Construction process for cantilever grouting beam

Abstract

The invention relates to the technical field of intelligent building data processing, and discloses a cantilever grouting beam construction process which is used for solving the problems of insufficient precision and potential safety hazards caused by data acquisition dispersion, model staticization and equipment control lag in traditional construction. The technology comprises the following steps: optimizing the node layout of an Internet of Things sensor based on a genetic algorithm, and generating high-coverage real-time monitoring data; constructing a digital twinborn model of multi-objective genetic optimization and exporting lightweight parameters; fusing Kalman filtering and LSTM prediction weight to generate an edge calculation control instruction; equipment parameters are cooperatively adjusted through real number coding, and the legality of the instruction is verified; optimizing data storage and a cross-chain tracing interface by adopting block chain integer coding; and dynamically updating the security early warning rule based on reinforcement learning. According to the method, dynamic acquisition, closed-loop optimization and credible storage of the construction data are realized, the construction precision and safety are improved, and cross-engineering experience reuse is supported.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent building data processing, and particularly to a construction technology for cantilever casting beams. Background Art

[0002] As a key technology in the field of bridge construction, the swing method of construction has been gradually popularized and applied in the construction of continuous beam bridges spanning existing lines such as railways and highways in recent years. The traditional swing method positions the bridge structure by overall rotation. Although it reduces the interference with the traffic on the existing lines compared with the full hall formwork method, there are still significant technical bottlenecks under complex working conditions: firstly, the conventional swing system adopts an equal-diameter and equal-elevation design, which is difficult to meet the requirements of adjusting the moment of inertia of bridges with different spans, resulting in insufficient control of construction accuracy; secondly, the existing weighing and counterweight technology mostly relies on empirical parameters and lacks a dynamic monitoring and feedback mechanism, which is prone to causing the risk of structural instability during the swing process; thirdly, the formwork erection plan for cast-in-place beams does not fully consider the space limitations adjacent to the existing line, and there is a potential safety hazard that the foundation settlement may induce the deformation of the existing line. Especially in dense road network areas such as the connecting line of a moving goods base, the existing swing construction technology still faces technical defects such as low accuracy of swing attitude control and imperfect protective measures for adjacent lines, and there is an urgent need to develop a new swing construction system to improve structural safety and achieve refined construction management and control.

[0003] During the construction process, there is a lack of a real-time dynamic monitoring and feedback mechanism for the collaborative data of multiple links. For example, the installation parameters of the hanging basket system (slideway positioning, tension of the hanging system, elevation of the bottom platform) rely on manual experience adjustment and no standardized data model is established, resulting in the risk of error accumulation; in the traditional construction process, key nodes (such as formwork installation positioning, coordinates of prestressed ducts, concrete curing parameters) adopt a discrete paper recording method, which cannot achieve full-cycle data traceability and deviation warning; in addition, when multiple devices cooperate in operation (positioning data of truck cranes, stress monitoring data of the hanging basket system, deformation data of the closure section), there is a lack of a unified data integration platform, resulting in the lag of construction parameter optimization behind the actual working conditions, affecting construction accuracy and structural safety. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a construction technology for cantilever casting beams, which is used to solve the problems of low construction accuracy and potential structural safety hazards in the existing cantilever casting beam construction due to the lack of real-time dynamic monitoring, insufficient standardization of data models, and lack of data integration for the collaborative operation of multiple devices.

[0005] To solve the above technical problems, the specific technical solution of the present invention is as follows: A construction technology for cantilever casting beams provided by the present invention includes: Step S1: Optimize the node layout of Internet of Things sensors based on a genetic algorithm to generate real-time monitoring data; Step S2: Construct a digital twin model based on the real-time monitoring data generated in Step S1, and output three-dimensional model parameters and model error data through the model verification module; Step S3: Optimize the edge computing strategy based on the three-dimensional model parameters and model error data output in Step S2, and generate device control instructions; Step S4: Analyze the device control instructions generated in Step S3, adjust the hydraulic compensation rate and positioning frequency, and record the device energy consumption data during the adjustment process; Step S5: Optimize the construction data storage structure of Steps S1 to S4 using the blockchain algorithm, and generate a cross-chain traceability interface; Step S6: Update the safety warning rules according to the real-time deformation data, identify risk patterns through the rule engine, and trigger the coordinated update of system parameters; Among them, the following synergistic effects form a closed loop between the steps: The model error data output in Step S2 serves as the input weight for optimization in Step S3; The device energy consumption data recorded in Step S4 is fed back to the fitness function calculation in Step S1; The risk patterns identified in Step S6 trigger the directional data traceability in Step S5; The optimized data of Steps S1 to S6 are stored through the blockchain and used as the initialization parameters for new projects.

[0006] Furthermore, in the cantilever-casting beam construction process of the present invention, in Step S1: The node layout optimization includes the following operations performed in sequence: performing binary coding modeling on the sensor positions; calculating the dynamic weight coefficient based on the coverage rate and energy consumption ratio; generating a layout plan through iterative calculation of the dynamic weight coefficient.

[0007] Furthermore, in the cantilever-casting beam construction process of the present invention, in Step S2: The construction of the digital twin model includes: Performing multi-objective genetic optimization on the tolerance of the prestressed duct deviation and the concrete curing parameters; Using the NSGA-II algorithm to perform Pareto optimization on the optimized model parameters, and exporting the three-dimensional model parameters in a lightweight format.

