Intelligent construction site auxiliary decision-making system based on augmented reality

Abstract

The invention relates to the technical field of intelligent construction and human-computer interaction, in particular to an intelligent construction site auxiliary decision-making system based on augmented reality. The system comprises a data acquisition module which acquires physiological and behavior data of an operator in real time through wearable sensing equipment and augmented reality glasses; the first processing module is used for generating a collaboration trust index for quantifying a man-machine collaboration state based on the collected data; the second processing module is used for evaluating and outputting a current overall failure risk through a nonlinear evolution model in combination with the cooperative trust index and a preset basic risk; the self-adaptive decision-making module is used for determining and outputting a final decision-making scheme by combining two utility functions of efficiency priority and toughness priority according to the total failure risk; according to the method, the cooperation trust index is introduced, the physiological and behavior data of the operator is converted into the quantitative state index, the problems that in the prior art, the man-machine cooperation state is fuzzy, and precision measurement is difficult are solved, and a data basis is provided for risk assessment and self-adaptive decision making.

Description

Technical Field

[0001] This invention relates to the field of intelligent construction and human-computer interaction technology, specifically to an intelligent construction site auxiliary decision-making system based on augmented reality. Background Technology

[0002] In the current intelligent construction environment, human-machine collaboration is becoming increasingly common, but a series of challenges exist. First, the physiological state of operators, such as fatigue, confusion, and trust in automated instructions, is dynamically changing, and these human factors are difficult to quantify precisely, leading to ambiguity in the state of human-machine collaboration. Second, existing risk assessment models are usually static, and they cannot effectively capture the non-linear and accelerating growth characteristics of system failure risks caused by changes in operator state, especially a decline in trust. Finally, decision support systems often employ fixed and rigid logic, and cannot adaptively adjust between ensuring construction efficiency and operational safety based on real-time risk assessment levels.

[0003] These problems collectively lead to sluggish system response and suboptimal decision-making in complex and ever-changing construction sites, making it difficult to effectively prevent safety accidents caused by human factors. Therefore, how to quantify the state of human-machine collaboration, dynamically assess human risk, and achieve adaptive switching of decision-making logic has become an urgent technical problem to be solved. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a smart construction site auxiliary decision-making system based on augmented reality. Specifically, the technical solution of this invention includes:

[0005] The data acquisition module is used to acquire multi-source physiological and behavioral data of the operator in real time through wearable sensing devices and augmented reality glasses;

[0006] The first processing module is used to generate a collaborative trust index that characterizes the human-machine collaboration status based on multi-source physiological and behavioral data acquired by the data acquisition module.

[0007] The second processing module is used to combine the preset basic failure risk of the augmented reality-based intelligent construction site auxiliary decision-making system with the collaborative trust index generated by the first processing module, and to evaluate and output the current overall failure risk through a nonlinear evolution model;

[0008] The adaptive decision-making module is used to respond to the overall failure risk output by the second processing module, and combine it with the preset efficiency-first utility function and resilience-first utility function to determine and output the final decision scheme.

[0009] Preferably, the first processing module generates a collaborative trust index, including:

[0010] Based on historical operational data, multi-source physiological and behavioral data are normalized to obtain normalized parameters;

[0011] Based on the preset risk correlation, the normalized parameters are converted into individual risk levels;

[0012] Based on individual risk levels and preset weighting coefficients, a collaborative trust index is generated through a linear weighting model.

[0013] Preferably, the second processing module assesses and outputs the overall failure risk, including:

[0014] A risk sensitivity coefficient is introduced to characterize the accelerated growth of risk with the loss of trust, and a nonlinear evolution model is constructed.

[0015] By inputting the collaboration trust index and the system's fundamental failure risk into the nonlinear evolution model, the overall failure risk is determined.

[0016] Preferably, the efficiency-first utility function is used to perform a weighted summation of the progress utility score and cost utility score of the alternative decision-making schemes to evaluate the economy and timeliness of the decision-making schemes.

