A digital steel structure embedded part precision positioning system

Abstract

This invention discloses a precise positioning system for embedded parts in digital steel structures, belonging to the fields of computer-aided design and digital twin technology. It includes a data acquisition module for acquiring correction vectors for the digital twin model, the current measured local positioning accuracy, and the current global attitude consistency deviation; a model entropy calculation module for calculating the model authority entropy, representing the cumulative deviation of the digital twin model, based on the correction vectors and a preset reference correction scale; a model authority calculation module for determining the model authority of the digital twin model based on the model authority entropy and a preset attenuation sensitivity coefficient; and a correction decision module for generating a correction acceptance probability by combining the model authority and local positioning accuracy, and for probabilistically filtering the execution of the correction vector based on the correction acceptance probability. This invention provides a solid quantitative foundation for subsequent judgment of the model's health status.

Description

Technical Field

[0001] This invention relates to the fields of computer-aided design and digital twin technology, specifically to a digital steel structure embedded part precision positioning system. Background Technology

[0002] In existing technologies, the accuracy and reliability of digital positioning using Building Information Modeling (BIM) and surveying robots are built upon a stable and reliable coordinate reference. However, in extreme engineering scenarios such as deep-sea immersed tunnels, the structure experiences continuous and nonlinear creep due to water pressure and its own weight, causing dynamic changes in the coordinate reference itself. Simultaneously, the reference of internal sensors, such as inertial measurement units (IMUs), may also 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. It can even lead to a vicious cycle of deviation-correction-greater deviation due to incorrect attribution, ultimately resulting in a collapse of system credibility.

[0003] To address this issue, a new technical solution has been proposed. The core of this solution lies in establishing a mechanism that can quantify model credibility and make adaptive decisions accordingly. This mechanism no longer blindly pursues the apparent self-consistency of data, but rather conducts a comprehensive assessment of the system's health in a dynamic equilibrium, and includes a final circuit breaker protection to ensure that it can proactively switch to a safe and redundant mode before the model's authority collapses.

[0004] The information disclosed in the background section above is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a digital steel structure embedded part precision positioning system to solve the problems mentioned in the background art.

[0006] The technical solution of the present invention includes:

[0007] The data acquisition module is used to obtain the correction vector for the digital twin model, the current measured local positioning accuracy, and the current global attitude consistency deviation.

[0008] The model entropy calculation module is used to calculate the model authority entropy, which represents the cumulative deviation of the digital twin model, based on the correction vector and the preset reference correction scale.

[0009] The model authority calculation module is used to determine the model authority of the digital twin model based on the model authority entropy and the preset decay sensitivity coefficient.

[0010] The correction decision module is used to combine the model authority and local positioning accuracy to generate a correction acceptance probability, and to perform probabilistic filtering on the execution of the correction vector based on the correction acceptance probability.

[0011] The risk assessment module is used to integrate the model's authoritative entropy and the global attitude consistency deviation to calculate the system instability index, and to trigger the circuit breaker protection mechanism when the system instability index exceeds the preset circuit breaker threshold.

[0012] Preferably, the model entropy calculation module includes:

[0013] 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 authoritative entropy of the model by recursively accumulating the dimensionless entropy increment.

[0014] Preferably, the model authority calculation module maps model authority entropy to model authority based on an exponential decay model, and the model authority and model authority entropy have an exponential decay relationship.

[0015] Preferably, the correction decision module is used to generate a corrected acceptance probability, including:

[0016] The local positioning accuracy is compared with the preset highest acceptable threshold for local accuracy to generate a local accuracy performance factor; the corrected acceptance probability is determined based on the product of the model authority and the local accuracy performance factor.

[0017] Preferably, the risk assessment module is used to calculate the system instability index, including:

[0018] The global attitude consistency deviation is compared with the global deviation reference threshold to determine the first relative risk level; the model authority entropy is compared with the model entropy reference threshold to determine the second relative risk level; and the system instability index is determined based on the weighted sum of the first and second relative risk levels.

[0019] Preferably, when the risk assessment module triggers the circuit breaker protection mechanism, the system switches to a safe redundancy mode and stops all modifications to the global model; when the system instability index does not exceed the preset circuit breaker threshold, the system maintains the current operating mode.

[0020] Preferably, the correction decision module achieves negative feedback suppression of correction behavior through probabilistic screening;

[0021] The acceptance of the correction vector leads to an increase in the model authority entropy, which in turn leads to a decrease in the model authority. The decrease in the model authority, in turn, leads to a decrease in the probability of correction acceptance, thereby suppressing subsequent corrections.

[0022] Preferably, the reference correction scale, attenuation sensitivity coefficient, highest acceptable threshold for local accuracy, global deviation reference threshold, model entropy reference threshold, and circuit breaker threshold are all parameters preset according to engineering design tolerance or previous simulation calibration.

[0023] This invention provides an improved digital steel structure embedded part precision positioning system, which has the following improvements and advantages compared with the prior art:

[0024] 1. This solution transforms the abstract concept of model deviation into a measurable and accumulative physical quantity, namely the model authority entropy, by setting up a model entropy calculation module. The underlying mechanism is achieved through the quantitative accumulation of each correction action. To address the issue that the correction vector may contain components with different physical dimensions such as displacement and rotation, the solution provides the system with an objective and continuous memory that can record the cumulative impact of all historical corrections on the original state of the model, providing a solid quantitative basis for subsequent judgment of the model's health status.

