Accurate positioning system for digital steel structure embedded part

Abstract

The invention discloses a digital steel structure embedded part accurate positioning system, and belongs to the technical field of computer aided design and digital twinning, and the system comprises a data acquisition module which is used for obtaining a correction vector of a digital twinning model, the currently measured local positioning precision and the current global attitude consistency deviation; the model entropy calculation module is used for calculating a model authority entropy representing the accumulated deviation degree of the digital twin model based on the correction vector and a preset reference correction scale; the model authority calculation module is used for determining the model authority of the digital twin model according to the model authority entropy and a preset attenuation sensitivity coefficient; and the correction decision module is used for generating a correction acceptance probability in combination with the model authority and the local positioning precision, and performing probabilistic screening on the execution of the correction vector based on the correction acceptance probability, thereby providing a solid quantitative basis for subsequent judgment of the health condition of the model.

Description

Technical Field

[0001] The present invention relates to the field of computer-aided design and digital twin technology, and in particular to a digital steel structure embedded parts precise positioning system. Background Art

[0002] When existing technologies use building information models and surveying robots for digital positioning, their accuracy and reliability are built on a stable and reliable coordinate base. In extreme engineering scenarios, such as deep-sea immersed tube tunnels, the structure undergoes continuous and nonlinear creep due to water pressure and its own weight, causing the coordinate base itself to change dynamically. Furthermore, the base of internal sensors, such as the inertial measurement unit, may shift due to unexpected structural torsion. This causes the system to continuously modify its internal digital twin model when correcting global positioning deviations, gradually eroding the original authority of the model design. This can even lead to a vicious cycle of deviation-correction-further deviation due to misattribution, ultimately leading to a collapse of system credibility.

[0003] To solve this problem, a new technical solution has been proposed. The core of this solution is to establish a mechanism that can quantify the credibility of the model and make adaptive decisions based on it. This mechanism no longer blindly pursues the apparent self-consistency of the data, but instead conducts a comprehensive assessment of the system health in a dynamic balance and has a final circuit breaker protection to ensure that it can actively switch to a safe redundancy mode before the model's authority collapses.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0005] The purpose of the present invention is to provide a digital steel structure embedded parts precise positioning system to solve the problems raised in the above background technology.

[0006] The technical solution of the present invention is, comprising: The data acquisition module is used to obtain the correction vector of the digital twin model, the current measured local positioning accuracy, and the current global posture consistency deviation; The model entropy calculation module is used to calculate the model authority entropy that represents the cumulative deviation degree of the digital twin model based on the correction vector and the preset reference correction scale; A model authority calculation module is used to determine the model authority of the digital twin model based on the model authority entropy and a preset attenuation sensitivity coefficient; The correction decision module is used to combine the model authority and local positioning accuracy to generate the correction acceptance probability and perform probabilistic screening of the execution of the correction vector based on the correction acceptance probability; The risk assessment module is used to integrate the model authority entropy and the global posture consistency deviation, calculate the system instability index, and trigger the fuse protection mechanism when the system instability index exceeds the preset fuse threshold.

[0007] Preferably, the model entropy calculation module includes: The normalization unit is used to normalize each component of the correction vector according to its corresponding reference correction scale to generate a dimensionless entropy increment; the accumulation unit is used to determine the model authoritative entropy by recursively accumulating the dimensionless entropy increment.

[0008] Preferably, the model authority calculation module maps the model authority entropy to the model authority based on an exponential decay model, and the model authority and the model authority entropy are in an exponential decay relationship.

[0009] Preferably, the modified decision module is used to generate a modified acceptance probability, including: The local positioning accuracy is compared with the preset maximum acceptable threshold of local accuracy to generate a local accuracy performance factor; the correction acceptance probability is determined based on the product of the model authority and the local accuracy performance factor.

[0010] Preferably, the risk assessment module is used to calculate the system instability index, including: The global posture consistency deviation is compared with the global deviation reference threshold to determine a first relative risk; the model authority entropy is compared with the model entropy reference threshold to determine a second relative risk; and the system instability index is determined based on the weighted sum of the first relative risk and the second relative risk.

[0011] Preferably, when the risk assessment module triggers the fuse protection mechanism, the system switches to a safety redundancy mode and terminates all modifications to the global model; when the system instability index does not exceed the preset fuse threshold, the system maintains the current operating mode.

[0012] Preferably, the correction decision module realizes negative feedback suppression of the correction behavior through probabilistic screening; The acceptance of the correction vector leads to an increase in the model authority entropy, the increase in the model authority entropy leads to a decrease in the model authority, and the decrease in the model authority leads to a decrease in the probability of correction acceptance to inhibit subsequent corrections.

[0013] Preferably, the reference correction scale, attenuation sensitivity coefficient, local accuracy maximum acceptable threshold, global deviation reference threshold, model entropy reference threshold and fuse threshold are all parameters preset according to engineering design tolerance or early simulation calibration.