[0008] Furthermore, in the cantilever-casting beam construction process of the present invention, the optimization of the edge computing strategy includes: Constructing a hybrid-coded chromosome integrating the Kalman filter coefficient and the LSTM prediction weight; Generating device control instructions based on the hybrid-coded chromosome and synchronizing them to the device side through a preset protocol.

[0009] Furthermore, in the cantilever-casting beam construction process of the present invention, in Step S4: The adjustment operations include: Genetic collaborative calculation with real - number encoding for hydraulic compensation rate and positioning frequency; After completing the genetic collaborative calculation, verify the legality of the instruction set through the blockchain smart contract.

[0010] Furthermore, in the cantilever - casting beam construction process described in the present invention, in step S5: The blockchain algorithm optimization includes: Perform integer encoding on the block capacity and the levels of the Merkle tree; Generate a cross - chain traceability interface based on the integer encoding result to shorten the data query response time.

[0011] Furthermore, in the cantilever - casting beam construction process described in the present invention, in step S6: The safety warning rule update includes: Construct a real - number - encoded chromosome of the deformation threshold and the stability coefficient; Update the warning rules according to the real - number - encoded chromosome, triggering the reconstruction of the sensor network and the iteration of system parameters.

[0012] The beneficial effects of the present invention: Through the optimization of sensor node layout driven by the genetic algorithm (step S1) and multi - objective digital twin modeling (step S2), high - coverage acquisition of monitoring data and dynamic adaptation of model parameters are achieved. Combining the hybrid encoding optimization with edge - computing strategies (step S3) and real - time collaborative adjustment of equipment parameters (step S4), it breaks through the traditional experience dependence and static model defects, significantly improves the construction accuracy (such as displacement compensation accuracy and real - time deformation prediction), and reduces the manual intervention error and the risk of structural instability.

[0013] Relying on the synergistic effect of steps S2 to S6, a dynamic closed - loop of "data acquisition - model optimization - equipment control - safety warning" is formed. The model error data inversely optimizes the edge - computing strategy, the equipment energy - consumption data is fed back to the sensor layout in real time, and the risk mode triggers blockchain traceability and system parameter iteration. This adaptive feedback mechanism solves the problems of data islands and optimization lag in traditional construction, and realizes the full - life - cycle dynamic management and active risk prevention of the construction process.

[0014] Through the blockchain integer - encoding storage optimization (step S5) and cross - chain traceability interface, the immutability and high - efficiency query ability of construction data are ensured. The warning rule evolution mechanism in step S6, combined with the elite chromosome data stored in the blockchain, supports the automatic inheritance of historical construction experience and the rapid initialization of new - project parameters. This solution breaks through the limitations of traditional paper records and isolated data systems, provides reusable standardized data assets for complex projects, shortens the project deployment cycle and reduces the repeated development cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] To more clearly illustrate the technical solution of the present invention, the following will briefly introduce the drawings required for use in the embodiments. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on the drawings.

[0016] Figure 1 It is a construction process flow chart of a cantilever-cast beam provided by an embodiment of the present invention. Specific embodiments

[0017] To make the purpose, technical solution and advantages of the present invention clearer, the following will clearly and completely describe the technical solution of the present invention in combination with specific embodiments and corresponding drawings of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. The following will detail the technical solutions provided by each embodiment of the present invention in combination with the drawings.

[0018] To better understand the purpose of the present invention, the following will give a more detailed description of the present invention.

[0019] As Figure 1 shown, a cantilever-cast beam construction process provided by the present invention includes: Step S1: Optimize the node layout of Internet of Things sensors based on genetic algorithms to generate real-time monitoring data; Model the candidate installation points of sensors using binary coding. Each gene bit represents the enabled state of a point. Iteratively screen the optimal layout plan through genetic algorithms. The dynamic weight coefficient adjusts the priorities of monitoring coverage rate and energy consumption balance according to the construction stage. After deployment, dynamically adjust the monitoring density through self-organizing network. The sensor layout optimization discretizes the physical space into a binary coding model, and the dynamic weight coefficient drives the genetic algorithm to find a balance between coverage rate and energy consumption. The generated real-time monitoring data includes key parameters such as stress and displacement, providing input for subsequent modeling and solving the problems of high redundancy and large data blind spots in traditional deployments.

[0020] Step S2: Construct a digital twin model based on the real-time monitoring data generated in step S1, and output three-dimensional model parameters and model error data through the model verification module; Encode the tolerance of prestressed duct installation deviation and the temperature and humidity parameters of concrete curing as chromosome gene segments to construct a multi-objective optimization model. The model verification module compares the predicted values of the twin model with the actual sensor data to generate a model error data set. The optimized three-dimensional model retains the geometric topological relationship of the main load-bearing structures through lightweight processing.

[0021] Real-time monitoring data drives the construction of digital twin models, and the multi-objective genetic algorithm balances model accuracy and computational efficiency. Model error data reflects the prediction deviation of the twin model and serves as the basis for subsequent edge computing strategy optimization. Lightweight models reduce computing loads and meet real-time requirements.

[0022] Step S3: Optimizing the edge computing strategy based on the three-dimensional model parameters and model error data output in step S2, and generating device control instructions; A hybrid coding chromosome is constructed, in which the Kalman filter coefficient segment is used for noise filter parameter optimization, and the LSTM prediction weight segment is used for time series deformation prediction. The optimized edge computing strategy generates equipment control instructions, including parameters such as weight adjustment amount and displacement compensation value, which are synchronized to the equipment control terminal through the industrial protocol.