[0025] 2. A potentially infinitely accumulating entropy value is non-linearly mapped to a credibility index between 0 and 1 that aligns with engineering intuition. The preset decay sensitivity coefficient is designed as a regulator, its value determined through preliminary simulations to define the system's overall tolerance to 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 continuously subjected to negative influences, making model authority a dynamic index sensitive to risk.

[0026] 3. Effective negative feedback is formed, which suppresses subsequent correction behavior, so that the algorithm recalculates the correction vector and the probability of the system accepting and executing it is significantly reduced, thereby effectively breaking the potential chain of deviation-correction-greater deviation.

[0027] 4. It provides an insurmountable safety baseline for the system, ensuring that it can proactively switch to a safe redundancy mode before the model's authority collapses, thus preventing catastrophic consequences caused by the system's confident but erroneous corrections. Attached Figure Description

[0028] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0029] Figure 1 This is a flowchart of the system of the present invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0031] Example 1

[0032] See also Figure 1 This invention provides a digital steel structure embedded part precision positioning system, comprising:

[0033] The data acquisition module is used to obtain the correction vector for the digital twin model, the current measured local positioning accuracy, and the current global attitude consistency deviation.

[0034] The model entropy calculation module is used to calculate the model authority entropy, which represents the cumulative deviation of the digital twin model, based on the correction vector and the preset reference correction scale.

[0035] The model authority calculation module is used to determine the model authority of the digital twin model based on the model authority entropy and the preset decay sensitivity coefficient.

[0036] The correction decision module is used to combine the model authority and local positioning accuracy to generate a correction acceptance probability, and to perform probabilistic filtering on the execution of the correction vector based on the correction acceptance probability.

[0037] The risk assessment module is used to integrate the model's authoritative entropy and the global attitude consistency deviation to calculate the system instability index, and to trigger the circuit breaker protection mechanism when the system instability index exceeds the preset circuit breaker threshold.

[0038] This invention provides a digital steel structure embedded component precision positioning system. The system aims to solve the technical problem in extreme engineering scenarios, such as deep-sea immersed tunnels, where the reliability of digital twin models gradually decreases due to dynamic changes in coordinate references, and may even fall into a vicious cycle of deviation-correction-greater deviation. The inherent logic of the technical solution lies in establishing a mechanism that can quantify model reliability, make adaptive decisions, and have final circuit breaker protection, thereby ensuring positioning accuracy while maintaining the long-term stability and reliability of the system.

[0039] The system in this embodiment forms a complete technical closed loop with its overall architecture and data flow process. The initial step of the system starts the workflow through a data acquisition module. The data acquisition module is the unit responsible for acquiring raw system status data from various sensors and measuring devices, and its purpose is to provide 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:

[0040] Correction vector for digital twin model: The correction vector refers to the adjustment amount that needs to be applied to the current digital twin model to compensate for measurement deviations, calculated by the positioning algorithm. It may include multiple components such as displacement and rotation, and is the direct driving force for changes in the model state.

[0041] Local positioning accuracy as measured: Local positioning accuracy refers to the actual performance of a single or short-term positioning task. It is usually measured by the error between the measured point and the corresponding point in the model. Its role is to characterize the real-time performance of the system in a specific task.

[0042] Current global attitude consistency deviation: Global attitude consistency deviation refers to the overall attitude deviation of the current physical structure relative to its digital twin model, which is assessed through a holistic analysis of multiple positioning points. Its role is to reflect the system-level, macroscopic deviation status.

[0043] After data acquisition, the process enters the model entropy calculation module. This module is one of the core quantization units of the system, aiming to transform discrete, multi-dimensional correction vectors into a single scalar that characterizes the cumulative deviation of the digital twin model. In this embodiment, the module calculates based on the acquired correction vectors and a preset reference correction scale. The reference correction scale refers to parameters preset according to engineering design tolerances or previous simulation calibrations, providing a unified evaluation benchmark for correction components of different physical dimensions. The output of this module is a quantification index called model authority entropy. Model authority entropy is a dimensionless value that borrows from the concept of information entropy, used to characterize the cumulative information distortion or uncertainty of the current digital twin model compared to its initial design benchmark. It serves as a key basis for all subsequent decisions.

[0044] The model authority calculation module processes the model authority entropy. The purpose of this module is to transform the abstract model authority entropy into a more intuitive credibility index 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 decay sensitivity coefficient. The decay sensitivity coefficient is a dimensionless parameter calibrated through prior simulation, defining the rate at which model authority decreases as model authority entropy increases. Model authority is a normalized measure of the degree of conformity or reliability of the current digital twin model to its original design baseline; a higher value indicates that the model is closer to the initial design state and has higher credibility.

[0045] The system's correction decision module intervenes; the correction decision module is the adaptive control center of the system, and its purpose is to avoid unconditionally executing all correction instructions, thereby interrupting potential vicious correction cycles; in this embodiment, the 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 that the system decides to adopt and execute the current correction vector; the module then performs probabilistic screening on the execution of the correction vector based on this probability, that is, accepts or rejects the correction with a certain probability;

[0046] To ensure overall system safety, the risk assessment module operates in parallel. This module serves as the system's safety guarantee and final line of defense, aiming to integrate risk indicators from different dimensions to comprehensively assess the overall stability of the system and trigger protection mechanisms when the risk is too high. In this embodiment, the module integrates model authority entropy, representing the chronic, hidden risk of model health decline, and global attitude consistency deviation, representing the immediate, observable global deviation risk, to calculate a system instability index. The system instability index is a comprehensive risk indicator used to quantify the probability of system collapse. When this index exceeds a preset circuit breaker threshold, the system will trigger a circuit breaker protection mechanism. The circuit breaker threshold is a critical value preset based on simulation or engineering experience, indicating that the system is about to enter an irreversible failure state.