[0017] Preferably, the resilience priority utility function is used to perform a weighted summation based on the schedule utility score, cost utility score, and interpretability index of the alternative decision schemes to evaluate the safety and reliability of the decision schemes.

[0018] Preferably, the adaptive decision-making module determines and outputs the final decision scheme, including:

[0019] Based on the overall failure risk and a preset switching risk threshold, a mode switching factor is generated using a logistic function.

[0020] A hybrid decision utility function is constructed by dynamically weighting the efficiency-first utility function and the resilience-first utility function using a mode-switching factor.

[0021] Iterate through all alternative decision schemes and select the scheme that maximizes the mixed decision utility function as the final decision scheme.

[0022] Preferably, the generation of the interpretability index includes:

[0023] Determine the similarity score between each instruction step in the decision-making scheme and the standard operating procedure;

[0024] Determine the cognitive complexity of each instruction step;

[0025] By combining the similarity scores of each instruction step with the cognitive complexity, an interpretability index is generated through weighted averaging.

[0026] Preferably, it also includes a trust repair module, used for:

[0027] Based on the collaborative trust index generated by the first processing module, the injection probability of the trust repair instruction is determined, wherein the injection probability increases rapidly as the collaborative trust index decreases;

[0028] Based on the injection probability, inject trust repair instructions into the current task sequence.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] 1. By introducing a collaborative trust index, the physiological and behavioral data of operators are transformed into quantitative indicators of human-machine collaboration status, which solves the problem of fuzzy and difficult-to-measure human-machine collaboration status in existing technologies, and provides a reliable data foundation for risk assessment and adaptive decision-making.

[0031] 2. By constructing a nonlinear evolution model, the overall system failure risk caused by the decline in operator trust can be dynamically assessed, accurately depicting the accelerated growth of risk with the loss of trust, and improving the accuracy and foresight of risk assessment;

[0032] 3. By establishing two utility functions—efficiency-first and resilience-first—and using a risk-level-based mode-switching factor for dynamic weighting, the decision-making logic achieves adaptive switching between pursuing efficiency and ensuring safety, overcoming the problem of rigid logic in traditional decision-making systems.

[0033] 4. By introducing an interpretability index into decision-making and proactively injecting trust repair instructions when trust levels decline, not only is the safety and reliability of decision-making schemes under high risk ensured, but a proactive trust intervention mechanism is also established, thereby improving the long-term stability and robustness of the human-machine collaboration system. Attached Figure Description

[0034] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0035] Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0036] 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.

[0037] Example 1:

[0038] Please see Figure 1 An augmented reality-based intelligent construction site decision support system includes:

[0039] The data acquisition module is used to acquire multi-source physiological and behavioral data of the operator in real time through wearable sensing devices and augmented reality glasses;

[0040] The first processing module is used to generate a collaborative trust index that characterizes the human-machine collaboration status based on multi-source physiological and behavioral data acquired by the data acquisition module.

[0041] The second processing module combines the preset basic failure risk of the augmented reality-based intelligent construction site auxiliary decision-making system with the collaborative trust index generated by the first processing module, and evaluates and outputs the current overall failure risk through a nonlinear evolution model.

[0042] The adaptive decision-making module responds to the overall failure risk output by the second processing module and, in conjunction with preset efficiency-first utility functions and resilience-first utility functions, determines and outputs the final decision scheme. To provide a basis for decision-making, the system also includes a decision scheme generation module, which operates before the adaptive decision-making module. Its function is to search for multiple alternative construction paths and operation sequences with varying costs based on the current construction task, resource status, and environmental constraints. For example, using the A* algorithm, the system uses the construction site map as a grid network, with the current task point as the starting point and the target point as the ending point. It comprehensively considers path length, obstacles, and resource availability as heuristic cost functions to form alternative decision schemes. A set of options; or a set of feasible alternative decision-making options a is generated by searching a database of historical success cases; these options provide input options for subsequent utility evaluation and final decision.