[0023] The three-dimensional model parameters and model error data are input into the optimization engine together, and the genetic algorithm dynamically adjusts the filtering and prediction parameters to improve the real-time performance of deformation prediction. The generated equipment control instructions drive the hydraulic system and positioning equipment to form the first "perception-computation-control" closed loop. Step S4: parsing the equipment control instructions generated in step S3, adjusting the hydraulic compensation rate and positioning frequency, and recording the equipment energy consumption data during the adjustment process; The hydraulic compensation rate and positioning frequency are encoded with real numbers to construct a composite fitness function of displacement deviation and energy consumption balance. The equipment energy consumption data generated by the adjustment operation is recorded in the local database, and the legitimacy of the instruction is verified through the blockchain smart contract to ensure that the operation complies with the preset safety specifications.

[0024] The equipment control instructions are parsed into specific parameters of the hydraulic system and positioning equipment, and the real-time adjustment data is fed back to the energy consumption calculation module. Blockchain smart contract verification prevents illegal instruction execution, and the recorded energy consumption data is used to calculate the fitness function for optimizing the sensor layout.

[0025] Step S5: Use blockchain algorithm to optimize the construction data storage structure of step S1 to step S4, and generate a cross-chain traceability interface; use integer coding to optimize block capacity and Merkle tree hierarchy, small-capacity blocks store high-frequency sensor data, and large-capacity blocks store three-dimensional models and video data. The cross-chain traceability interface parses query requests and realizes multi-chain data positioning and format conversion through relay nodes. Construction data is classified and stored in different blockchains, and the cross-chain interface supports rapid tracing of equipment operation logs and model parameters. When a risk event is triggered, the associated data is retrieved in a targeted manner to support historical case matching and cause analysis of the safety early warning system.

[0026] Step S6: updating safety warning rules according to real-time deformation data, identifying risk patterns through the rule engine, and triggering coordinated update of system parameters; Encode the deformation threshold and stability coefficient into a real - number chromosome, and optimize the warning rules through the reward and punishment mechanism of reinforcement learning. After the rule engine identifies the risk pattern, it triggers the instruction for sensor network reconstruction and the iterative update of device parameters, and at the same time drives the blockchain to trace relevant data directionally.

[0027] The real - time deformation data is input into the rule engine, matches the current warning threshold, and dynamically updates the rules to improve the false - alarm filtering ability. The risk pattern triggers the collaborative update of the whole - system parameters, and stores the elite chromosomes through the blockchain to achieve cross - project knowledge transfer and rapid deployment.

[0028] Among them, the following synergistic effects form a closed - loop among the steps: The model error data output by step S2 is used as the input weight for optimization in step S3. The device energy consumption data recorded in step S4 is fed back to the fitness function calculation in step S1. The risk pattern identified in step S6 triggers the directional data traceability in step S5. The optimized data from step S1 to step S6 is stored through the blockchain and used as the initialization parameters for new projects.

[0029] Data - processing closed - loop: Forward data flow: Sensor layout (S1) → Modeling (S2) → Edge computing (S3) → Device control (S4) → Data storage (S5) → Security warning (S6).

[0030] Feedback optimization flow: Model error (S2) → Optimize edge - computing weight (S3); Device energy consumption (S4) → Dynamically adjust sensor layout (S1); Risk pattern (S6) → Trigger blockchain traceability (S5) → Update system parameters (S1 - S6).

[0031] Knowledge inheritance: The optimized data of each step is stored through the blockchain and used as the initialization parameters for new projects to achieve cross - project reuse of construction experience.

[0032] Specifically, in the cantilever - casting beam construction process described in the present invention, in step S1: The node layout optimization includes the following operations executed in sequence: Binary - coding modeling of sensor positions; Calculating the dynamic weight coefficient based on the coverage rate and energy - consumption ratio; Generating a layout scheme through iterative calculation of the dynamic weight coefficient.

[0033] Step S1.1: Binary - coding modeling of sensor positions The slides, hanging points, and bottom platforms in the construction area are discretized into grid-like candidate installation points, and each point is assigned a unique binary coding bit (0 indicates that no sensor is deployed, and 1 indicates that a sensor is deployed). The coding length is the same as the total number of candidate points, and the coding sequence represents a sensor layout scheme. For example, in an area with 100 candidate points, a 100-bit binary coding string corresponds to a specific node deployment combination.

[0034] The binary coding abstracts the sensor layout in the physical space into a computable mathematical model, and each coding bit corresponds to the enabled state of a candidate point. Through coding mapping, the complex space optimization problem is transformed into a digital sequence that can be processed by the genetic algorithm, providing a data input basis for subsequent weight calculation and iterative optimization.

[0035] Step S1.2: Calculate the dynamic weight coefficient based on the coverage rate and energy consumption ratio The coverage rate is calculated by the proportion of the overlapping area of the sensor monitoring ranges, and the energy consumption ratio is determined based on the ratio of the node communication distance to the battery loss. The dynamic weight coefficient adjusts the priority according to the construction stage: Initial construction (beam pouring stage): The weight proportion of the coverage rate is increased to ensure no monitoring blind spots in key areas; Late construction (prestressed tensioning stage): The weight proportion of the energy consumption ratio is increased to optimize the network life cycle.