[0047] This embodiment constructs a complete closed-loop control system from state quantification to adaptive decision-making and risk circuit breaking through the collaborative work of the above modules; it no longer blindly pursues the apparent self-consistency of data, but comprehensively evaluates the health of the system in a dynamic balance; the system can effectively quantify and suppress the erosion of authority of the digital twin model caused by continuous correction, avoid the vicious cycle of deviation-correction-greater deviation, ensure accurate positioning capability, and greatly improve the long-term operational reliability and credibility of the system in extreme engineering environments;

[0048] This invention provides a digital steel structure embedded part precision positioning system. Its advancement over existing technologies lies in changing the logic of the system in handling positioning deviations. Existing technologies, when using building information models and surveying robots for positioning, rely on a static coordinate reference that is assumed to be absolutely reliable for accuracy and reliability. However, when encountering extreme engineering scenarios such as deep-sea immersed tunnels, the nonlinear creep of the structure caused by water pressure and its own weight can shake the stability of the coordinate reference. Existing technologies will fall into a cycle of continuously modifying the digital twin model to forcibly fit the observed data. This process will gradually erode the original authority of the model design and may eventually lead to a complete collapse of the system's credibility due to incorrect attribution.

[0049] The core advancement of this invention lies in the introduction of a quantitative self-reflection and proactive defense mechanism; it no longer assumes the absolute authority of the digital twin model, but instead continuously evaluates its credibility as a dynamic variable and makes adaptive decisions based on the evaluation results, setting an insurmountable security baseline.

[0050] Example 2

[0051] The model entropy calculation module includes:

[0052] 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 authoritative entropy of the model by recursively accumulating the dimensionless entropy increment.

[0053] In this embodiment, the specific implementation of the model entropy calculation module is defined. The purpose of this module is to solve the problem that the correction vector may contain components with different physical dimensions such as translation and rotation, which makes it impossible to directly accumulate them. To this end, the module is divided into two core units: a normalization unit and an accumulation unit.

[0054] The normalization unit aims to eliminate the physical dimensions of each component in the correction vector, allowing for unified mathematical processing. To further clarify, this unit is used to calculate the dimensionless entropy increment at discrete time step t. , will correct the vector Each component According to their respective corresponding reference correction scales Perform normalization processing;

[0055] This refers to the dimensionless increment introduced by the correction action at the current moment, representing an increase in model uncertainty. It serves as the basic unit for accumulating the model's authoritative entropy; the calculation formula is:

[0056] ;

[0057] in, Refers to the correction vector The i-th component, for example, the displacement correction in the x-axis direction or the rotation correction around the z-axis, is derived from the system's positioning algorithm.

[0058] This refers to the reference correction scale for the i-th component, such as the maximum allowable displacement correction or maximum attitude correction. Its source is based on engineering design tolerances or parameters preset in previous simulation calibrations, and its dimensions are similar to... same;

[0059] i: Correction vector The index of the i-th component; t: discrete time step or time point;

[0060] It refers to a global entropy scaling factor, which is a dimensionless adjustable parameter. It is preset after the sensitivity of the system to comprehensive correction is calibrated through early simulation.

[0061] By dividing each correction component by its corresponding reference scale, the unit ensures that every calculated term is a dimensionless value, thereby generating a dimensionless entropy increment.

[0062] Square operation here It serves a dual purpose: it ensures that all entropy increments are positive, consistent with the physical intuition that entropy always increases or remains constant; it imposes a stronger penalty weight on larger relative deviations, meaning that corrections that exceed the tolerance are considered to have a greater impact on the model's original authority, thus significantly increasing the model's authority entropy. (Summarization sign) This means that the entropy increments contributed by all correction components, such as x, y, z translations and rotations, are summed to form the total entropy increment at the current moment. ;

[0063] The purpose of the accumulation unit is to transform the instantaneous entropy increment into an indicator that reflects the historical cumulative effect; in this embodiment, this unit calculates the dimensionless entropy increment generated by the normalization unit. Perform recursive accumulation to determine the final model authority entropy. The implementation follows the recursive formula below:

[0064] ;

[0065] in, The dimensionless authoritative entropy of the model at time t is a core indicator for measuring the cumulative deviation of the model; the data type is a dimensionless floating-point number; the source is the model entropy of the current accumulation unit at the previous time step. and the entropy increment at the current moment Calculated based on the above;

[0066] The authoritative entropy of the model at the previous time step; the data type is a dimensionless floating-point number; the source is the result of previous calculations stored by the system.

[0067] : The dimensionless entropy increment at the current moment; its data type is a dimensionless floating-point number; it is obtained by calculation from the normalization unit within this module;

[0068] By introducing normalization and accumulation units, this embodiment provides a concrete, feasible, and logically rigorous method for calculating model authority entropy. It not only solves the fundamental problem of mixed calculation of heterogeneous physical quantities, such as length and angle, and ensures the mathematical consistency of model authority entropy as a dimensionless scalar, but also, through recursive accumulation, enables it to truly reflect the continuous impact of all historical correction operations on model authority, providing a more reliable and accurate input for subsequent authority calculation and decision-making.