[0014] The present invention provides a digital steel structure embedded parts precise positioning system through improvement, which has the following improvements and advantages compared with the prior art: 1. This solution incorporates a model entropy calculation module to transform the abstract concept of model deviation into a measurable and accumulative physical quantity, namely, the model's authoritative entropy. This mechanism is achieved through the quantitative accumulation of each correction action. To address the issue of correction vectors containing components of different physical dimensions, such as displacement and rotation, the solution provides the system with an objective and persistent memory that records the cumulative impact of all historical corrections on the model's original state, providing a solid quantitative foundation for subsequent assessments of the model's health. 2. A potentially infinitely cumulative entropy value is nonlinearly mapped to a credibility metric between 0 and 1 that aligns with engineering intuition. The preset decay sensitivity coefficient acts as a regulator, its value determined through early simulations, defining the system's overall tolerance for model modifications. A large k value means that even small cumulative corrections will lead to a rapid decline in model authority. This exponential decay relationship accurately simulates the accelerated decay of authority when negative influences persist, making model authority a dynamic indicator that is sensitive to risk. 3. This creates effective negative feedback, inhibiting subsequent corrections and enabling the algorithm to calculate the correction vector again. The probability of the system accepting and executing it is significantly reduced, effectively breaking the potential cycle of deviation-correction-further deviation. 4. It provides the system with an insurmountable safety bottom line, ensuring that it can actively switch to a safe redundancy mode before the model's authority collapses, thus eliminating catastrophic consequences caused by the system's confident error correction. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The present invention will be further explained below in conjunction with the accompanying drawings and Examples: Figure 1 It is a flow chart of the system of the present invention. DETAILED DESCRIPTION

[0016] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0017] Example 1 See also Figure 1 The present invention provides a digital steel structure embedded parts precise positioning system, comprising: The data acquisition module is used to obtain the correction vector of the digital twin model, the current measured local positioning accuracy, and the current global posture consistency deviation; The model entropy calculation module is used to calculate the model authority entropy that represents the cumulative deviation degree of the digital twin model based on the correction vector and the preset reference correction scale; A model authority calculation module is used to determine the model authority of the digital twin model based on the model authority entropy and a preset attenuation sensitivity coefficient; The correction decision module is used to combine the model authority and local positioning accuracy to generate the correction acceptance probability and perform probabilistic screening of the execution of the correction vector based on the correction acceptance probability; The risk assessment module is used to integrate the model authority entropy and the global posture consistency deviation to calculate the system instability index and trigger the fuse protection mechanism when the system instability index exceeds the preset fuse threshold; This invention provides a digital steel structure embedded component precision positioning system. This system aims to address the technical problem in extreme engineering scenarios, such as deep-sea immersed tube tunnels, where the credibility of digital twin models gradually decreases due to dynamic changes in coordinate references, and can even lead to a vicious cycle of deviation-correction-further deviation. The inherent logic of the technical solution lies in establishing a mechanism that can quantify model credibility, make adaptive decisions, and provide ultimate fuse protection, thereby maintaining long-term stability and reliability of the system while ensuring positioning accuracy. The system of this embodiment, with its overall architecture and data flow process, forms a complete technical closed loop. The system's initial steps initiate the workflow through a data acquisition module. The data acquisition module is responsible for acquiring raw system status data from various sensors and measuring devices, with the goal of providing real-time, quantitative input for subsequent analysis and decision-making. In this embodiment, this module is specifically used to acquire three types of key data: Correction vector for the digital twin model: The correction vector is the adjustment calculated by the positioning algorithm that needs to be applied to the current digital twin model to offset measurement deviations. It may contain multiple components such as displacement and rotation, and is the direct cause of the change in the model state. Current measured local positioning accuracy: Local positioning accuracy refers to the actual performance of a single or short-term positioning task. It is usually measured as the error between the measured point and the corresponding model point. Its role is to characterize the real-time performance of the system on a specific task. Current global posture consistency deviation: Global posture consistency deviation refers to the overall posture deviation of the current physical structure relative to its digital twin model evaluated through a holistic analysis of multiple positioning points. Its function is to reflect the system-level and macroscopic deviation status; After data collection, the process enters the model entropy calculation module; the model entropy calculation module is one of the core quantization units of this system. Its purpose is to convert discrete, multi-dimensional correction vectors into a single scalar that can characterize the cumulative deviation of the digital twin model. In this embodiment, the module performs calculations based on the collected correction vectors and a preset reference correction scale. The reference correction scale refers to the parameters preset based on engineering design tolerances or previous simulation calibration, which provides a unified evaluation benchmark for correction components of different physical dimensions. The output of this module is a quantitative indicator called model authority entropy. Model authority entropy refers to a dimensionless value that draws on the concept of information entropy and is used to characterize the cumulative information distortion or uncertainty of the current digital twin model compared to its initial design benchmark. It serves as the key basis for all subsequent decisions. The model authority calculation module processes the model authority entropy; the purpose of the model authority calculation module is to convert the abstract model authority entropy into a more intuitive credibility indicator between 0 and 1; in this embodiment, the module determines the model authority of the digital twin model based on the model authority entropy and a preset attenuation sensitivity coefficient; the attenuation sensitivity coefficient refers to a dimensionless parameter calibrated through previous simulations, and its function is to define the rate at which the model authority decreases as the model authority entropy increases; model authority refers to a normalized measure of the degree of conformity or trustworthiness of the current digital twin model with its original design benchmark. The higher the value, the closer the model is to the initial design state and the higher the credibility. The system's correction decision module intervenes; the correction decision module is the system's adaptive control center, and its purpose is to avoid unconditional execution of all correction instructions, thereby interrupting a potential vicious correction cycle. In this embodiment, this module combines the model authority calculated in the previous step with the local positioning accuracy obtained by the data acquisition module to jointly generate a correction acceptance probability. The correction acceptance probability refers to the probability value of the system deciding to adopt and execute the current correction vector. The module then performs probabilistic screening of the execution of the correction vector based on this probability, that is, accepting or rejecting the correction with a certain probability. To ensure the overall security of the system, the risk assessment module works in parallel. The risk assessment module is the system's security guarantee and final line of defense. Its purpose is to integrate risk indicators from different dimensions, conduct a comprehensive assessment of the overall stability of the system, and trigger a protection mechanism when the risk is too high. In this embodiment, the module integrates the model authority entropy, which represents the chronic and hidden risk of model health decline, and the global posture consistency deviation, which represents the immediate and observable global deviation risk, to calculate a system instability index. The system instability index is a comprehensive risk indicator used to quantify the possibility of system collapse. When the index exceeds the preset fuse threshold, the system will trigger the fuse protection mechanism. The fuse threshold is a critical value preset based on simulation or engineering experience that indicates that the system is about to enter an irreversible failure state. This embodiment, through the collaborative work of the above modules, constructs a complete closed-loop control system from state quantification to decision adaptation and risk fusing. Instead of blindly pursuing apparent self-consistency in data, it comprehensively evaluates system health in a dynamic balance. The system can effectively quantify and suppress the erosion of authority caused by continuous corrections to the digital twin model, avoiding the vicious cycle of deviation-correction-greater deviation. While ensuring precise positioning capabilities, it also greatly improves the long-term operational reliability and credibility of the system in extreme engineering environments. The present invention provides a digital steel structure embedded component precision positioning system. Compared to existing technologies, this system improves upon the logic used to handle positioning deviations. When using building information models and surveying robots for positioning, the accuracy and reliability of existing technologies are based on a static, assumedly absolutely reliable coordinate base. However, in extreme engineering scenarios such as deep-sea immersed tube tunnels, the nonlinear creep of the structure due to water pressure and deadweight can destabilize the coordinate base. Existing technologies often fall into a cycle of continuously revising the digital twin model to forcefully fit the observed data. This process gradually erodes the original authority of the model design and can ultimately lead to a complete collapse of the system's credibility due to misattribution. The core advancement of this invention lies in the introduction of quantitative introspection and active defense mechanisms; it no longer assumes the absolute authority of the digital twin model, but continuously evaluates its credibility as a dynamic variable, and makes adaptive decisions based on the evaluation results, setting an insurmountable safety bottom line.