[0036] The dynamic weight coefficient, as the core parameter of the fitness function of the genetic algorithm, guides the iterative direction of the algorithm. In the initial stage, it focuses on the coverage rate to avoid monitoring blind spots, and in the later stage, it focuses on energy consumption balance to extend the network life. The dynamic adjustment of the weight coefficient enables the layout scheme to adapt to the changing needs of different construction stages and improves the practicality of the optimization results.

[0037] Step S1.3: Iteratively generate the layout scheme through the dynamic weight coefficient Use the selection, crossover, and mutation operations of the genetic algorithm to iteratively optimize the binary coding population: Selection operation: Retain the individuals with high fitness (the weighted value of the coverage rate and energy consumption ratio); Crossover operation: Exchange segments of different coding strings to generate a new layout scheme; Mutation operation: Randomly flip some coding bits to introduce diversity. Finally, output the sensor layout scheme corresponding to the coding string with the highest fitness.

[0038] The genetic algorithm searches for the optimal layout scheme in the binary coding space by simulating the biological evolution process. The dynamic weight coefficient serves as the calculation basis for the fitness function, driving the algorithm to converge to the balance point of high coverage rate and low energy consumption. After the iteration is completed, the optimal coding string is decoded into the actual sensor deployment positions, generating real-time monitoring data and transmitting it to Step S2.

[0039] Description of data processing relationship Binary coding modeling (S1.1) → Dynamic weight calculation (S1.2): The binary coding of candidate points provides a structural basis for weight calculation, and the calculation of coverage rate and energy consumption ratio depends on the deployment scheme corresponding to the coding.

[0040] Dynamic weight calculation (S1.2) → Iterative layout generation (S1.3): The weight coefficient guides the fitness evaluation of the genetic algorithm and determines the screening and evolution direction of the coding string.

[0041] Iterative layout generation (S1.3) → Real-time monitoring data (output to S2): The final layout scheme is decoded into the actual sensor deployment, and the collected monitoring data is used as the input of step S2.

[0042] Specifically, in the cantilever casting beam construction process described in the present invention, in step S2: The digital twin model construction includes: Processing the tolerance of prestressed duct deviation and concrete curing parameters through multi-objective genetic optimization; Using the NSGA-II algorithm to perform Pareto optimization on the optimized model parameters and export the three-dimensional model parameters in a lightweight format.

[0043] Step S2.1: Processing the tolerance of prestressed duct deviation and concrete curing parameters through multi-objective genetic optimization Encode the tolerance range of prestressed duct installation deviation (such as ±5mm) and the temperature and humidity parameters during concrete curing into chromosome gene segments, and each gene segment corresponds to an adjustable range of a construction parameter. Construct a multi-objective optimization problem, and the optimization objectives include model geometric accuracy (minimizing deviation) and computational efficiency (minimizing model complexity). Generate parameter combination schemes through the crossover and mutation operations of the genetic algorithm, and screen candidate solutions that simultaneously meet the accuracy and efficiency requirements.

[0044] This step abstracts the construction parameters into a gene structure that can be genetically optimized, and balances model accuracy and computational resource consumption through multi-objective optimization. The actual deviation values and environmental parameters in the real-time monitoring data are input into the optimization model to drive the genetic algorithm to iteratively generate parameter combinations that take into account both accuracy and efficiency, providing a data basis for subsequent model lightweighting.

[0045] Step S2.2: Pareto optimization of the NSGA-II algorithm and export of the lightweight model; The NSGA-II algorithm is used to perform non-dominated sorting and crowding degree calculation on the candidate parameter combinations generated by multi-objective optimization, and screen the Pareto front solution set (i.e., the solution that cannot further optimize one objective without damaging other objectives). The selected parameter combinations are subjected to model lightweight processing, redundant meshes and non-critical components are removed, the geometric topological relationship of the main load-bearing structure is retained, and the three-dimensional model parameters in FBX or glTF format are exported.

[0046] The NSGA-II algorithm extracts the optimal Pareto solution set from the candidate solutions to ensure the best balance between the accuracy and efficiency of the model parameters. The lightweight processing reduces the computational load by simplifying the model structure while retaining the key construction features. The exported three-dimensional model parameters can be directly used for optimizing the edge computing strategy (step S3), and the model error data is fed back to the optimization process through the verification module to form an accuracy iteration closed-loop.

[0047] Description of data processing relationship; Parameter encoding and optimization (S2.1) → Pareto solution screening (S2.2): The output parameter combinations of multi-objective genetic optimization are used as the input of the NSGA-II algorithm, and the optimal solutions are further screened through non-dominated sorting.

[0048] Model lightweighting (S2.2) → Output of three-dimensional model parameters: The lightweighted model parameters are transferred to step S3 to support the generation of the edge computing strategy; at the same time, the model error data is fed back to the optimization process in step S2 to form a dynamic correction mechanism.

[0049] Real-time data linkage: The monitoring data updated during the construction process (such as the actual pipeline deviation value) is input into the multi-objective optimization model in real time, and the parameter range of the chromosome gene segment is dynamically adjusted to ensure that the model is synchronized with the construction progress.