[0069] This scheme transforms the abstract concept of model deviation into a measurable and accumulative physical quantity, namely the model authority entropy, by setting up a model entropy calculation module. Its underlying mechanism is achieved through the quantitative accumulation of each correction action. To address the issue that the correction vector may contain components with different physical dimensions such as displacement and rotation, the calculation of the entropy increment normalizes each component.

[0070] dimensionless entropy increment at the current moment Through formula ,in, Global entropy scaling factor. Correction vector The i-th component, 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 Each of them is a corresponding reference scale For example, the maximum allowable displacement or maximum attitude correction has been normalized; these reference scales All parameters are preset according to the design tolerance of the project, which ensures that each term in the summation is a dimensionless value;

[0071] Model Authority Entropy It is a dimensionless cumulative value, obtained through a recursive formula. ,in, Model authority entropy, Entropy increment, t: time is determined; 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 progress 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, providing a solid quantitative basis for subsequent judgment of the health status of the model.

[0072] Example 3

[0073] The model authority calculation module maps model authority entropy to model authority based on an exponential decay model. The model authority and model authority entropy have an exponential decay relationship.

[0074] In this embodiment, the implementation method of the model authority calculation module is limited; the purpose of this module is to calculate the model authority entropy, which has cumulative properties and can theoretically grow indefinitely. This maps to a model authority with clear boundaries, between 0 and 1, making it easier to make decisions and understand. ;

[0075] To achieve this mapping, the model authority calculation module in this embodiment employs a calculation method based on an exponential decay model. This method ensures that there is a non-linear exponential decay relationship between model authority and model authority entropy. The calculation logic is embodied in the following formula:

[0076] ;

[0077] in, Model authority is a normalized metric between 0 and 1, used to intuitively represent the current credibility of a digital twin model; its data type is a dimensionless floating-point number; and its source is determined by this module based on the current model authority entropy. Calculated;

[0078] e: the base of the natural logarithm;

[0079] k: A dimensionless attenuation sensitivity coefficient, used to adjust the model's authority. Authoritative entropy of the model Increased sensitivity; a larger k value means that even small cumulative corrections will lead to a rapid decrease in model authority; the data type is a dimensionless floating-point number; this is a parameter preset based on the system's overall tolerance to model modifications, and its value is usually calibrated through preliminary simulations;

[0080] The current dimensionless model authoritative entropy; the data type is a dimensionless floating-point number; the source is calculated by the preceding model entropy calculation module;

[0081] This exponential decay model ensures that when the system is in an ideal initial state, i.e., the model is unmodified, At this point, the model's authority The highest level of authority for the model is indicated by the number 1; as corrections accumulate, Gradually increase It then smoothly decays from 1 to 0, which accurately simulates the objective law that trust or authority gradually diminishes after being continuously affected by negative events.

[0082] An exponential decay model is employed to provide a physically and mathematically sound approach for calculating model authority. This exponential decay relationship accurately characterizes the nonlinear decay process of model authority under continuous correction: initially, small entropy increases have little impact on authority; however, as entropy accumulates, the decline in authority accelerates. This makes the model authority... This becomes an indicator that is sensitive to changes in the health status of the model and conforms to engineering logic, thereby providing more refined and dynamic control inputs for downstream correction decision modules;

[0083] Based on the calculated model authority entropy Authority of a normalized model This was further derived; the derivation process borrowed from the exponential decay model in reliability engineering, specifically through the formula...

[0084] ,in, Model authority, k: dimensionless attenuation sensitivity coefficient. The current implementation of the dimensionless model entropy;

[0085] The practical significance of this formula lies in nonlinearly mapping a potentially infinitely accumulating entropy value to a credibility index between 0 and 1 that aligns with engineering intuition. The preset decay sensitivity coefficient k is intended to act as a regulator, its value determined through preliminary simulations to define the system's overall tolerance to model modifications. A larger 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 continuously subjected to negative influences, making model authority a dynamic index sensitive to risk.

[0086] Example 4

[0087] The revised decision module is used to generate revised acceptance probabilities, including:

[0088] The local positioning accuracy is compared with the preset highest acceptable threshold for local accuracy to generate a local accuracy performance factor; the corrected acceptance probability is determined based on the product of the model authority and the local accuracy performance factor.

[0089] In this embodiment, the correction decision module is used to generate the correction acceptance probability. The specific logic is defined; the purpose of this module is to intelligently suppress further, potentially harmful, self-correcting behavior when the model's authority decreases, thereby forming effective negative feedback.

[0090] To achieve this goal, the computational logic of the revised decision module is designed to comprehensively consider information from two dimensions: the long-term health of the model and the model's authority. The real-time task performance of the system is reflected in the local positioning accuracy. The calculation process involves two steps:

[0091] Local positioning accuracy With a preset local accuracy highest acceptable threshold The comparisons are performed to generate a local precision performance factor; the local precision performance factor is a dimensionless value between 0 and 1 used to quantify the quality of the current positioning task, and its calculation formula is as follows:

[0092] ;

[0093] in, This refers to the local positioning accuracy of the current measurement, such as the root mean square error between the measured point and the model point. It is obtained in real time by the measuring equipment and its dimension is length.