[0018] Example 2 The model entropy calculation module includes: a normalization unit for normalizing each component of the correction vector according to its corresponding reference correction scale to generate a dimensionless entropy increment; an accumulation unit for determining the model authority entropy by recursively accumulating the dimensionless entropy increment; In this embodiment, the specific implementation of the model entropy calculation module is limited. The purpose of this module is to solve the problem that the correction vector may contain components of different physical dimensions such as translation and rotation, which makes it impossible to directly accumulate. To this end, the module is divided into two core units: a normalization unit and an accumulation unit. The purpose of the normalization unit is to eliminate the physical dimensions of each component in the correction vector so that it can be processed mathematically in a unified manner; to further clarify, this unit is used to calculate the dimensionless entropy increment at the discrete time step t. , the correction vector The components of Correct the scales according to their respective references Perform normalization processing; It refers to the dimensionless increment introduced by the correction action at the current moment, which represents the increase in model uncertainty. It serves as the basic unit of the accumulation of the model's authoritative entropy. The calculation formula is: ;

[0019] in, Refers to the correction vector The i-th component of , for example, the displacement correction in the x-axis direction or the rotation correction around the z-axis, is calculated by the system's positioning algorithm; Refers to the reference correction scale for the i-th component, for example, the maximum allowable displacement correction or maximum attitude correction, which is derived from the design tolerance of the project or the parameters preset in the early simulation calibration. Its dimension is the same as same; i: correction vector The index of the i-th component of ; t: discrete time step or moment; Refers to a global entropy increase scaling factor, which is a dimensionless adjustable parameter. It is preset after calibrating the system's sensitivity to comprehensive corrections through early simulations. By dividing each correction component by its corresponding reference scale, the unit ensures that each term in the calculation is a dimensionless value, thereby generating a dimensionless entropy increment; The square operation here It has a dual effect: it ensures that all entropy increments are positive, which is consistent with the physical intuition that entropy always increases or remains unchanged; it imposes a stronger penalty weight on larger relative deviations, which means that the more corrections that exceed the tolerance, the more damage they will be considered to the original authority of the model, thereby significantly increasing the model authority entropy. It means that all correction components, such as the entropy increments contributed by x, y, z translation and rotation, are accumulated to form the total entropy increment at the current moment. ; The accumulation unit is used to convert the instantaneous entropy increment into an indicator that can reflect the historical cumulative effect. In this embodiment, the unit generates the dimensionless entropy increment generated by the normalization unit. Perform recursive accumulation to determine the final model authority entropy ; The implementation follows the following recursive formula: ;