[0050] Specifically, for the cantilever beam construction process described in the present invention, the edge computing strategy optimization includes: Construct a hybrid-encoded chromosome integrating Kalman filter coefficients and LSTM prediction weights; Generate device control instructions based on the hybrid-encoded chromosome and synchronize them to the device side through a preset protocol.

[0051] Step S3.1: Construct a hybrid-encoded chromosome integrating Kalman filter coefficients and LSTM prediction weights The noise covariance matrix parameters of the Kalman filter (used for sensor data denoising) and the time series prediction weights of the LSTM neural network (used for deformation trend prediction) are respectively encoded as independent gene segments of the chromosome. The Kalman filter gene segment uses real number encoding, corresponding to the process noise and observation noise parameters of the filter; the LSTM gene segment uses floating point encoding, corresponding to the connection weight matrix of the neural network hidden layer. The collaborative optimization of the two gene segments is achieved through the crossover operation of the genetic algorithm.

[0052] The optimization of the Kalman filter coefficients improves the real-time data denoising ability and ensures the reliability of the input data; the optimization of the LSTM prediction weights enhances the long-term prediction accuracy of the beam deformation trend. The hybrid-encoded chromosome integrates these two types of heterogeneous parameters into a unified optimization object, and through the iteration of the genetic algorithm, the optimal parameter combination for filtering and prediction is selected, providing high-precision input for generating device control instructions.

[0053] Step S3.2: Generate device control instructions based on the hybrid-encoded chromosome and synchronize them to the device side Decode the optimized hybrid-encoded chromosome to obtain the specific values of the Kalman filter coefficients and the LSTM prediction weights. Based on the filtered sensor data and the predicted deformation values, calculate the counterweight adjustment amount of the hydraulic system and the displacement compensation value of the positioning device, and generate a structured control instruction (such as a JSON format instruction package). Synchronize the instruction package to the hydraulic control unit and the UWB positioning terminal through an industrial communication protocol (such as OPC UA) to ensure the real-time and reliability of the instruction transmission.

[0054] The decoded parameters drive the edge computing module to generate control instructions. The Kalman filter ensures the quality of the input data, and the LSTM prediction provides a basis for forward-looking adjustments. The structured instruction package realizes cross-device data intercommunication through a standardized protocol, ensuring the synchronous execution of hydraulic compensation and positioning adjustment. After receiving the instructions, the device responds in real time, forming a "perception - prediction - control" closed loop.

[0055] Explanation of data processing relationships; Parameter fusion optimization (S3.1) → Instruction generation (S3.2): The hybrid-encoded chromosome integrates the filtering and prediction parameters, and the optimization results directly determine the calculation logic of the control instructions.

[0056] Data flow closed loop: The filtered sensor data is input into the LSTM model for deformation prediction; The prediction results and the real-time data jointly generate control instructions; The status data (such as the actual displacement amount) of the device after executing the instructions is fed back to the parameter optimization process in step S3.1.

[0057] Protocol synchronization function: The preset protocol (such as OPC UA) ensures the reliable transmission of control instructions between heterogeneous devices, solving the problem of incompatible communication protocols among multiple devices in traditional construction.

[0058] Specifically, in step S4 of the cantilever beam construction process described in the present invention: The adjustment operation includes: Performing genetic collaborative calculation with real number encoding for the hydraulic compensation rate and positioning frequency; After completing the genetic collaborative calculation, verify the legality of the instruction set through the blockchain smart contract.

[0059] Step S4.1: Performing genetic collaborative calculation with real number encoding for the hydraulic compensation rate and positioning frequency; Map the hydraulic compensation rate (value range 0 - 10 mm / s) and the positioning frequency (value range 1 - 100 Hz) to two independent gene segments of the real number encoded chromosome respectively, and the value of each gene segment corresponds to specific device parameters. Construct a composite fitness function to simultaneously evaluate the displacement compensation accuracy (goal: minimizing the deviation between the actual displacement and the theoretical value) and the energy consumption efficiency (goal: minimizing the energy consumption per unit time). Generate multiple groups of parameter combinations through the crossover and mutation operations of the genetic algorithm, and screen out the solution set with the best comprehensive accuracy and energy consumption.

[0060] Real number encoding converts continuous device parameters into a numerical form that can be processed by the genetic algorithm, and the composite fitness function ensures that the optimization process takes into account both control accuracy and energy consumption. Genetic collaborative calculation generates the optimal parameter combination through multiple generations of iteration, solving the problems of response lag and energy consumption waste caused by traditional isolated adjustment, and providing an accurate instruction basis for device operation.

[0061] Step S4.2: Verifying the legality of the instruction set through the blockchain smart contract; Package the device parameter combination output by the genetic algorithm into a structured instruction set and input it into the blockchain smart contract for legality verification: Parameter range verification: Check whether the hydraulic compensation rate exceeds the device safety threshold (such as > 10 mm / s), and whether the positioning frequency is within the allowable range (such as < 100 Hz); Operation logic verification: Verify whether the instruction sequence conforms to the preset construction specifications (such as the displacement compensation sequence, the positioning frequency gradient rule).

[0062] The verified instruction set triggers the device to execute, and the illegal instruction triggers an alarm and is recorded in the immutable blockchain log.

[0063] The blockchain smart contract automatically verifies the legality of instructions through preset rules, preventing equipment damage or safety accidents caused by incorrect parameters. The verified instructions are synchronized to the device side for execution, while illegal operations are recorded as blockchain evidence, providing data support for subsequent risk tracing.