[0094] This refers to the preset, highest acceptable threshold for local accuracy. It represents the maximum positioning error that the engineering requirements can tolerate, and it originates from parameters preset according to the engineering design requirements; its dimensions are... The fact that they are identical ensures that the ratio between them is dimensionless;

[0095] Based on model authority The corrected acceptance probability is finally determined by multiplying the result with the local precision performance factor generated in the previous step. The complete calculation formula is as follows:

[0096] ;

[0097] in, : Corrected acceptance probability, representing the probability that the system accepts and executes the current corrected vector; its data type is a dimensionless floating-point number, a probability value; its source is calculated from the authority and local precision performance of the comprehensive model in this module; A: subscript, indicating acceptance;

[0098] : Model authority; data type is dimensionless floating-point number; source is calculated by the model authority calculation module; M: subscript, referring to the model;

[0099] : Current local positioning accuracy; dimension is length; source is obtained in real time by the data acquisition module; L: subscript, referring to the local area;

[0100] : The preset maximum acceptable threshold for local accuracy; the dimension is length; the source is a parameter preset according to engineering requirements; ref: subscript, referring to a reference;

[0101] Therefore, this formula constructs a decision gate with logic: only when the model itself has good long-term health, i.e. Large and currently performs well in real-time measurements, i.e., local positioning accuracy. Much smaller than the threshold When the ratio term is close to 1, the acceptance probability is adjusted. Only then will it approach its upper limit. If either party performs poorly, the final acceptance probability will be significantly reduced, thus achieving intelligent and prudent decision-making.

[0102] This embodiment provides an adaptive and dynamic decision-making mechanism; it ensures that the corrective decision is not an isolated judgment, but a result of comprehensive consideration: when the model has high authority and accurate local localization, near The system tends to accept corrections to optimize the model; however, when the model's authority has significantly decreased, or the local positioning accuracy is close to the allowable limit, The value will approach 0, and the system will likely reject the correction request. This design not only makes the decision-making logic more complete, but more importantly, it directly links the core indicator of model authority with the actual behavior of the system, whether or not to make corrections, forming an effective negative feedback loop. This can effectively break potential vicious cycles and improve the system's adaptability and robustness.

[0103] The correction decision module in this scheme no longer executes all correction instructions unconditionally, but instead uses a probabilistic gating mechanism for selection. Its core function is to generate negative feedback to suppress potential vicious cycles; the acceptance probability of correction actions... Based on model authority And the system's local task performance, i.e., local positioning accuracy To decide together;

[0104] Its decision-making logic is solidified in formulas ,in, : Adjust the acceptance probability, Model authority The current local positioning accuracy. The formula's advancement lies in establishing a trade-off mechanism: when model authority is high and local accuracy is good, the system tends to accept corrections for improvement; conversely, if model authority has significantly decreased, or local accuracy has approached its allowable limit, the system will not accept corrections. At that time, the acceptance probability will approach 0, and the system will likely reject the correction request. It is a rigid indicator preset according to the project requirements;

[0105] This mechanism creates an effective negative feedback loop: corrective behavior leads to... Increase, which in turn leads to The decline eventually led to This reduces the initial value, thereby suppressing subsequent corrective actions; to illustrate with a concrete example: assuming the initial... The local accuracy is good, and the system accepts corrections with a high probability; after multiple consecutive corrections... Accumulated from 0 to 2.0, resulting in The value drops to 0.4; at this point, even if the algorithm recalculates the correction vector, the probability of the system accepting and executing it has been significantly reduced, thus effectively breaking the potential chain of deviation-correction-greater deviation.

[0106] Example 5

[0107] The risk assessment module is used to calculate the system instability index, including:

[0108] The global attitude consistency deviation is compared with the global deviation reference threshold to determine the first relative risk; the model authority entropy is compared with the model entropy reference threshold to determine the second relative risk; and the system instability index is determined based on the weighted sum of the first and second relative risk.

[0109] In this embodiment, the risk assessment module is used to calculate the system instability index. The method was limited; the purpose of this module is to integrate two different dimensions of risk in the system—immediate global bias risk and chronic model health decline risk—in order to derive a comprehensive indicator that can fully reflect the possibility of the system being on the verge of failure.

[0110] To achieve this integration of multi-dimensional risks, the risk assessment module in this embodiment adopts a weighted summation model derived from the field of multi-criteria decision analysis; the calculation process of this method includes the following steps:

[0111] Global attitude consistency deviation With a global deviation reference threshold A comparison is made to determine the first relative risk level; the first relative risk level refers to a normalized measure of the currently observable global deviation risk, calculated as follows: ;

[0112] in, It refers to the current global attitude consistency deviation, which is usually the root mean square error of all positioning points relative to an optimal fitting reference surface. It is derived from real-time measurement and calculation by the system and its dimension is length.

[0113] This refers to the global deviation reference threshold, which is derived from previous simulations or historical data analysis. It represents a critical statistical value that indicates the system's immediate performance is about to enter an unacceptable state. Its dimensions are... same;

[0114] The model's authority entropy With a model entropy reference threshold A comparison is made to determine the second relative risk; the second relative risk refers to a normalized measure of the chronic risk of model health deterioration, and it is calculated as follows: ;

[0115] in, This refers to the current authoritative entropy of the model, which is continuously calculated by the model entropy calculation module and is a dimensionless value.

[0116] It refers to the model entropy reference threshold, which is determined through previous simulations and indicates that the cumulative deviation of the model is about to reach an irreversible level. It is also a dimensionless value.

[0117] Finally, the system instability index is determined by a weighted sum of the first and second relative risk levels. The complete calculation formula is as follows:

[0118] ;

[0119] in, The system instability index is a comprehensive, dimensionless risk indicator; it is derived from a weighted summation calculation performed by this module.