[0020] in, : The dimensionless model authority entropy at time t is the core indicator for measuring the degree of cumulative deviation of the model; the data type is a dimensionless floating point number; the source is the model entropy of this accumulation unit at the previous moment and the entropy increment at the current moment Calculated on the basis of : The authoritative entropy of the model at the previous moment; the data type is a dimensionless floating point number; the source is the previous calculation result stored by the system; : The dimensionless entropy increment at the current moment; its data type is a dimensionless floating point number; the source is calculated by the normalization unit in this module; By introducing normalization and accumulation units, this embodiment provides a specific, feasible, and logically rigorous method for calculating model authority entropy. This not only solves the fundamental problem of mixed calculations of heterogeneous physical quantities, such as length and angle, and ensures the mathematical consistency of the model authority entropy as a dimensionless scalar, but also, through recursive accumulation, enables it to truly reflect the ongoing impact of all historical correction operations on the model authority, providing more reliable and accurate input for subsequent authority calculations and decision-making. This solution uses a model entropy calculation module to transform the abstract concept of model deviation into a measurable and accumulative physical quantity, namely the model authority entropy. Its internal mechanism is achieved through the quantified accumulation of each correction behavior. To address the problem that the correction vector may contain components of different physical dimensions, such as displacement and rotation, the calculation of the entropy increment normalizes each component. The dimensionless entropy increment at the current moment By formula ,in, : Global entropy increase scaling coefficient, :Correction vector The i-th component of : The reference correction scale for the i-th component is calculated; the practical significance of this formula is that it quantifies the destructiveness of each correction according to the engineering design tolerance; each correction component are all referenced by their corresponding scales , such as the maximum displacement or maximum attitude correction allowed, are normalized; these reference scales These are all preset parameters based on the design tolerance of the project, which ensures that each term in the summation is a dimensionless value; Model Authority Entropy is a dimensionless cumulative value, which is expressed by the recursive formula ,in, :Model authority entropy, : entropy increment, t: is determined at the moment; the derivability of this formula in the specification is reflected in its definition itself, that is, the total entropy is the sum of the historical entropy and the current entropy increment; the advancement of this mechanism is that it provides the system with an objective and continuous memory, which can record the cumulative impact of all historical corrections on the original state of the model, and provide a solid quantitative basis for subsequent judgment of the health status of the model.

[0021] Example 3 The model authority calculation module maps the model authority entropy to the model authority based on the exponential decay model. The model authority and the model authority entropy have an exponential decay relationship.

[0022] In this embodiment, the implementation of the model authority calculation module is limited; the purpose of this module is to convert the model authority entropy, which has a cumulative nature and can theoretically grow infinitely, into , mapped to a model authority with clear boundaries, between 0 and 1, which is easier to make decisions and understand ; To achieve this mapping, the model authority calculation module of this embodiment adopts a calculation method based on an exponential decay model. This method ensures that the model authority and the model authority entropy exhibit a nonlinear exponential decay relationship. The calculation logic is reflected by the following formula: ;

[0023] in, : Model authority, a normalized indicator between 0 and 1, is used to intuitively represent the current credibility of the digital twin model; its data type is a dimensionless floating point number; its source is the current model authority entropy calculated by this module. Calculated; e: base of natural logarithm; k: dimensionless attenuation sensitivity coefficient, used to adjust the authority of the model Entropy of model authority Increased sensitivity; a larger k value means that even small cumulative corrections will lead to a rapid decline in model authority; the data type is a dimensionless floating point number; this is a preset parameter based on the system's overall tolerance for model modifications, and its value is usually calibrated through early simulations; : The current dimensionless model authority entropy; the data type is a dimensionless floating point number; the source is calculated by the previous model entropy calculation module; The exponential decay model ensures that when the system is in an ideal initial state, that is, the model has not been modified, , at this time the model authority , which represents the highest authority of the model; with the accumulation of corrections, Gradually increase, It decays smoothly from 1 to 0, which accurately simulates the objective law that trust or authority gradually loses after being continuously affected by negative events. The exponential decay model provides a path for calculating the authority of the model with clear physical meaning and mathematical basis. This exponential decay relationship can accurately characterize the nonlinear decay process of the model authority under the influence of continuous correction: in the initial stage, the small entropy increase has little effect on the authority; but as the entropy continues to accumulate, the decline of the authority will accelerate; this makes the authority of the model It becomes an indicator that is sensitive to changes in the model's health status and conforms to engineering logic, thereby providing more refined and dynamic control input for downstream correction decision modules; Based on the calculated model authority entropy , a normalized model authority It is further derived; its derivation process draws on the exponential decay model in reliability engineering, specifically through the formula ,in, : Model authority, k: dimensionless attenuation sensitivity coefficient, : Current dimensionless model entropy implementation; The practical significance of this formula is to nonlinearly map an entropy value that may accumulate infinitely to a credibility indicator between 0 and 1 that conforms to engineering intuition; the preset attenuation sensitivity coefficient k works as a regulator, and its value is determined through preliminary simulation to define the system's overall tolerance for model modifications; a larger k value means that even a small cumulative correction will lead to a rapid decline in the model's authority; this exponential decay relationship accurately simulates the law of accelerated attenuation of authority when it is continuously negatively affected, making model authority a dynamic indicator that is sensitive to risks.