[0064] Explanation of data processing relationship; Parameter optimization (S4.1) → Instruction verification (S4.2): The parameter combinations generated by the genetic algorithm are used as an instruction set to input into the blockchain smart contract to complete the "optimization - verification" process.

[0065] Feedback loop: The actual displacement data after the device executes is fed back to the fitness function calculation in S4.1, driving the next round of parameter optimization; The records of illegal instructions are transmitted to S5 (data storage optimization) through the blockchain, triggering cross - chain tracing and rule updates.

[0066] Security control linkage: The verification result of the smart contract is synchronized to S6 (security warning system) in real - time, and abnormal instructions trigger the update of warning rules.

[0067] Specifically, in the cantilever - casting beam construction process described in the present invention, in step S5: The blockchain algorithm optimization includes: Performing integer encoding on the block capacity and the levels of the Merkle tree; Generating a cross - chain tracing interface based on the integer encoding result to shorten the data query response time.

[0068] Step S5.1: Performing integer encoding on the block capacity and the levels of the Merkle tree; Classify and store the construction data into different blockchains according to types (such as sensor data, model parameters, equipment operation logs), and assign unique integer encoding values to the block capacity (1 - 10MB) and the levels of the Merkle tree (3 - 7 layers) for each type of data. The specific rules are: Block capacity encoding: Small - capacity blocks (1 - 5MB) store high - frequency sensor data, and large - capacity blocks (6 - 10MB) store 3D models and video surveillance data; Merkle tree level encoding: High - frequency query data is assigned low levels (3 - 4 layers), and low - frequency data is assigned high levels (5 - 7 layers).

[0069] The integer encoding binds the block storage structure with data characteristics (access frequency, data volume), and assigns the optimal storage configuration for different data types. Small - capacity blocks improve the writing efficiency of high - frequency data, and low - level Merkle trees accelerate the data verification process, jointly optimizing the blockchain storage performance. The encoded data is associated through a hash chain to ensure data immutability and fast positioning.

[0070] Step S5.2: Generate a cross-chain traceability interface based on integer coding Construct a cross-chain query index table according to the integer coding value. The index table records the coding mapping relationship between the block capacity and the Merkle tree level. When receiving a data traceability request, the cross-chain interface parses the data characteristics (such as data type, time range) in the request, matches the coding value in the index table to locate the target blockchain and the corresponding node, and realizes the joint query of multi-chain data through the relay gateway.

[0071] The cross-chain traceability interface uses the index information of integer coding to quickly locate the target data storage location, avoiding the inefficient operation of full-chain traversal. For example, a sensor data request directly locates to a small-capacity block and a low-level Merkle tree, shortening the hash verification path. The interface converts different blockchain protocols (such as Ethereum and Hyperledger Fabric) through relay nodes to achieve data interconnection between heterogeneous chains.

[0072] Explanation of data processing relationship; Integer coding (S5.1) → Generation of cross-chain interface (S5.2): The coding results of the block capacity and the Merkle tree level provide a structured index basis for the cross-chain traceability interface, and the coding value is directly mapped to the target data storage location.

[0073] Data flow closed-loop: Construction data (generated by S1 - S4) is assigned a coding value according to the type and stored in the blockchain; When a safety warning (S6) triggers a traceability request, the cross-chain interface quickly retrieves the target data based on the coding value; The traceability result is fed back to the risk pattern recognition module of S6 to drive the update of the warning rules.

[0074] Performance optimization logic: High-frequency data (such as real-time sensor data) is stored in small-capacity blocks to reduce the block packaging time; the low-level Merkle tree reduces the verification steps and improves the query efficiency.

[0075] Specifically, in step S6 of the cantilever casting beam construction process described in the present invention: The update of the safety warning rules includes: Construct a real-number coding chromosome of the deformation threshold and the stability coefficient; Update the warning rules according to the real-number coding chromosome, triggering the reconstruction of the sensor network and the iteration of system parameters.

[0076] Step S6.1: Construct a real-number coding chromosome of the deformation threshold and the stability coefficient; Map the deformation threshold of the cantilever casting beam (the allowable range of the beam deflection corresponding to different construction stages) and the stability coefficient (reflecting the safety redundancy of the support structure) to independent gene segments of the real-number coding chromosome respectively. The numerical range of each gene segment is set according to engineering specifications: Deformation threshold gene segment: Encodes the critical deformation values at different monitoring points. For example, the deflection threshold at the beam end is set to 0.1% - 1.5%. Stability coefficient gene segment: Encodes the safety margin parameters of the support structure. For example, the value range is 0.8 - 1.2. Through the crossover and mutation operations of the genetic algorithm, various combinations of thresholds and coefficients are generated, and the chromosome with the lowest false alarm rate is selected.

[0077] The real - number - encoded chromosome transforms the dynamic adjustment of the early - warning rule into a computable optimization problem. The deformation threshold determines the sensitivity of early - warning triggering, and the stability coefficient controls the safety redundancy. The collaborative optimization of the two is achieved through the genetic algorithm. The data of the chromosome gene segment comes from historical construction cases and real - time monitoring data to ensure that the rule update matches the actual engineering requirements.