[0120] Risk weighting factor, a dimensionless parameter between 0 and 1, reflects the decision-maker's trade-off between immediate performance risk and long-term health risk; it is derived from the preset tolerance for different risks based on the specific project.

[0121] Risk weighting factor Essentially, it acts as a risk preference regulator. For example, in the early stages of an engineering project or during critical docking phases, when extremely high real-time positioning accuracy is required, a relatively large risk preference can be set. Values, such as 0.7, are used to focus on monitoring global attitude consistency deviations. In the long-term settlement observation phase, more attention may be paid to the long-term health drift of the model, in which case a smaller threshold can be set. Values, such as 0.3, are used to emphasize the authoritative entropy of the monitoring model. This accumulated flexibility allows the system to adapt to different engineering stages and application scenarios.

[0122] The definition is the same as above;

[0123] This embodiment provides a structured and quantifiable method for generating the system instability index through this weighted summation calculation method; it successfully integrates two completely different types of risk sources, immediate / macro vs. chronic / cumulative, within the same mathematical framework, and uses weighting factors... It provides flexible adjustment capabilities, allowing the risk assessment model to be adapted to different engineering scenarios; this enables... 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 basis for decision-making in triggering the circuit breaker protection mechanism.

[0124] To achieve ultimate system-level safety redundancy, a system instability index This index is introduced to integrate two different dimensions of risk in a system: immediate, observable global bias risk, and chronic, hidden risk of declining model health. The index calculation model comes from multi-criteria decision analysis in the field of risk assessment, and integrates multiple risk factors of different natures through weighted summation;

[0125] Its calculation formula is: ,in, System instability index Risk weighting factor Current global attitude consistency deviation Global deviation reference threshold Current model entropy Model entropy reference threshold; the practical significance of this formula lies in passing two risks of different natures through their respective reference thresholds. and Normalize it to make it a dimensionless relative risk level, and then apply risk weighting factors. A trade-off is made; these reference thresholds are critical statistical values ​​determined through prior simulations that indicate the system is about to enter an irreversible failure state.

[0126] once Exceeding the preset circuit breaker threshold The system then triggers a circuit breaker mechanism, proactively switching from a self-consistent mode that pursues global optimization to a primitive mode that only guarantees core security, ceasing all modifications to the global model and awaiting external manual intervention. This final circuit breaker protection mechanism is a decisive advancement of this solution compared to existing technologies. It provides the system with an insurmountable security baseline, ensuring that it can proactively switch to a safe redundancy mode before the model's authority collapses, thus preventing catastrophic consequences caused by the system's confident but erroneous modifications.

[0127] Example 6

[0128] When the risk assessment module triggers the circuit breaker protection mechanism, the system switches to a safe redundancy mode and suspends all modifications to the global model; when the system instability index does not exceed the preset circuit breaker threshold, the system maintains the current operating mode.

[0129] In this embodiment, the behavior of the risk assessment module after triggering the circuit breaker protection mechanism is limited; the purpose of this is to clarify how the system can safely transition from the conventional mode of pursuing optimal performance to the protection mode with the primary goal of ensuring core functions and data integrity after detecting an irreversible risk.

[0130] The risk assessment module calculates the system instability index. Then, it will be compared with a preset circuit breaker threshold. Perform real-time comparisons; It is a dimensionless constant, for example, it can be set to 1.0, and its source is the maximum risk limit that the system can withstand, determined according to engineering safety specifications or simulation experiments;

[0131] Based on this comparison result, the system will perform one of the following two mutually exclusive operations:

[0132] Triggering the circuit breaker mechanism: When the system instability index... Exceeding the preset circuit breaker threshold When the system reaches a critical point, it indicates that the overall risk has reached a critical point and there is a very high probability that it will fall into an uncontrollable state. At this time, the system will automatically trigger the circuit breaker protection mechanism. Under this mechanism, the system will switch to the safety redundancy mode. The safety 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 stop all modifications to the global model. This means that the modification decision module will be bypassed or forced to output a rejection result, thereby completely cutting off the modification path that may lead to further expansion of risk. After that, the system will remain in the safety mode and issue an alarm, waiting for external manual intervention or review.

[0133] This circuit breaker protection mechanism is the ultimate safety bottom line that distinguishes this system from traditional best-effort correction systems. It fundamentally prevents the system from continuing to make confident and erroneous corrections based on a potentially seriously distorted digital twin model under highly uncertain conditions, thereby avoiding potential catastrophic consequences and ensuring the absolute safety of physical assets and engineering data.

[0134] Maintain current operating mode: (Based on system instability index) The preset circuit breaker threshold has not been exceeded. In this case, it indicates that the overall risk of the system is still within an acceptable range. At this time, the system will maintain its 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.

[0135] 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 concrete and decisive system behavior; by automatically switching to a safe redundancy mode and aborting all corrections, this mechanism can effectively stop damage before the system state deteriorates completely, avoiding catastrophic consequences caused by incorrect attribution; this not only protects the security of physical facilities and digital assets, but also preserves 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.

[0136] Example 7

[0137] The correction decision module uses probabilistic filtering to suppress negative feedback on correction actions;

[0138] The acceptance of the correction vector leads to an increase in the model authority entropy, which in turn leads to a decrease in the model authority. The decrease in the model authority, in turn, leads to a decrease in the probability of correction acceptance, thereby suppressing subsequent corrections.