[0024] Example 4 The revised decision module is used to generate the revised acceptance probability, including: Compare the local positioning accuracy with the preset maximum acceptable threshold of local accuracy to generate a local accuracy performance factor; determine the probability of acceptance of the correction based on the product of the model authority and the local accuracy performance factor; In this embodiment, the correction decision module is used to generate the correction acceptance probability The specific logic of the module is limited; the purpose of this module is to intelligently suppress its further, potentially harmful, self-correction behavior when the model's authority decreases, thereby forming effective negative feedback; To achieve this goal, the calculation logic of the revised decision module is designed to comprehensively consider two dimensions of information: the long-term health of the model, which is determined by the model authority. Reflects and system's immediate task performance, by local positioning accuracy The calculation process consists of two steps: Local positioning accuracy With a preset maximum acceptable threshold for local accuracy Compare to generate a local accuracy performance factor; the local accuracy performance factor refers to a dimensionless value between 0 and 1, which is used to quantify the completion quality of the current positioning task. Its calculation formula is: ;

[0025] in, Refers to the local positioning accuracy of the current measurement, such as the root mean square error between the measurement point and the model point. Its source is collected in real time by the measurement equipment; its dimension is length; Refers to the preset maximum acceptable threshold of local accuracy, which represents the maximum positioning error that can be tolerated by engineering requirements. The source is the parameters preset according to engineering design requirements; its dimension is the same as The same ensures that the ratio of the two is dimensionless; According to the model authority The product of the local accuracy performance factor generated in the previous step is used to finally determine the modified acceptance probability. ; The complete calculation formula is as follows: ;

[0026] in, : Correction acceptance probability, which indicates the probability that the system accepts and executes the current correction vector; its data type is a dimensionless floating point number, a probability value; the source is calculated by the comprehensive model authority and local accuracy performance of this module; A: subscript, refers to acceptance; : Model authority; data type is dimensionless floating point number; source is calculated by the model authority calculation module; M: subscript, refers to the model; : The local positioning accuracy of the current measurement; the dimension is length; the source is obtained in real time by the data acquisition module; L: subscript, refers to the local; : The preset maximum acceptable threshold of local accuracy; the dimension is length; the source is the parameter preset according to engineering requirements; ref: subscript, referring to reference; Therefore, the formula constructs a decision gate with logic: the model is only considered to be in good long-term health if its own long-term health is good, i.e. Large and excellent current instantaneous measurement performance, i.e. local positioning accuracy Far less than the threshold , so that when the ratio term is close to 1, the modified acceptance probability It will approach its upper limit , if any party performs poorly, the final probability of acceptance will be significantly lowered, thus achieving intelligent and prudent decision-making; This embodiment provides an adaptive and dynamic decision-making mechanism; it ensures that the correction decision is not an isolated judgment, but the result of comprehensive trade-offs: when the model authority is high and the local positioning is accurate, near , the system tends to accept corrections to optimize the model; when the model authority has been significantly reduced, or the local positioning accuracy has approached the allowable limit, If the error rate approaches 0, the system will likely reject the correction request. This design not only makes the decision logic more complete, but more importantly, it directly links the core indicator of model authority with the actual system behavior and whether to correct the error. This forms an effective negative feedback loop, effectively interrupting potential vicious cycles and improving the system's adaptability and robustness. The correction decision module of this solution no longer executes all correction instructions unconditionally, but screens them through a probabilistic gating mechanism. Its core is to form negative feedback to suppress potential vicious cycles; the acceptance probability of the correction action is By model authority and the local task performance of the system, i.e. local positioning accuracy , jointly decide; Its decision logic is solidified in the formula ,in, : Corrected acceptance probability, :Model authority, : The local positioning accuracy of the current measurement, : The preset maximum acceptable threshold of local accuracy; the advancement of this formula is that it establishes a trade-off mechanism: when the model authority is high and the local accuracy is good, the system tends to accept corrections for improvement; on the contrary, if the model authority has been significantly reduced or the local accuracy has approached its allowable limit When , the acceptance probability will approach 0, and the system will most likely reject the modification request; It is a hard indicator preset according to engineering requirements; This mechanism creates an effective negative feedback loop: correcting behavior leads to increase, leading to Decline, eventually leading to Reduced, thus inhibiting subsequent correction behavior; Take a specific example to illustrate: Assume that the initial , the local accuracy is good, and the system accepts corrections with a high probability; after multiple consecutive corrections, From 0 to 2.0, the cumulative It drops to 0.4; at this point, even if the algorithm calculates the correction vector again, the probability of the system accepting and executing it has been significantly reduced, thus effectively interrupting the potential deviation-correction-larger deviation chain.