[0078] Step S6.2: Update the early - warning rule and trigger system reconstruction; Decode the optimized real - number - encoded chromosome, extract the new parameter combination of the deformation threshold and the stability coefficient, and update the safety early - warning rule library. Evaluate the rule effectiveness through the Q - learning mechanism: Reward for correct early - warning: Retain the corresponding chromosome and increase its genetic probability in the population; Penalty for false alarm: Eliminate the relevant chromosome and trigger the sensor network reconstruction instruction. The reconstruction instruction drives the redeployment of sensor nodes or adjusts the monitoring density, and at the same time synchronizes the updated parameters to the optimization module in steps S1 - S5 to achieve the iteration of the whole - system parameters.

[0079] The early - warning rule update is dynamically adjusted based on the chromosome decoding result and real - time data feedback. The Q - learning mechanism ensures the continuous evolution of the rule library. The sensor network reconstruction instruction dynamically adjusts the layout of monitoring nodes through the self - organizing network protocol to solve the local blind - area problem; the system parameter iteration feeds the optimized data back to the blockchain storage module (step S5) to form a cross - project knowledge transfer closed - loop.

[0080] Explanation of data - processing relationships; Chromosome construction (S6.1) → Rule update (S6.2): The real - number - encoded chromosome provides a parameter basis for the early - warning rule and directly drives the update of the rule library after decoding.

[0081] Feedback optimization closed - loop: The early - warning result (correct / false alarm) is fed back to the chromosome optimization process to drive the evolution of the next - generation population; The reconstruction instruction triggers the synchronous update of the sensor network (S1) and device control (S4) parameters.

[0082] Cross - system linkage: The updated early - warning rule is stored through the blockchain (S5). When a new project is launched, the historical elite chromosomes are called to initialize the rule library, shortening the deployment cycle.

[0083] The core pain points solved by the technical solution and the data processing flow; Data acquisition standardization and dynamic optimization; In traditional construction, the layout of sensors depends on manual experience, resulting in monitoring blind spots and redundancy problems, and the data of multiple devices are scattered and isolated. In the present invention, the layout of sensor nodes is optimized by the genetic algorithm in step S1, the physical space is discretized into a binary coding model, and the dynamic weight coefficient adjusts the coverage rate and energy consumption priority according to the construction stage to generate high-precision real-time monitoring data. In step S2, a digital twin model is constructed based on this data, and the multi-objective genetic algorithm is used to balance the model accuracy and calculation efficiency, and lightweight three-dimensional parameters are derived. This process solves the defects of traditional data acquisition discretization and model staticization, and provides a standardized and dynamic data input basis for subsequent links.

[0084] Real-time processing and closed-loop feedback control; In traditional construction, the control of equipment depends on empirical parameters and lacks a real-time feedback mechanism, resulting in response lags and safety hazards. In step S3, the Kalman filter and LSTM prediction weights are fused, and the edge computing strategy is optimized through hybrid coding chromosomes to generate equipment control instructions that take into account noise suppression and deformation prediction. In step S4, the instructions are parsed and the hydraulic compensation and positioning frequency are adjusted, and the energy consumption data is recorded at the same time, and the legality of the operation is verified through the blockchain smart contract. The actual displacement data and energy consumption data after the equipment execution are fed back to the sensor layout optimization (S1) and model parameter update (S2) in real time to form a "perception - calculation - control" closed loop. This process breaks through the limitations of traditional isolated regulation and realizes the dynamic safety control of the construction process.

[0085] Trusted data storage and cross-system collaboration; In traditional construction, data is stored in paper form or dispersed, resulting in difficulties in traceability and data island problems. In step S5, the blockchain storage structure is optimized by integer coding, the block capacity and Merkle tree levels are allocated according to the data type, high-frequency data is written quickly, low-frequency data is verified efficiently, and a cross-chain traceability interface is generated to support multi-chain data joint query. In step S6, the early warning rules are updated based on the real-time deformation data, and the deformation threshold and stability coefficient are dynamically optimized through real-time coding chromosomes, triggering the reconstruction of the sensor network and the iteration of the whole system parameters. After the risk mode is identified, the blockchain traces historical data in a targeted manner, and the optimized parameters are stored through the blockchain and used as the initialization seeds for new projects. This process realizes the trusted storage of data throughout the life cycle and the transfer of cross-project knowledge, and solves the problems of traditional data islands and experience dependence. Specific embodiments Implementation scenarios; In the swivel construction of a continuous girder bridge spanning a railway, the construction technology of this invention is applied to the cantilever casting of the girder. The construction area is adjacent to the connecting line of the freight base, and it is necessary to ensure zero interference to the existing line during the swivel process and achieve millimeter-level displacement control.

[0087] Implementation steps; Dynamic deployment of sensors (Step S1); Divide the slideway, hanging points and bottom platform into 100 candidate installation points, and use binary coding for modeling (1 means deploying sensors, 0 means not deploying). At the initial stage of construction (during the girder casting stage), set the dynamic weight coefficients α = 0.7 (focusing on coverage rate) and β = 0.3 (energy consumption ratio). After deployment, dynamically increase the node density in the slideway area through Zigbee self-organizing network; in the later stage (tensioning stage), adjust to α = 0.4 and β = 0.6, and reduce the monitoring frequency in the edge area to optimize energy consumption. The genetic algorithm iteratively generates the optimal layout plan, with the coverage rate of key areas ≥ 98% and the network energy consumption reduced by 32%.