[0139] In this embodiment, the negative feedback suppression mechanism implemented by the correction decision module is explained in detail. This mechanism is one of the core logics that distinguishes this invention from traditional correction systems. Its purpose is to actively suppress excessive correction behavior that may lead to the collapse of the model's authority through an inherent, interconnected causal chain.

[0140] The negative feedback suppression relies on a logical closed loop constructed between multiple modules and parameters within the system; the correction decision module, through its probabilistic filtering mechanism, directly constitutes the execution end of this negative feedback loop; the entire negative feedback chain transmission process is as follows:

[0141] The acceptance of the modified vector leads to an increase in the model's authority entropy: when the modification decision module bases its modified acceptance probability... When deciding to accept a correction vector, this correction behavior itself is captured by the model entropy calculation module; this module quantifies this correction as a dimensionless entropy increment. And add it to the model authority entropy. Therefore, every successful correction inevitably leads to... The increase in this value, in a physical sense, represents an increase in the uncertainty of the model.

[0142] An increase in model authority entropy leads to a decrease in model authority: Increased model authority entropy It will be passed as input to the model authority calculation module; according to the exponential decay formula:

[0143] ;

[0144] The increase will inevitably lead to a decrease in the model's authority. The exponential decrease; this step transforms the accumulated entropy into an intuitive level of distrust;

[0145] A decrease in model authority leads to a decrease in the probability of correction acceptance: the decreased model authority It will then be returned as input to the correction decision module itself, used to calculate the correction acceptance probability for the next round. According to the formula:

[0146] ;

[0147] As a product factor, its own decrease will directly lead to The reduction;

[0148] This process forms a complete negative feedback loop: correction → entropy increase → authority decrease → correction probability decrease → suppression of subsequent corrections;

[0149] This embodiment clearly reveals the system's internal self-regulation and self-stabilization mechanisms. This negative feedback suppression mechanism enables the system to possess inherent self-regulation capabilities. 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 break the vicious cycle of deviation-correction-greater deviation common in traditional systems. Especially when facing continuous and nonlinear benchmark drift, it can prevent the system from blindly pursuing data fitting and causing the model to be completely corrected beyond recognition, thereby ensuring the long-term effectiveness of the digital twin model and the robustness of the system.

[0150] Example 8

[0151] The reference correction scale, attenuation sensitivity coefficient, maximum acceptable threshold for local accuracy, global deviation reference threshold, model entropy reference threshold, and circuit breaker threshold are all parameters preset according to engineering design tolerance or previous simulation calibration.

[0152] In this embodiment, the sources and properties of several 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, thereby fully implementing this technical solution; these parameters are the basis for the system to perform quantitative calculations and decisions, and the rationality of their setting is directly related to the performance and reliability of the entire system;

[0153] All parameters are preset based on engineering design tolerances or preliminary simulation calibration; this means that they are not variables that are dynamically calculated during system operation, but constants that are set once or determined through a series of standardized experiments by engineers or designers before system deployment, based on the characteristics and requirements of specific application scenarios.

[0154] These parameters include:

[0155] 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 parts to be 5 mm, then the reference correction scale of the displacement component can be set to 5 mm; this ensures that the entropy calculation is based on the allowable deviation range of the actual project.

[0156] Attenuation sensitivity coefficient (k): This parameter is used to control the rate at which authority decays with increasing entropy in the model authority calculation module; it is derived from the previous simulation calibration; technicians can build a simulation system, input correction signals of different intensities and frequencies, observe the process of model failure, and thus calibrate a suitable k value to define the system's overall tolerance to model modification;

[0157] The highest acceptable threshold for local accuracy This parameter is used to evaluate immediate task performance in the corrective decision module; it originates from engineering design tolerances and directly corresponds to the accuracy requirements of the positioning task; for example, if the contract stipulates that the local positioning error must not exceed 2 mm, then... It can be set to 2 millimeters;

[0158] Global Deviation Reference Threshold ( ) and model entropy reference threshold ( These two parameters are used to normalize different risk factors in the risk assessment module; they are derived from previous simulation calibration. Through simulation experiments, critical statistical values ​​that indicate the system is about to enter an irreversible failure state can be found. For example, simulations show that when the global root mean square error exceeds 10 mm or the model entropy accumulates to 50, the system is likely to collapse. In this case, these two values ​​can be used as... and The basis for its setting;

[0159] The initial simulation calibration process can be implemented as follows: Construct a virtual immersed tunnel model that includes known nonlinear creep or external disturbances, run the system in this simulation environment, and continuously record key indicators; through multiple experiments, it is possible to observe under what conditions... and Under these combined conditions, the system's positioning error begins to diverge uncontrollably. Taking the statistical average of the critical values ​​leading to this divergence, or a lower limit with a safety margin, can be used as the setting... and The scientific basis;

[0160] Circuit breaker threshold ( This parameter is the threshold for triggering the final safety protection; its source can be engineering design tolerances, such as mandatory provisions in safety specifications or preliminary simulation calibration. A value with a safety margin, such as 1.0, is set based on the system crash point in the simulation.

[0161] This embodiment greatly enhances the feasibility and reproducibility of the technical solution by clearly defining the source and determination method of all key preset parameters. It clearly indicates to those skilled in the art that these parameters are not guesswork, but are based on evidence and methods, and their setting is closely linked to the specific engineering background and scientific experimental methods. This not only provides those skilled in the art with a clear parameter configuration guide, but also ensures that different users can configure the system reasonably and in a standardized manner according to their own engineering environment when deploying the system, thereby achieving the expected technical effect.