[0027] Example 5 The risk assessment module is used to calculate the system instability index, including: Comparing the global posture consistency deviation with the global deviation reference threshold to determine a first relative risk; comparing the model authority entropy with the model entropy reference threshold to determine a second relative risk; and determining a system instability index based on a weighted sum of the first relative risk and the second relative risk;

[0028] In this embodiment, the risk assessment module is used to calculate the system instability index The purpose of this module is to integrate the two different dimensions of system risk—immediate global deviation risk and chronic model health degradation risk—to derive a comprehensive indicator that can fully reflect the possibility of system failure. To achieve this multi-dimensional risk integration, the risk assessment module of this embodiment adopts a weighted sum model derived from the field of multi-criteria decision analysis. The calculation process of this method includes the following steps: The global pose consistency deviation Deviation from a global reference threshold Compare to determine the first relative risk; the first relative risk refers to the normalized measure of the currently observable global deviation risk, calculated as: ; in, Refers to the current global posture consistency deviation, which is usually the root mean square error of all positioning points relative to a best-fit reference plane. It is obtained by real-time measurement and calculation by the system and has the dimension of length. Refers to the global deviation reference threshold, which is determined by early simulation or historical data analysis and is a critical statistical value that indicates that the system's immediate performance is about to enter an unacceptable state. same; The model authority entropy With a model entropy reference threshold The comparison is performed to determine the second relative risk, which is a normalized measure of the chronic risk of deterioration in the model's health and is calculated as: ; in, Refers to the authoritative entropy of the current model, which is continuously calculated by the model entropy calculation module and is a dimensionless value; Refers to the model entropy reference threshold, which is determined by the previous simulation and indicates that the cumulative deviation of the model is about to reach an irreversible critical value. It is also a dimensionless value. Finally, the system instability index is finally determined based on the weighted sum of the first relative risk and the second relative risk. ; The complete calculation formula is as follows: ;

[0029] in, : System instability index, a comprehensive dimensionless risk indicator; the source is calculated by this module through weighted summation; : Risk weight factor, a dimensionless parameter between 0 and 1, reflects the decision maker's balance between immediate performance risk and long-term health risk; its source is preset based on the tolerance of different risks for specific projects; Risk Weight Factor Essentially, it plays the role of a risk preference regulator. For example, in the early stages of a project or at the critical docking stage, when the real-time positioning accuracy is extremely high, a larger A value such as 0.7 to focus on monitoring global pose consistency deviation During the long-term sedimentation observation phase, we may pay more attention to the long-term health drift of the model. In this case, we can set a smaller A value such as 0.3 to focus on monitoring the authority entropy of the model This flexibility enables the system to be adapted to different engineering stages and application scenarios. : Same definition as above; Through this weighted summation calculation method, this embodiment provides a structured and quantifiable method for generating the system instability index; it successfully integrates two completely different risk sources, immediate / macro and chronic / cumulative, into the same mathematical framework and uses weight factors to Provides flexible adjustment capabilities so that the risk assessment model can be adapted to different engineering scenarios; this makes The index becomes a quantitative indicator that can comprehensively and dynamically reflect the overall health status of the system, providing a more reliable and comprehensive decision-making basis for subsequent triggering of the circuit breaker protection mechanism; To achieve ultimate system-level safety redundancy, a system instability index This index is introduced to integrate two different dimensions of risk in the system: immediate, observable global bias risk, and chronic, hidden model health degradation risk; The index calculation model is derived from multi-criteria decision analysis in the field of risk assessment, integrating multiple risk factors of different natures through weighted summation; The calculation formula is ,in, : System instability index, :Risk weight factor, : Current global posture consistency deviation, : global deviation reference threshold, : Current model entropy, : Model entropy reference threshold; the practical significance of this formula is to combine two risks of different natures through their respective reference thresholds and Normalize it to make it dimensionless relative risk, and then use the risk weight factor These reference thresholds are determined through early simulations and are critical statistical values ​​that indicate that the system is about to enter an irreversible failure state. once Exceeding the preset fuse threshold , the system triggers the circuit breaker mechanism and actively switches from the self-consistent mode that pursues global optimization to the primitive mode that only ensures core security, stops all corrections to the global model, and waits for external manual intervention; this final circuit breaker protection mechanism is a decisive improvement of this solution compared to existing technologies. It provides the system with an insurmountable safety bottom line, ensuring that it can actively switch to the safe redundancy mode before the authority of the model collapses, and eliminating the catastrophic consequences caused by the system's confident error correction.

[0030] Example 6 When the risk assessment module triggers the fuse protection mechanism, the system switches to the safety redundancy mode and suspends all modifications to the global model; when the system instability index does not exceed the preset fuse threshold, the system maintains the current operating mode; This embodiment defines the behavior of the risk assessment module after the fuse protection mechanism is triggered. This is to clarify how the system, after detecting an irreversible risk, safely transitions from a normal mode that pursues optimal performance to a protection mode that prioritizes core functionality and data integrity. The risk assessment module calculates the system instability index After that, it will be compared with a preset fuse threshold Perform real-time comparisons; It is a dimensionless constant, for example, it can be set to 1.0. It is derived from the maximum risk limit that the system can withstand, determined according to engineering safety specifications or simulation experiments. Based on the result of this comparison, the system performs one of two mutually exclusive actions: Triggering the fuse protection mechanism: When the system instability index Exceeding the preset fuse threshold , indicating that the system's comprehensive risk has reached a critical point, with a high probability of falling into an uncontrollable state. At this point, the system will automatically trigger the fuse protection mechanism. Under this mechanism, the system will switch to safe redundancy mode. Safe redundancy mode refers to a predefined, functionally degraded operating mode. Its core task is no longer to optimize the global model, but to ensure the most basic positioning function and data security. At the same time, the system will suspend all corrections to the global model. This means that the correction decision module will be bypassed or forced to output a veto result, completely cutting off the correction path that may lead to further escalation of risks. After that, the system will remain in safe mode and issue an alarm, waiting for external manual intervention or review. This fusing mechanism is the ultimate safety barrier that distinguishes this system from traditional best-effort correction systems. It fundamentally prevents the system from confidently correcting errors based on a potentially severely inaccurate digital twin model under highly uncertain conditions, thus avoiding potentially catastrophic consequences and ensuring the absolute safety of physical assets and engineering data. Maintain current operating mode: System instability index The preset fuse threshold is not exceeded In the case of , it indicates that the current comprehensive risk of the system is still within an acceptable range; at this time, the system will maintain the current operating mode, that is, the data acquisition module, model entropy calculation module, model authority calculation module and correction decision module will continue to work normally according to their design logic and perform probabilistic corrections to continuously optimize positioning accuracy; This embodiment provides the ultimate security guarantee for the system; it clarifies the specific triggering conditions and execution consequences of circuit breaking, transforming abstract risk assessment into specific and decisive system behavior; by automatically switching to safe redundancy mode and suspending all corrections, this mechanism can effectively stop damage before the system status completely deteriorates, avoiding catastrophic consequences caused by incorrect attribution; this not only protects the security of physical facilities and digital assets, but also maintains a relatively stable and predictable baseline state for subsequent manual troubleshooting and system recovery, greatly enhancing the fault response capability and ultimate reliability of the entire system.