[0043] This embodiment provides an augmented reality-based intelligent construction site auxiliary decision-making system, which aims to solve the problems of ambiguous human-machine collaboration status, static risk assessment, and rigid decision-making in intelligent construction by dynamically evaluating the human-machine collaboration status and achieving adaptive switching of the system's decision-making mode.

[0044] The system's technical solution consists of a data acquisition module, a first processing module, a second processing module, and an adaptive decision-making module working together.

[0045] The data acquisition module is designed to capture underlying data reflecting the operator's actual working status in a seamless and real-time manner. In this embodiment, this module is implemented through a monitoring unit integrated into the augmented reality glasses and wearable sensing device worn by the operator. Specifically, the eye-tracking module built into the augmented reality glasses is used to acquire the gaze duration per unit time in real time. With video scanning rate The interaction log records the delay in command confirmation. Wearable sensing devices collect heart rate variability The augmented reality glasses' cameras and positioning system can also record the number of times the operator makes minor adjustments to the device. These data collectively constitute multi-source physiological and behavioral data characterizing the physiological and behavioral state of operators; this multi-dimensional data collection method can provide earlier and more accurate insights into potential fatigue, confusion, or distrust among operators.

[0046] The first processing module is designed to extract and integrate the previously collected discrete and multidimensional raw data into a single, standardized indicator that can macroscopically characterize the state of human-machine collaboration, namely, the collaboration trust index. This collaboration trust index provides a quantitative and standardized input for subsequent risk assessment, thereby solving the problem that the state of human-machine collaboration is difficult to measure directly.

[0047] The second processing module is responsible for dynamically and non-linearly assessing the current overall risk level of the system; this module presets a basic system failure risk. This risk represents the probability that the system might fail due to inherent factors such as hardware and software, even under an ideal state of complete trust between humans and machines. Its value can be determined based on historical data and system design safety redundancy. To further clarify its internal logic, this module uses the collaborative trust index generated by the first processing module as a core dynamic variable, calculates it through a nonlinear evolution model, evaluates and outputs the current overall failure risk that reflects the sharp amplification of risk caused by the deterioration of the human-machine collaboration state. This design more realistically simulates the nonlinear impact of trust breakdown on system security in the real world.

[0048] The adaptive decision-making module intelligently selects the most appropriate decision logic based on the system's current risk level to achieve a dynamic balance between efficiency and safety. Responding to the overall failure risk output by the second processing module, and based on preset efficiency-first utility functions and resilience-first utility functions, the module determines and outputs a final decision scheme through a smooth switching mechanism. This enables the system to maximize construction efficiency and economy under low-risk conditions, while automatically switching to a decision mode that prioritizes safety, reliability, and ease of operator understanding under high-risk conditions, thus achieving adaptive intelligent decision-making.

[0049] This embodiment constructs a complete technical closed loop through the collaborative work of the aforementioned modules, from operator status perception to system risk assessment, and then to adaptive adjustment of decision-making modes. It solves the problems of existing technologies, such as the inability to quantify human-machine collaboration status, the inability to dynamically assess human-caused risks, and rigid decision-making logic. By introducing a collaboration trust index, it transforms the ambiguous human-machine relationship into calculable engineering parameters. Through a nonlinear evolution model, it accurately characterizes the amplification effect of trust deficiency on system security. Through dual utility functions and adaptive switching mechanisms, it endows the system with intelligent decision-making capabilities that balance efficiency and resilience in complex and ever-changing construction sites, significantly improving the safety, reliability, and smoothness of human-machine collaboration in intelligent construction systems.

[0050] Example 2:

[0051] The first processing module generates a collaboration trust index, including:

[0052] Based on historical operational data, multi-source physiological and behavioral data are normalized to obtain normalized parameters;

[0053] Based on the preset risk correlation, the normalized parameters are converted into individual risk levels;

[0054] Based on individual risk levels and preset weighting coefficients, a collaborative trust index is generated through a linear weighting model.

