distribution maps, and material laying maps; The intermediate layer uses a gated recurrent neural network to improve the model's prediction accuracy regarding the direction and rate of error evolution.

Images