[0031] Example 7 The correction decision module realizes negative feedback suppression of correction behavior through probabilistic screening; The acceptance of the correction vector leads to an increase in the model authority entropy, the increase in the model authority entropy leads to a decrease in the model authority, and the decrease in the model authority leads to a decrease in the probability of correction acceptance to inhibit subsequent corrections;

[0032] This embodiment describes in detail the negative feedback suppression mechanism implemented by the correction decision module. This mechanism is one of the core logics that distinguishes the present invention from traditional correction systems. Its purpose is to proactively suppress excessive correction behaviors that may lead to a collapse of the model's authority through an inherent, interconnected chain of cause and effect. The negative feedback suppression relies on the logical closed loop constructed between multiple modules and parameters in the system. The correction decision module directly constitutes the execution end of the negative feedback loop through its probabilistic screening mechanism. The conduction process of the entire negative feedback chain is as follows: The acceptance of the correction vector leads to an increase in the model authority entropy: when the correction decision module is based on its correction acceptance probability When a correction vector is decided, the correction behavior itself will be captured by the model entropy calculation module; the module will quantify this correction into a dimensionless entropy increment , and add it to the model authority entropy Therefore, every successful correction inevitably leads to The increase in , which in physical terms means that the uncertainty of the model has increased by one point; The increase in model authority entropy leads to a decrease in model authority: It will be passed as input to the model authority calculation module; according to the exponential decay formula: ;

[0033] The increase in model authority will inevitably lead to This step converts the accumulated entropy into an intuitive level of distrust; The decrease in model authority leads to a decrease in the probability of accepting the correction: the decreased model authority It will be used as input and returned to the correction decision module itself to calculate the probability of acceptance of the next round of corrections ; According to the formula: ;

[0034] As a multiplying factor, its own reduction will directly lead to reduction; This process forms a complete negative feedback loop: correction → entropy increase → authority decrease → correction probability decrease → inhibition of subsequent corrections; This embodiment clearly reveals the self-regulation and self-stabilization mechanism within the system; this negative feedback inhibition mechanism gives the system an inherent self-regulation capability; it is no longer a tool that passively executes instructions, but an adaptive system that can dynamically adjust its behavior and correct tendencies based on its own health status and model authority; this mechanism can effectively interrupt the vicious cycle of deviation-correction-greater deviation that is common in traditional systems, especially when facing continuous, nonlinear benchmark drift. It can prevent the system from blindly pursuing data fitting, which leads to the model being completely corrected beyond recognition, thereby ensuring the long-term effectiveness of the digital twin model and the robustness of the system.

[0035] Example 8 The reference correction scale, attenuation sensitivity coefficient, local accuracy maximum acceptable threshold, global deviation reference threshold, model entropy reference threshold, and fuse threshold are all preset parameters based on engineering design tolerances or previous simulation calibration;