[0055] Based on Example 1, this embodiment provides further limitations and optimizations on how the first processing module specifically generates the collaborative trust index;

[0056] The specific implementation of this module includes normalizing multi-source physiological and behavioral data; normalization refers to mapping raw parameters with different dimensions and numerical ranges to a unified standard. The process of dividing the data into intervals serves to eliminate dimensional differences, making parameters of different dimensions comparable. In this embodiment, the calculation method is as follows:

[0057]

[0058] in, Let be the i-th raw parameter value collected at time t. and These are the maximum and minimum values ​​of this parameter in the historical database, respectively. These are the normalized parameters obtained through calculation; to ensure the robustness of the calculation, when... In the case of normalization parameters It can be set to a constant value, such as 0.5; in addition, the data acquisition module can also incorporate outlier detection and smoothing filtering algorithms to handle sensor data loss or transient noise interference, ensuring the accuracy of the input to the model. The stability and rationality of the value;

[0059] Based on normalized parameters and according to pre-defined risk correlations, they are converted into individual risk levels. Risk correlation refers to the pre-determined relationship between the changing trend of a parameter and the increase or decrease of risk, based on prior physiological or behavioral knowledge. Its purpose is to unify the direction of all parameters' contribution to risk. The conversion method is: if a parameter is positively correlated with risk, then... If the parameter is negatively correlated with risk, then Through this transformation, all individual risk levels The value range is Furthermore, the larger the value, the higher the risk;

[0060] Based on individual risk levels and preset weighting coefficients, a collaborative trust index is generated through a linear weighted model; weighting coefficients This refers to the numerical values ​​assigned to each individual risk level, reflecting its importance and sensitivity in representing a decline in trust, with the sum of all weighting coefficients equal to 1. The initial values ​​of these weighting coefficients can be set based on the prior knowledge of domain experts and dynamically optimized during system operation through machine learning regression analysis using historical security event data. (Collaborative Trust Index) The calculation formula is

[0061]

[0062] This formula involves weighting and summing the risk levels of all dimensions to obtain a total risk score, then subtracting this score from 1 to obtain a positive trust index. This technical solution provides a systematic and reproducible collaborative trust index modeling method, which greatly enhances the reliability and interpretability of the trust index and provides high-quality input for subsequent risk assessment and decision-making. It should be noted that although the linear weighted model is a simplification, it provides an effective and feasible framework for quantifying trust. Those skilled in the art will understand that more complex nonlinear models can be used to replace the linear weighted model to capture more complex interactions between indicators, which may further improve the accuracy of trust assessment.

[0063] Example 3:

[0064] The second processing module assesses and outputs the overall failure risk, including:

[0065] A risk sensitivity coefficient is introduced to characterize the accelerated growth of risk with the loss of trust, and a nonlinear evolution model is constructed.

[0066] By inputting the collaboration trust index and the system's fundamental failure risk into the nonlinear evolution model, the overall failure risk is determined.

[0067] Based on Example 1, this embodiment elaborates in detail on how the second processing module specifically assesses and outputs the overall failure risk;

[0068] To characterize the accelerating growth of risk with a loss of trust, a risk sensitivity coefficient was introduced, and a nonlinear evolution model was constructed. This is a constant greater than 1, which serves as an index for the trust deficiency term, used to adjust the steepness of the risk growth curve. To calibrate this coefficient, an independent calibration dataset containing observations of the collaborative trust index during historical human-machine collaboration processes can be used. Corresponding system failure probability statistics By using regression analysis methods such as least squares, the formula was analyzed.

[0069] By performing a fitting process, the result can be determined. The possible values ​​of ;

[0070] The collaborative trust index and the system's fundamental failure risk are input into this nonlinear evolution model to determine the overall failure risk; the specific mathematical expression of this model is as follows:

[0071]

[0072] in, It represents the overall failure risk at time t. It is a risk of system fundamental failure. This represents the degree of trust deficiency. This nonlinear evolution model can more profoundly and realistically reflect the complex dynamics of human-machine collaboration on intelligent construction sites. By accurately simulating the nonlinear and accelerated impact of trust collapse on system security, the system can provide earlier and more sensitive warnings and responses to the potential catastrophic consequences of declining trust, significantly improving the accuracy and foresight of risk assessment.