[0088] Construction of the twin model and optimization (Step S2); Real-time monitoring data (stress, temperature and humidity) is input into the digital twin system, and the deviation tolerance of the prestressed duct (±5mm) and the temperature and humidity parameters of concrete curing are encoded as chromosome gene segments. Use the NSGA-II algorithm to perform multi-objective optimization on the model parameters, and export the lightweight FBX model. The model response speed is increased by 38%, and the error data (model prediction VS actual value) is transmitted to Step S3 Generation of edge computing instructions (Step S3); The hybrid-encoded chromosome fuses the Kalman filter coefficient (noise covariance matrix) and the LSTM deformation prediction weight. After optimization, the filtering delay ≤ 50ms, and the deformation prediction accuracy reaches 91%. Generate control instructions in JSON format (such as hydraulic compensation amount +2.3mm, positioning frequency 45Hz), and synchronize them to the hydraulic control terminal and the UWB positioning system through the OPC UA protocol.

[0089] Device collaborative control and verification (Step S4); The hydraulic system performs compensation according to the instructions and records the energy consumption data in real time (such as the energy consumption of a single compensation is 0.8kW·h).

[0090] The blockchain smart contract verifies the legality of the instructions: it is detected that the positioning frequency "60Hz" exceeds the device threshold (preset ≤ 50Hz), triggers an alarm and interrupts the execution, and the abnormal log is encrypted and stored in the blockchain.

[0091] Data storage and cross-chain traceability (Step S5); Sensor data (high frequency) is stored in 1-3MB blocks with 3 layers of Merkle trees; 3D model data (low frequency) is stored in 8-10MB blocks with 6 layers.

[0092] An abnormal deformation was detected during the rotation process. The cross-chain interface located the historical data of the slideway area sensor within 5 seconds (block ID: 0x3a7d), and the traceability showed that the model prediction deviation was caused by temperature and humidity fluctuations.

[0093] Safety warning and system iteration (step S6); Update of the deformation threshold chromosome: The beam end deflection threshold is dynamically adjusted from 1.2% to 0.9%, and the stability coefficient is reduced from 1.1 to 0.95.

[0094] After a false alarm is detected, the Q-learning mechanism eliminates the corresponding chromosome, triggers the sensor network reconstruction instruction, and redeploys 12 bottom platform nodes through self-organizing networking.

[0095] The optimized parameters are synchronized to the blockchain and used as the initialization seed for the new project "XX High-Speed Railway Overhead Bridge".

Claims

1. A suspended grouting beam construction process, characterized in that: include: Step S1: Optimize the node layout of IoT sensors based on genetic algorithm to generate real-time monitoring data; Step S2: construct a digital twin model according to the real-time monitoring data generated in step S1, and output three-dimensional model parameters and model error data through a model verification module; Step S3: Optimizing the edge computing strategy based on the three-dimensional model parameters and model error data output in step S2, and generating device control instructions; Step S4: parsing the equipment control instructions generated in step S3, adjusting the hydraulic compensation rate and positioning frequency, and recording the equipment energy consumption data during the adjustment process; Step S5: Use blockchain algorithm to optimize the construction data storage structure from step S1 to step S4, and generate a cross-chain traceability interface; Step S6: updating safety warning rules according to real-time deformation data, identifying risk patterns through the rule engine, and triggering coordinated update of system parameters; The steps form a closed loop through the following synergy: The model error data outputted from step S2 is used as the input weight for optimization in step S3; The equipment energy consumption data recorded in step S4 is fed back to the fitness function calculation in step S1; The risk pattern identified in step S6 triggers the directional data tracing in step S5; The optimized data from step S1 to step S6 are stored via blockchain and used as initialization parameters for new projects.

2. The suspended grouting beam construction process according to claim 1, characterized in that: In step S1: the node layout optimization includes the following operations performed in sequence: binary coding modeling of sensor positions; calculating dynamic weight coefficients based on coverage and energy consumption ratio; and iteratively generating a layout plan through dynamic weight coefficients.

3. The suspended grouting beam construction process according to claim 1, characterized in that: In step S2: the digital twin model construction includes: Processing prestressed pipe deviation tolerance and concrete curing parameters through multi-objective genetic optimization; The NSGA-II algorithm is used to perform Pareto optimization on the optimized model parameters and export the three-dimensional model parameters in a lightweight format.

4. The suspended grouting beam construction process according to claim 1, characterized in that: The edge computing strategy optimization includes: Construct a hybrid coding chromosome that integrates Kalman filter coefficients and LSTM prediction weights; Generate device control instructions based on hybrid coding chromosomes and synchronize them to the device end through a preset protocol.

5. The suspended grouting beam construction process according to claim 1, characterized in that: In step S4: The adjustment operation includes: Genetic collaborative calculation of hydraulic compensation rate and positioning frequency with real number encoding; After completing the genetic collaborative computing, the legitimacy of the instruction set is verified through the blockchain smart contract.

6. The suspended grouting beam construction process according to claim 1, characterized in that: In step S5: the blockchain algorithm optimization includes: Integer encoding of block capacity and Merkle tree levels; Generate a cross-chain traceability interface based on integer coding results to shorten data query response time.

7. The suspended grouting beam construction process according to claim 1, characterized in that: In step S6: the security warning rule update includes: Construct real-number coded chromosomes of deformation thresholds and stability coefficients; The early warning rules are updated according to the real-coded chromosomes to trigger the reconstruction of the sensor network and the iteration of system parameters.

Images