[0036] In this embodiment, the sources and properties of multiple key preset parameters involved in the system are uniformly explained to ensure that those skilled in the art can understand the basis for setting these parameters and thus fully implement this technical solution; these parameters are the basis for the system to perform quantitative calculations and decision-making, and the rationality of their settings is directly related to the performance and reliability of the entire system; The parameters are preset based on engineering design tolerances or early simulation calibration. This means they are not variables that are dynamically calculated during system operation. Instead, they are constants that are set once or determined through a series of standardized experiments by engineers or designers based on the characteristics and requirements of the specific application scenario before system deployment. These parameters include: Reference correction scale : This parameter is used to normalize the correction vector components of different dimensions in the model entropy calculation module. Its source is the engineering design tolerance. For example, if the engineering specification requires the maximum installation displacement tolerance of the embedded part to be 5 mm, the reference correction scale of the displacement component can be set to 5 mm. This ensures that the entropy calculation is based on the deviation range allowed by the actual project. Decay sensitivity coefficient (k): This parameter is used in the model authority calculation module to control the rate at which authority decays as entropy increases. Its source is previous simulation calibration. Technicians can build a simulation system, input correction signals of varying strengths and frequencies, and observe the progression of model failure to calibrate an appropriate k value, defining the system's overall tolerance to model modifications. The highest acceptable threshold for local accuracy : This parameter is used to evaluate the immediate task performance in the correction decision module; the source is the engineering design tolerance, which directly corresponds to the accuracy requirement of the positioning task; for example, the contract stipulates that the local positioning error shall not exceed 2 mm, then It can be set to 2 mm; Global deviation reference threshold ( ) and the model entropy reference threshold ( ): These two parameters are used to normalize different risk factors in the risk assessment module; their source is the previous simulation calibration; through simulation experiments, we can find the critical statistical value that indicates that the system is about to enter an irreversible failure state; for example, the simulation shows that when the global root mean square error exceeds 10 mm or the model entropy accumulates to 50, the system will most likely crash, then these two values ​​can be used as and The basis for setting The early simulation calibration process can be implemented as follows: build a virtual immersed tube tunnel model with known nonlinear creep or external interference, let the system run in the simulation environment, and continuously record key indicators; through multiple experiments, it can be observed under what conditions and Under the combination of , the positioning error of the system begins to diverge uncontrollably. The statistical average of the critical value that causes divergence or a lower limit with a safety margin can be used as the setting and The scientific basis of Circuit breaker threshold ( ): This parameter is the threshold for triggering the final safety protection; its source can be engineering design tolerance, such as mandatory requirements of safety regulations or early simulation calibration. A value with a safety margin, such as 1.0, is set based on the system crash point in the simulation; This embodiment greatly enhances the feasibility and reproducibility of the technical solution by clarifying the sources and determination methods of all key preset parameters; it clearly indicates to technical personnel in this field that these parameters are not guesswork, but are based on evidence and laws, and their settings are closely linked to specific engineering backgrounds and scientific experimental methods; this not only provides technical personnel in this field with a clear parameter configuration guide, but also ensures that different users can reasonably and standardizedly configure the system according to their own engineering environment when deploying the system, thereby achieving the expected technical effect.

[0037] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A digital steel structure embedded parts precise positioning system, characterized by: include: The data acquisition module is used to obtain the correction vector of the digital twin model, the current measured local positioning accuracy, and the current global posture consistency deviation; The model entropy calculation module is used to calculate the model authority entropy that represents the cumulative deviation degree of the digital twin model based on the correction vector and the preset reference correction scale; A model authority calculation module is used to determine the model authority of the digital twin model based on the model authority entropy and a preset attenuation sensitivity coefficient; The correction decision module is used to combine the model authority and local positioning accuracy to generate the correction acceptance probability and perform probabilistic screening of the execution of the correction vector based on the correction acceptance probability; The risk assessment module is used to integrate the model authority entropy and the global posture consistency deviation, calculate the system instability index, and trigger the fuse protection mechanism when the system instability index exceeds the preset fuse threshold.

2. A digital steel structure embedded parts precise positioning system according to claim 1, characterized in that: The model entropy calculation module includes: The normalization unit is used to normalize each component of the correction vector according to its corresponding reference correction scale to generate a dimensionless entropy increment; the accumulation unit is used to determine the model authoritative entropy by recursively accumulating the dimensionless entropy increment.

3. A digital steel structure embedded parts precise positioning system according to claim 1, characterized in that: The model authority calculation module maps the model authority entropy to the model authority based on an exponential decay model, and the model authority and the model authority entropy are in an exponential decay relationship.

4. A digital steel structure embedded parts precise positioning system according to claim 1, characterized in that: The modified decision module is used to generate a modified acceptance probability, including: The local positioning accuracy is compared with the preset maximum acceptable threshold of local accuracy to generate a local accuracy performance factor; the correction acceptance probability is determined based on the product of the model authority and the local accuracy performance factor.

5. The digital steel structure embedded parts precise positioning system according to claim 1 is characterized in that: The risk assessment module is used to calculate the system instability index, including: The global posture consistency deviation is compared with the global deviation reference threshold to determine a first relative risk; the model authority entropy is compared with the model entropy reference threshold to determine a second relative risk; and the system instability index is determined based on the weighted sum of the first relative risk and the second relative risk.

6. The digital steel structure embedded parts precise positioning system according to claim 1, characterized in that: When the risk assessment module triggers the fuse protection mechanism, the system switches to a safety redundancy mode and terminates all modifications to the global model; when the system instability index does not exceed the preset fuse threshold, the system maintains the current operating mode.

7. The digital steel structure embedded parts precise positioning system according to claim 1, characterized in that: The correction decision module realizes negative feedback suppression of the correction behavior through probabilistic screening; The acceptance of the correction vector leads to an increase in the model authority entropy, the increase in the model authority entropy leads to a decrease in the model authority, and the decrease in the model authority leads to a decrease in the probability of correction acceptance to inhibit subsequent corrections.

8. The digital steel structure embedded parts precise positioning system according to claim 1, characterized in that: The reference correction scale, attenuation sensitivity coefficient, local accuracy maximum acceptable threshold, global deviation reference threshold, model entropy reference threshold and fuse threshold are all parameters preset based on engineering design tolerances or previous simulation calibration.

Images