[0073] Example 4:

[0074] The efficiency-first utility function is used to perform a weighted sum of the schedule utility score and cost utility score of alternative decision-making options to evaluate the economy and timeliness of the decision-making options.

[0075] Based on Example 1, this embodiment provides a detailed explanation of the composition of the efficiency-first utility function. This function is used to provide the decision-making system with an evaluation standard that prioritizes economy and timeliness when the system is running stably with low risk and high trust.

[0076] Efficiency-first utility function The schedule utility score and cost utility score of the alternative decision-making options are weighted and summed; schedule utility score This refers to the normalized score obtained by comparing the expected progress contribution of option a with the best and worst values ​​among all alternative options, used to quantify the advantage of option a in the time dimension; its specific calculation formula can be...

[0077]

[0078] in, The expected time for scheme a is... and These represent the maximum and minimum expected time among all alternatives, respectively. This formula ensures that the shorter the time, the higher the score; cost-utility score. This refers to the score obtained by inversely normalizing the expected resource cost of option a, used to quantify the economic advantage of option a; its specific calculation formula can be:

[0079]

[0080] in, The expected resource cost of option a, and Let be the maximum and minimum expected costs among all alternatives, respectively. This formula uses inverse normalization to ensure that lower costs result in higher scores; the mathematical expression of this function is:

[0081]

[0082] in, and These are the weighting coefficients for schedule and cost, respectively, whose values ​​can be preset according to the strategic focus of the project management party, and the sum of the two is 1; This embodiment clarifies the quantitative basis for decision optimization under normal system operation, and provides a clear and calculable mathematical model for the system to automatically select the optimal economic and time benefit scheme under low-risk conditions.

[0083] Example 5:

[0084] The resilience-first utility function is used to perform a weighted summation of the schedule utility score, cost utility score, and interpretability index of alternative decision-making options to assess the safety and reliability of the decision-making options.

[0085] Based on Example 1, this embodiment provides a detailed explanation of the composition of the resilience priority utility function. This function is used to provide the decision-making system with an evaluation criterion that prioritizes safety, reliability, and ease of operator acceptance when the system enters a high-risk, low-trust alert state.

[0086] Resilience priority utility function The calculation involves a weighted sum of the schedule utility score, cost utility score, and interpretability index of the alternative decision-making options; its key innovation lies in the introduction of the interpretability index dimension; the interpretability index... This is a quantitative assessment of the operator-friendliness and suitability of the instruction sequence in decision scheme a for operators, taking into account their operating habits and cognitive abilities. Its purpose is to ensure that, under high-risk conditions, the system-generated instructions are ones that operators can accurately understand and execute. The mathematical expression of this function is:

[0087]

[0088] in, These are the weighting coefficients for schedule, cost, and interpretability in this model. The values ​​of these weights are set relatively high to highlight the focus on safety. These weights can be set according to safety management procedures and human factors engineering principles, and their sum is 1. This embodiment creatively introduces an interpretability index and assigns it a high weight, ensuring that the decision-making scheme generated by the system is highly safe, reliable and easy for operators to execute in emergency or uncertain situations, thereby greatly enhancing the resilience and stability of the human-machine system when facing risk shocks.

[0089] Example 6:

[0090] The adaptive decision-making module determines and outputs the final decision scheme, including:

[0091] Based on the overall failure risk and a preset switching risk threshold, a mode switching factor is generated using a logistic function.

[0092] A hybrid decision utility function is constructed by dynamically weighting the efficiency-first utility function and the resilience-first utility function using a mode-switching factor.

[0093] Iterate through all alternative decision schemes and select the scheme that maximizes the mixed decision utility function as the final decision scheme.