[0162] 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 part precision positioning system, characterized in that, include: The data acquisition module is used to obtain the correction vector for the digital twin model, the current measured local positioning accuracy, and the current global attitude consistency deviation. The model entropy calculation module is used to calculate the model authority entropy, which represents the cumulative deviation of the digital twin model, based on the correction vector and the preset reference correction scale. The model authority calculation module is used to determine the model authority of the digital twin model based on the model authority entropy and the preset decay sensitivity coefficient. The correction decision module is used to combine the model authority and local positioning accuracy to generate a correction acceptance probability, and to perform probabilistic filtering on the execution of the correction vector based on the correction acceptance probability. The risk assessment module is used to integrate the model's authoritative entropy and the global attitude consistency deviation to calculate the system instability index, and to trigger the circuit breaker protection mechanism when the system instability index exceeds the preset circuit breaker threshold. 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 formula for calculating the dimensionless entropy increment is: ； in, Refers to the correction vector The i-th component, the displacement correction in the x-axis direction or the rotation correction around the z-axis, is derived from the system's positioning algorithm. This refers to a reference correction scale for the i-th component, derived from parameters preset based on engineering design tolerances or prior simulation calibration. Its dimensions are similar to... Same; i: Correction vector The index of the i-th component; t: discrete time step or time point; It refers to a global entropy scaling factor, which is a dimensionless adjustable parameter. It is preset after the sensitivity of the system to comprehensive correction is calibrated through early simulation. The accumulation unit is used to determine the model authority entropy by recursively accumulating the dimensionless entropy increment; the formula for calculating the model authority entropy is: ； in, The dimensionless authoritative entropy of the model at time t is a core indicator for measuring the cumulative deviation of the model; the data type is a dimensionless floating-point number; the source is the model entropy of the current accumulation unit at the previous time step. and the entropy increment at the current moment Calculated based on the above; The authoritative entropy of the model at the previous time step; the data type is a dimensionless floating-point number; the source is the result of previous calculations stored by the system. : The dimensionless entropy increment at the current moment; its data type is a dimensionless floating-point number; it is obtained by calculation from the normalization unit within this module; The model authority calculation module maps model authority entropy to model authority based on an exponential decay model, where model authority and model authority entropy exhibit an exponential decay relationship. The formula for calculating model authority is: ； in, Model authority is a normalized metric between 0 and 1, used to intuitively represent the current credibility of a digital twin model; its data type is a dimensionless floating-point number; and its source is determined by this module based on the current model authority entropy. Calculated; e: base of the natural logarithm; k: dimensionless attenuation sensitivity coefficient, used to adjust the model's authority. Authoritative entropy of the model Increased sensitivity; data type is dimensionless floating-point number; The current dimensionless model authoritative entropy; the data type is a dimensionless floating-point number; the source is calculated by the preceding model entropy calculation module; The revised decision module is used to generate a revised acceptance probability, including: The local positioning accuracy is compared with a preset highest acceptable threshold for local accuracy to generate a local accuracy performance factor; the formula for calculating the local accuracy performance factor is: ； in, It refers to the local positioning accuracy of the current measurement, the root mean square error between the measured point and the model point, which is collected in real time by the measuring equipment; its dimension is length. This refers to the preset maximum acceptable threshold for local accuracy, which is derived from parameters preset according to engineering design requirements; The corrected acceptance probability is determined by multiplying the model authority and the local accuracy performance factor; the formula for calculating the corrected acceptance probability is as follows: ； in, : Corrected acceptance probability, representing the probability that the system accepts and executes the current corrected vector; its data type is a dimensionless floating-point number, a probability value; it is derived from the authority and local precision performance of the comprehensive model in this module; A: subscript, indicating acceptance; : Model authority; data type is dimensionless floating-point number; source is calculated by the model authority calculation module; M: subscript, referring to the model; : Current local positioning accuracy; dimension is length; source is obtained in real time by the data acquisition module; L: subscript, referring to the local area; The preset local accuracy is the highest acceptable threshold; the dimension is length; the source is a parameter preset according to engineering requirements. Subscript: refers to a reference; The risk assessment module is used to calculate the system instability index, including: The global attitude consistency deviation is compared with a global deviation reference threshold to determine the first relative risk level; the model authority entropy is compared with a model entropy reference threshold to determine the second relative risk level; the system instability index is determined by a weighted sum of the first and second relative risk levels; the formula for calculating the system instability index is as follows: ； in, System instability index Risk weighting factor Current global attitude consistency deviation Global deviation reference threshold Current model entropy Model entropy reference threshold; When the risk assessment module triggers the circuit breaker protection mechanism, the system switches to a safe redundancy mode and stops all modifications to the global model; when the system instability index does not exceed the preset circuit breaker threshold, the system maintains the current operating mode.

2. The digital steel structure embedded part precise positioning system according to claim 1, characterized in that, The correction decision module achieves negative feedback suppression of correction behavior through probabilistic screening; The acceptance of the correction vector leads to an increase in the model authority entropy, which in turn leads to a decrease in the model authority. The decrease in the model authority, in turn, leads to a decrease in the probability of correction acceptance, thereby suppressing subsequent corrections.

3. The digital steel structure embedded part precise positioning system according to claim 1, characterized in that, The reference correction scale, attenuation sensitivity coefficient, highest acceptable threshold for local accuracy, global deviation reference threshold, model entropy reference threshold, and circuit breaker threshold are all parameters preset according to engineering design tolerances or previous simulation calibration.

Images