[0094] Based on Example 1, this embodiment elaborates in detail on the internal working mechanism of how the adaptive decision-making module determines and outputs the final decision scheme;

[0095] This module generates a mode switching factor based on the overall failure risk and a preset switching risk threshold using a logistic function. To avoid abrupt changes in the decision mode at a certain risk point, this embodiment uses a logistic function to generate a continuously changing mode switching factor. Switch risk thresholds This is a preset risk level, defined by the system as the critical point at which the risk becomes unacceptable. The threshold is set based on the specific construction scenario and the criticality of the task. The mathematical expression for this function is:

[0096]

[0097] in, The slope of a dimensionless switching function controls the speed of switching, and its value can be determined according to the task criticality of the specific construction scenario.

[0098] Furthermore, a hybrid decision-making utility function is constructed by dynamically weighting the efficiency-first utility function and the resilience-first utility function using a mode-switching factor; its mathematical expression is as follows:

[0099]

[0100] Among them, through weight and The ebb and flow of these factors enabled a smooth shift in the focus of decision-making.

[0101] Finally, after iterating through all alternative decision options, the option that maximizes the mixed decision utility function is selected as the final decision option.

[0102]

[0103] This solution achieves a seamless and smooth transition between decision-making modes through logistic functions and hybrid utility functions, making the system's response to risks more refined and stable, and greatly improving the intelligence level and environmental adaptability of the entire auxiliary decision-making system.

[0104] Example 7:

[0105] The generation of the interpretability index includes:

[0106] Determine the similarity score between each instruction step in the decision-making scheme and the standard operating procedure;

[0107] Determine the cognitive complexity of each instruction step;

[0108] By combining the similarity scores of each instruction step with the cognitive complexity, an interpretability index is generated through weighted averaging.

[0109] Based on Example 5, this embodiment elaborates in detail the specific method for generating the interpretability index, a key parameter.

[0110] To generate this index, it is necessary to determine the similarity score between each instruction step in the decision-making scheme and the standard operating procedure (SOP); the SOP refers to the validated best practice procedures preset in the system's knowledge base; the similarity score... It uses natural language processing or structured instruction comparison technology to calculate the degree of matching between the k-th instruction step and the most relevant entry in the standard operating procedure knowledge base;

[0111] At the same time, the cognitive complexity of each instruction step needs to be determined; cognitive complexity This refers to evaluating the amount of information contained in the k-th instruction step itself, the complexity of the interaction, and the requirements on the operator's memory and reasoning abilities. This parameter is also normalized to... interval;

[0112] Based on the above scores and complexity, an interpretability index is generated through a weighted average; the calculation formula is as follows:

[0113] in, It is the final interpretability index of decision option a. This represents the total number of instruction steps contained in scheme a. It is the similarity score at step k. It is the cognitive complexity of the k-th step. These are weighting coefficients, which sum to 1, and their values ​​can be calibrated through small-scale user experiments. This embodiment provides a specific and operable quantification method for the interpretability index, greatly enhancing the scientificity and effectiveness of the resilience-first decision-making model.

[0114] Example 8:

[0115] It also includes a trust repair module, used for:

[0116] Based on the collaborative trust index generated by the first processing module, the injection probability of the trust repair instruction is determined, wherein the injection probability increases rapidly as the collaborative trust index decreases;

[0117] Based on the injection probability, inject trust repair instructions into the current task sequence.

[0118] This embodiment adds a trust repair module to the system in embodiment 1. The purpose of this module is to take proactive measures to intervene when a decline in the trust level of human-machine collaboration is detected, so as to repair and rebuild the trust relationship.

[0119] The trust repair module determines the injection probability of trust repair instructions based on the collaborative trust index generated by the first processing module. Trust repair instructions refer to non-critical instructions that serve to confirm, reassure, or provide additional clear explanations. The injection probability... The design motivation is that the lower the level of trust, the more urgent the need for remedial measures; its calculation formula is...

[0120]

[0121] in, It is a collaboration trust index. This is the maximum injection probability, used to limit the highest frequency of repair commands. It is a sensitivity coefficient greater than 0, and its function is to accelerate the increase in injection probability as the trust level decreases; and The settings are determined by the system designer based on the tolerance for task interruption and the urgency of trust restoration;

[0122] Based on the calculated injection probability, this module injects trust repair instructions into the current task sequence; in each decision cycle, the system will... A random judgment is made. If the judgment is successful, a suitable instruction is selected from the preset trust repair instruction library and inserted into the task sequence to be issued to the operator. This proactive intervention strategy works in parallel with the decision mode switching to form a dual protection mechanism for the system in high-risk, low-trust environments. This helps prevent the continuous decline in trust and improves the long-term stability and robustness of the human-machine collaboration system.

[0123] 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 it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An intelligent construction site auxiliary decision-making system based on augmented reality, characterized in that... ,include: The data acquisition module is used to acquire multi-source physiological and behavioral data of the operator in real time through wearable sensing devices and augmented reality glasses; The first processing module is used to generate a collaborative trust index that characterizes the human-machine collaboration status based on multi-source physiological and behavioral data acquired by the data acquisition module. The second processing module is used to combine the preset basic failure risk of the augmented reality-based intelligent construction site auxiliary decision-making system with the collaborative trust index generated by the first processing module, and to evaluate and output the current overall failure risk through a nonlinear evolution model; The adaptive decision-making module is used to respond to the overall failure risk output by the second processing module, and combine it with the preset efficiency-first utility function and resilience-first utility function to determine and output the final decision scheme.

2. The augmented reality-based intelligent construction site auxiliary decision-making system according to claim 1, characterized in that... The first processing module generates a collaboration trust index, including: Based on historical operational data, multi-source physiological and behavioral data are normalized to obtain normalized parameters; Based on the preset risk correlation, the normalized parameters are converted into individual risk levels; Based on individual risk levels and preset weighting coefficients, a collaborative trust index is generated through a linear weighting model.

3. The augmented reality-based intelligent construction site auxiliary decision-making system according to claim 1, characterized in that... The second processing module assesses and outputs the overall failure risk, including: A risk sensitivity coefficient is introduced to characterize the accelerated growth of risk with the loss of trust, and a nonlinear evolution model is constructed. By inputting the collaboration trust index and the system's fundamental failure risk into the nonlinear evolution model, the overall failure risk is determined.

4. The augmented reality-based intelligent construction site auxiliary decision-making system according to claim 1, characterized in that... The efficiency-first utility function is used to perform a weighted summation of the schedule utility score and cost utility score of the alternative decision-making schemes to evaluate the economy and timeliness of the decision-making schemes.

5. The augmented reality-based intelligent construction site auxiliary decision-making system according to claim 1, characterized in that... The resilience priority utility function is used to perform a weighted summation based on the schedule utility score, cost utility score, and interpretability index of the alternative decision schemes to evaluate the safety and reliability of the decision schemes.

6. The augmented reality-based intelligent construction site auxiliary decision-making system according to claim 1, characterized in that... The adaptive decision-making module determines and outputs the final decision scheme, including: Based on the overall failure risk and a preset switching risk threshold, a mode switching factor is generated using a logistic function. A hybrid decision utility function is constructed by dynamically weighting the efficiency-first utility function and the resilience-first utility function using a mode-switching factor. Iterate through all alternative decision schemes and select the scheme that maximizes the mixed decision utility function as the final decision scheme.

7. The augmented reality-based intelligent construction site auxiliary decision-making system according to claim 5, characterized in that... The generation of the interpretability index includes: Determine the similarity score between each instruction step in the decision-making scheme and the standard operating procedure; Determine the cognitive complexity of each instruction step; By combining the similarity scores of each instruction step with the cognitive complexity, an interpretability index is generated through weighted averaging.

8. The augmented reality-based intelligent construction site auxiliary decision-making system according to claim 1, characterized in that... It also includes a trust repair module, used for: Based on the collaborative trust index generated by the first processing module, the injection probability of the trust repair command is determined, wherein the injection probability increases rapidly as the collaborative trust index decreases; Based on the injection probability, inject trust repair instructions into the current task sequence.

Images